SW1451 c002

Chapter 2
Analysis and design of machine

foundations
2.1 INTRODUCTION
This chapter deals with vibration analysis and design of machine foundations subjected
to dynamic load. As a pre-requisite to this chapter, you should be thoroughly familiar
with concepts that are put in chapter 5 (Vol. 1) on
• Basic concepts in structural dynamics.

• Basic concepts in soil dynamics.
Armed with these basics, we believe you will find this chapter interesting and find
design of machine foundation a challenging and intellectually stimulating task.
Machine foundations are one of the most important features of industrial devel-
opment. In both developed and developing countries, growth of economy is largely
attributed to development of industry and infra-structure facilities.
In industrial facilities like Power Plants, Steel Plants, Petrochemical Complexes,
Fertiliser Plants etc., consist of a number of centrifugal and reciprocating machines
and these play an important role to ensure smooth operation of the process and that
the output product is of right quality. If any of these equipments starts malfunc-
tioning or breaks down due to excessive vibration or settlement of the foundations,
cascading effect on the overall performance on engineering could be catastrophic at
times.
We give two case histories below making you aware of how far reaching could be
the consequences.
2.1.1 Case history #1

In Middle-East, in one of the oil producing nation there was a plant which had been
operating for last 25 years smoothly, sweetening the sour gas that was being pumped
into the complex from a nearby gas-field. The authorities hit a new source of natural
gas nearby this complex and the obvious choice was to pump gas from this new gas
field to the existing plant for further processing. This called for upgrading the plant
capacity. On engineering evaluation it was found that it necessitated certain changes in
diameter of the pipes, re-routing some of the existing pipes with new supports and also
changing the rating of the two-stroke reciprocating compressor which was existing at
the plant. The company management wanted to expedite the issue for they perceived
© 2009 Taylor & Francis Group, London, UK

84 Dynamics of Structure and Foundation: 2. Applications
that each day lost in production, the company stood to loose about 100,000 thousand
US dollars in profit.
In haste nobody thought to re-check performance of the foundation of the
compressor under dynamic load, now that its rating was changed!
When the plant started after this modification with additional gas being pumped
from the new gas field, whole pipe rack started shacking violently and the compressor
foundation started showing vibration amplitude that was well beyond acceptable limit.
The vibration became so high at 80% production level (at which the plant would
operate at most of the time) that the operation manager had no alternative but to stop
the plant completely.
Subsequent investigation revealed that with new rating of the compressor, the oper-
ating frequency now hovered very near to natural frequency of the foundation resulting
in a resonant condition and also induced additional excitation to the fluid flowing
through the pipe resulting in a force which the piping system was not capable of
taking care off, without undue distress.
This resulted in complete overhauling of the compressor foundation and stiffening
the pipe racks by additional bracings and all these re-engineering resulted in a delay
of about 5 months for full scale production and also a total revenue loss to company
in the tune of 300 million dollars1 .
2.1.2 Case history #2

In another case a medium scale factory requiring heavy duty power for its production
opened a new unit in an industrial area (in India), where during peak summer season
power supply was reported to be sporadic. To maintain optimum production level,
the owner procured 3 numbers of standby generator sets to supply power during
periods of power cuts. During soil investigation nobody thought of doing a test for
dynamic property of the soil and the foundation was designed based on obtaining
the dynamic properties by theoretical co-relation with other static engineering soil
parameters. After generators were installed and started operating it was found that
amplitude of vibration was well beyond tolerable limits resulting in tripping of the
machines quite often. As the generator failed to meet the optimal power demand,
production output nose dived quite substantially.
On investigation of the problem by a consultant hired by the owner it was found that
field observed dynamic properties varied widely with those considered from theoretical
co-relation for the amplitude and resonance check.
The consultant suggested that the generator foundation be modified by providing
additional mass of concrete, adjusting the height of foundation and partially re-routing
the cables adding to an additional cost of only 20% of original installation cost. The
suggestion was vetoed by the owner arguing that company was going through difficult
times financially and it was not possible to generate further funds for such additional
capital expenditures.
The owner hired a couple of mechanics and with some adjustments coerced the
machines into operation but still the performance did not improve significantly. Within
1 This was time when oil was priced at 25 dollars per barrel. In todays index the loss would be 4 to 5
times the actual loss incurred.

Analysis and design of machine foundations 85
a period of six months the machines broke down completely. The equipment supplier
refused to replace the machines (though they failed within the warranty period) arguing
that conditions put in the contract in terms of amplitude and frequency restrictions
were violated from the very outset and as such they were not responsible for bad
performance of the machine.
By this time the company was in such a poor condition financially due to failure
of production target, that it could not generate fund to replace and overhaul the
equipment and its foundations and had no other option but to declare it sick and close
the unit.
So lesson learnt from the above two cases are that if proper attention is not paid to
the design of these type of foundations, consequences could be quite far reaching and
serious in nature.
2.2 DIFFERENT TYPES OF FOUNDATIONS
Foundations supporting machines can be classified into the following categories:
• Block foundations resting on soil or piles

• Frame foundations
• Wall foundations
• Spring mounted machines resting on rafts/grade slabs
2.2.1 Block foundations resting on soil/piles

These types of foundations usually consist of massive RCC blocks resting on soil or
on piles and are as shown in Figures 2.2.1 and 2.
Block type of foundation are usually used to support machines like
1 Pumps
2 Motors
3 Generators
Figure 2.2.1 Block foundation resting on soil.

Figure 2.2.2 Block foundation on piles.
4 Coal mill foundations

5 Gas Turbines
6 Motor Driven Boiler feed pumps
7 Centrifugal/Reciprocating type Compressors etc.
2.2.2 How does a block foundation supporting rotating

machines differ from a normal foundation?
The function of a foundation is basically to transfer the load coming on it from super-
structure or items like vessels, tanks, skids etc. to the underlying soil. For a normal
foundation supporting systems like building structure, vessels, tanks etc. the major
load coming on it is static in nature. For geometric sizing of the foundation the stress
induced on soil being less than the allowable bearing capacity of the soil suffice. The
dynamic force coming if any are quite rare and are mostly those due to earthquake
forces and in case of very tall structures may be a bit more often due to wind induced
vibration. In majority of the time for a normal foundation static load pre-dominates.
While for machine foundation it is just the reverse. In most of the industrial facilities
production being round the clock, the major load coming on the foundation is dynamic
in nature and the foundation should be so designed that it is capable enough to sustain
this dynamic loads over and above the static loads without causing any distress to
underlying soil or to the machine it is supporting.
So the question boils down to what are these conditions for which the foundation
can safely sustain the dynamic load coming from the machines in operation?
There are usually two conditions that are checked for while designing a machine
foundation:

• Resonance check
• Amplitude check
2.2.2.1 Resonance check

All machines under operation usually induce a periodic dynamic load on the founda-
tion and in most of the case can be represented by a function like P0 sin ωm t, where
P0 = magnitude of the unbalanced force from the machine during its operation and
ωm = operating frequency of the machine usually expressed in
• radians/sec.
• Hertz or revolution per minute (rpm).
Due to this induced dynamic load from the machine the block foundation including
some portion of the soil underlying the foundation is subjected to vibration and it is
essential that the natural frequency (ωn ) of this vibration should be well away from
the operating frequency of the machine i.e. resonance condition should not prevail.
Irrespective of any code the normal practise is to design the foundation in such a
way that its operating frequency is at least ±20% away from the natural frequency of
the foundation.
2.2.2.2 Amplitude check

Under this condition, it is usually checked that the amplitude of vibration of the block
foundation is well within the acceptable limits of engineering practise.
The acceptable or the tolerable limits are usually suggested by the vendor supply-
ing the equipment or in absence of such data are usually obtained from the codal
stipulations.
If the amplitude of vibration is more than this acceptable limit can mar the
performance of the equipment in the following way
• Rapid deterioration of the machine due to heavy wear and tear.

• Excess amplitude of vibration inducing fatigue in the coupling and the connecting
shafts leading to repeated breakdowns.
• Damage to the piping system connected rigidly to the equipment.
• Increase in decibel level during operation causing discomfort to the operators.
• Accumulation of explosive gases which at times could be dangerous to human life
and property.
Based on the above discussion it is imperative that for a foundation designed for
dynamic load the above two conditions are met.
Now let us see how we mathematically model the soil-foundation system to
theoretically check the two conditions as mentioned above.
A block foundation as shown earlier constitute of a massive RCC block resting on
the ground supporting the machine aligned over it.
For all practical purpose the block and the machine is considered as a rigid lumped
mass supported on an elastic base constituting the underlying soil/pile.

Kθ Kh
Kv
Figure 2.2.3 Mathematical model of foundation in 2D.
1
1
Figure 2.2.4 Degrees of freedom in space.
Where, for analysis purpose the soil is modelled as equivalent linear springs.
Shown in Figure 2.2.3 is the mathematical model of a machine foundation with soil
modelled as linear springs based on mechanical analog of elastic half space theory in
2D, and 6 degrees of freedom it has on space (Figure 2.2.4).
Before we go into further details of the state of the art theory for design of such
foundations, it would possibly be worthwhile to look back at its evolution and study
its subsequent metamorphosis to the various techniques used in present day design
office practices.
2.2.3 Foundation for centrifugal or rotary type

of machine: Different theoretical methods
for analysis of block foundation
2.2.3.1 Tschebotarioff’s (1953) method

This is one of the early methods used for calculating the natural frequency of a founda-
tion. If fn is the natural frequency of the machine plus foundation in terms of contact

area then Tschebotarioff defined a term as reduced natural frequency fnr given by
√ W
fnr = fn σ , where σ = t/ft 2 (2.2.1)
Af
where W is the weight of the machine plus foundation and Af is the base area.
2.2.3.2 Alpan’s (1961) method

Alpan made use of Tschebotarioff’s theory and developed an expression of natural
frequency as
α 1/4
fn = √ (Af ) (2.2.2)
W
where, fn = natural frequency of the foundation in cycles per minute; W = weight

of the machine plus foundation in Kilogram; Af = contact area of the foundation in
square meter, and α = a constant whose value depends on the nature of the soil as
given in Table 2.2.1.
2.2.3.3 Newcomb’s (1951) method

Newcomb developed an empirical equation for natural frequency based on the static
deflection of the soil data and is expressed as

1
fn = 188 (2.2.3)
δst
where, fn = natural frequency in cycles per minute and, δst = static deflection in
inches.
The displacement parameter δst can be obtained from plate load test for any design
bearing pressure.
The above theories were mostly based on observation and experience and as such
are empirical in nature. The theories can be put to use to check the resonant condition
only no check for amplitude is possible by these methods. As such they shall only be
used for preliminary design or sizing of the foundation only.
Table 2.2.1 Value of constant, α.
Sl. No. Soil type Value of α
1 Peat 3900
2 Plastic clay 69000
3 Sand 82000
4 Sand stone 111000

2.2.4 Analytical methods
2.2.4.1 Hsieh’s (1962) method

Considering the soil as semi-infinite elastic medium, Hsieh put forward an analytical
treatment for vibration of circular foundation as given hereunder

For Translation: mẍ + r20 F2 ρGẋ + r0 GF1 x = P0 sin ωm t (2.2.4)

For rocking: ϕ θ̈ + r4θ F2 ρGẋ + Gr3θ F1 θ = M0 sin ωm t (2.2.5)
where, m = mass of the foundation plus the machine; x, θ = displacement vectors in

translational and rotational mode respectively; ϕ = mass moment of inertia; ρ = mass
density of soil; G = dynamic shear modulus of the soil; r0 = radius
of the circular foun-
dation in translational mode which for rectangular foundations is LB/π ; rθ = radius

3
of the circular foundation in rocking mode which for rectangular foundations is 4 LB3π

4 L3 B
or 3π as the case may be; a0 = frequency factor and is given by ωm r0 ρ/G; ωm =
the operating frequency of the machine; P0 , M0 = amplitude of exciting force in trans-
lation and rocking mode, and F1 , F2 = are functions whose values are given below:
F1 is usually represented in the form, α1 − α2 a20 .
Now substituting this value of F1 in the above differential equation we have,

For translation mode,

(m + α2 ρr30 )ẍ + r20 F2 ρGẋ + r0 Gα1 x = P0 sin ωm t (2.2.6)
For rocking mode,

(ϕ + α2 ρr5θ )θ̈ + r4θ F2 ρGẋ + α1 Gr3θ θ = M0 sin ωm t (2.2.7)
Representing the above equations as
mx ẍ + cx ẋ + kx x = P0 sin ωm t and I θ̈ + cθ θ̇ + kθ θ = M0 sin ωm t (2.2.8)
√
where, mx = (m + α2 ρr30 ); cx = r20 F2 ρG; kx = r0 Gα1 ; I = (ϕ + α2 ρr5θ ); cθ =
√
r4θ F2 ρG; kθ = α1 Gr3θ .
We had already seen earlier2 that the solution of such equation can be repre-
sented as
2 Refer Chapter 3 (Vol. 1) for solution of such equations having damped single degree of freedom.

P0
kx
sin ωm t
x̄max = (2.2.9)
[(1 − r2 )2 + (2Dr)2 ]

in which, r = ωm /ωnx
and D = c/cc where, ωnx = kx /mx ; and cc = critical damping
of the system and is 2 mx kx .
And for rocking mode,
M0
kθ
sin ωm t
θmax = (2.2.10)
[(1 − r2 )2 + (2Dr)2 ]

in which, r = ωm /ωnθand D = c/cc where, ωnθ = kθ /I; and cc = critical damping
of the system and is 2 Ikθ .
Here one point needs to be noticed is the additional term in the inertial coefficient
mx = (m + α2 ρr30 ) and I = (ϕ + α2 ρr5θ ) (2.2.11)
Here the original mass and mass moment of inertia terms get added up with an
additional term of α2 ρr30 and α2 ρr5θ respectively.
This can be attributed as added mass of soil which starts vibrating with the
foundation in same phase.
This looks quite logical for it has indeed been observed during field observation
that a part of soil below foundation do indeed participates in the vibration of the
foundation system.
Table 2.2.2 gives values of F1 and F2 for various modes given by Hsieh.
For uniform and parabolic distribution of pressure, Hsieh suggest to use an effective
radius αr0 where α is 0.78 and 0.59 respectively.
Table 2.2.2 Values of F1 and F2 .
Mode Poisson’s ratio F1 F2
Vertical (0 < a0 < 1.5) 0.0 4.0 − 0.5a02 3.3 + 0.4a0

0.25 5.3 − 1.0a02 4.4 + 0.8a0
0.50 8.0 − 2.0a02 6.9
Horizontal (0 < a0 < 2.0) 0.0 4.5 − 0.2a02 2.4 + 0.3a0
0.25 4.8 − 0.2a02 2.5 + 0.3a0
0.50 5.3 − 0.1a02 2.8 + 0.4a0
Rocking (0 < a0 < 1.5) 0.0 2.5 − 0.4a02 0.4a0
Torsion (0 < a0 < 2.0) All 5.1 − 0.3a02 0.5a0

2.2.4.2 Barkan or IS-2974 method

This is by far the most popular method in the design office. Barkan (1962) developed
this method way back in 60s and is still in vogue for design of machine foundations
under rotating loads.
In this method Barkan assumed the block foundation, shown in Figure 2.2.5, as
a rigid lumped mass (i.e. he assumed the concrete block to have infinite stiffness in
comparison to the soil and neglected any internal deformation of the concrete block
itself) having three degrees of freedom.
The soil medium he idealised as linear springs which he defined in terms of soil
parameter cz , cx & cφ which are otherwise known as coefficient of elastic uniform
compression, coefficient of elastic uniform shear, coefficient of elastic non-uniform
compression respectively.
2.2.4.2.1 Vertical direction

Here in the vertical direction the spring constant is considered as
kz = cz · Af (2.2.12)
where, kz = equivalent spring in vertical direction; cz = coefficient of elastic uniform

compression, and Af = plan area of the foundation.
Pz sin t
m
C/L of machine shaft
h Mx sin t
m
mx
x Px sin m
t
H
Zc
mz
H
x0
V
Figure 2.2.5 Mathematical model of Barkan for vertical and coupled sliding and rocking motion about
the minor axes of the foundation.

The natural frequency of the foundation is given by

kz
ωz = and, (2.2.13)
m
the amplitude of vertical vibration is given by
Pz sin ωm t
kz
δz = (2.2.14)
1 − r2
where, r = ωm /ωn ; ωm = operating frequency of the machine.
2.2.4.2.2 For coupled horizontal and rocking mode

It has been observed by Barkan that when a foundation has horizontal force along its
minor axis the foundation undergoes sliding and rocking simultaneously. When the
foundation starts vibrating resistance is mobilised in the soil in terms of forces V and
H as shown in Figure 2.2.5.
The resistive force may thus be expressed as

VR = Cφ dA (2.2.15)
and the resistive moment is expressed

MR = Cφ 2 φdA = Cφ IA φ (2.2.16)
where = distance between rotation axis and the element of area dA; φ = angular
rotation of the machine foundation; IA = second moment of area of the foundation
contact surface with respect to the axis passing through the centroid of the area and
perpendicular to the plane of the vibration.
2.2.4.2.3 For horizontal force, H
H = cτ Af x0 = cτ Af (x − Zc φ) (2.2.17)
where, A = area of base contact; Zc , x, x0 etc. are shown in Figure 2.2.5.

Now applying D’Alembert’s equation for dynamic equilibrium3 , we have
mẍ + H = Px sin ωm t or mẍ + cτ Af (x − Zc φ) = Px sin ωm t (2.2.18)
where m is the mass of the machine foundation.
3 Refer Chapter 5 (Vol. 1) for definition of D’Alembert’s equation.

Similarly for the moment equation about the minor axis of the foundation we have
Jxφ φ̈ − cτ Af Zc x + φ(cφ IA − WZc + cτ AZc2 ) = Mx sin ωm t (2.2.19)
where Jxφ = mass moment of inertia of the machine-foundation block about the minor
axis of rotation.
From Equations (2.2.18) and (2.2.19), we see that they contain both x and φ, so a
coupled sliding and rocking motion will develop along this direction.
Using the above equations and considering free vibrations, Barkan developed the
following equation for calculation of the frequencies.
J0 (ωφ2 + ωx2 )ω2 ωφ2 ωx2 J0

ω4 − + =0 (2.2.20)
Jxφ Jxφ
c I −WZ c A
where J0 = Jxφ + mZc2 ; ωφ2 = φ A J0 c and ωx2 = τm f .
Based on the above, the two principal frequencies for the coupled vibration is
given by
⎡ ⎤
J0 ⎣ 2 4Jxφ ωφ2 ωx2
2
ω1,2 = ωφ + ωx2 ± (ωx2 + ωφ2 )2 − ⎦ (2.2.21)
2Jxφ J0
Considering the forced vibration, the amplitudes Ax , Aφ may be expressed as
(cφ IA − WZc + cτ Af Zc2 − Jxφ ωm

2 )P ± c A Z M
x τ f c x
Ax = sin ωm t
mJxφ (ω12 − ωm
2 )(ω2 − ω2 )
2 m
cτ Af Zc Px ± (cτ Af − mωm
2 )M
x
Aφ = sin ωm t
mJxφ (ω12 − ωm
2 )(ω2 − ω2 )
2 m
(2.2.22)
2.2.4.2.4 Torsional mode

For this mode, again the foundation considered is a lumped mass having single degree
of freedom when the frequency and amplitude are given by
T sin ωm t
Kψ kψ
ωψ = and ψ = (2.2.23)
Iψ 1 − r2
where, Kψ = cψ Iψ ; r = ωm /ωn .
This method is also recommended by IS 2974 “Code and design practices for
machine foundation” and still remains the most popular method for vibration analysis
of block foundations.
But let us see the limitations of Barkans method with respect to the reality under
field conditions.

2.2.4.2.5 The Limitations

The major limitations that can be attributed to Barkan’s method are as follows:
• Barkan’s model does not take damping into consideration. It has been observed
from field instrumentation data that damping plays a significant role in the overall
response of the foundation especially when the operating frequency of the machine
is low.
• It does not account for the embedment effect of the surrounding soil which could
play a significant role on the magnitude of soil stiffness and damping.
• It does not take into cognisance the virtual mass of soil which vibrates in same
phase with the machine and the foundation.
• Barkan suggested spring value (usually the coefficient of uniform elastic compres-
sion) of the soil to be obtained from dynamic plate load test4 and may only give
correct values for a shallow depth below the surface while this may not be valid
for layered soil and also when the contact area of the foundation is large.
So based on the above limitations it was felt to upgrade the mathematical process
in the design of machine foundation.
Before we study the further enhancements it would be worth to write Barkan’s
equation in a more generic form.
The soil spring stiffness is described by the terms
Kx = cτ Af and Kxφ = cφ IA , (2.2.24)
while, the equations of equilibrium are defined as
mẍ + cτ Af (x − Zc φ) = Px sin ωm t, and
Jxφ φ̈ − cτ Af Zc x + φ(cφ IA − WZc + cτ Af Zc2 ) = Mx sin ωm t (2.2.25)
Substituting the values of Kx and Kφ , we have
mẍ + Kx (x − Zc φ) = Px sin ωm t and

Jxφ φ̈ − Kx Zc x + φ(Kφ − WZc + Kx Zc2 ) = Mx sin ωm t (2.2.26)
The above on writing in matrix form can be represented as

m 0 ẍ Kx −Kx Zc x Px
+ = sin ωm t (2.2.27)
0 Jxφ φ̈ −Kx Zc Kφ + Kx Zc2 − WZc φ Mx
Based on the above equation we will see later how we develop further realistic model
of the coupled horizontal and rocking mode5 .
4 Usually carried out with a plate of size 300 mm × 300 mm.

5 Structural Engineers be alert from this point.What we are going to apply herein subsequently are the
theories of structural dynamics for system with two degrees of freedom.

2.2.4.3 Richart and Lysmer’s model

Richart et al. (1970) idealised the foundation as a lumped mass supported on soil
which is idealised as frequency independent springs which he described in terms of
soil parameter dynamic shear modulus or shear wave velocity of the soil for circular
footing when footings having rectangular shape in plan can be converted into a footing
having equivalent circular radius.
Tables 2.2.3 and 4 along with Figure 2.2.6 show the different values of spring and
damping value as per Richart and Lysmer.
In which, G = dynamic shear modulus of the soil and is given by, G = ρs Vs2 ;
ν = Poisson’s ratio of the soil; ρs = mass density of the soil; Vs = shear wave
velocity of the soil obtained from soil testing; g = acceleration due to gravity;
m = mass of the machine and foundation; J = mass moment of inertia of the
machine and the foundation about the appropriate axes; K = equivalent spring
stiffness of the soil; C = damping value of the soil; B = inertial factor contribut-
ing to the damping factor; D = damping ratio of the soil; r = equivalent radius
of a circular foundation; L = length of the foundation, and, B = width of the
foundation.
Many engineers in design offices prefer to use Richart’s springs neglecting the damp-
ing and use Barkan’s formulation in matrix form as shown earlier to find out the
natural frequency and amplitude of the foundation. But, by neglecting the damp-
ing, he could significantly over-estimate the amplitude of vibration (specially for low
tuned machines) thus adding to the cost by trying to restrict it within the acceptable
limits.
Let us now see as to what form the equations take when damping is introduced in
the system.
Table 2.2.3 Values of soil springs as per Richart and Lysmer (1970) model.
Sl. No. Direction Spring value Equivalent radius Remarks

4Grz LB
1 Vertical Kz = rz = This is in vertical Z direction
(1 − υ) π

32(1 − υ)Gr x LB
2 Horizontal Kx = rx = This induce sliding in horizontal
(7 − 8υ) π X or Y direction

8Gr 3φx 4 LB3
3 Rocking Kφ x = rφx = This produces rocking about
3(1 − υ) 3π Y axis

8Gr 3φy 4 L3 B
3.1 Rocking Kφy = rφy = This produces rocking about
3(1 − υ) 3π X axes

16Gr 3ψ 4 L3 B + BL3
4 Twisting Kψ = rψ = This produces twisting about
3 6π vertical Z axis

Table 2.2.4 Values of soil damping as per Richart and Lysmer (1970) model.
Damping ratio and

Sl. No. Direction Mass ratio (B) Damping value Remarks
0.25 m(1 − υ)g 0.425
1 Vertical Bz = ζz = √ , This is damping value is in
ρs rz3 Bz vertical Z direction.
√
Cz = 2ζ z Kz m
(7 − 8υ)mg 0.288
2 Horizontal Bx = ζx = √ , This damping value is in
32(1 − υ)ρs rx3 Bx lateral X or Y direction
√
Cx = 2ζx Kx m
0.375(1 − υ)Jφx g 0.15
3 Rocking Bφx = ζφx = , This damping value is for
ρs rφx
5
(1 + Bφx ) Bφx rocking about Y direc-
tion
Cφx = 2ζφx Kφx Jφx
0.375(1 − υ)Jφy g 0.15
3.1 Rocking Bφy = ζφy = , This damping value is for
ρs rφy
5
(1 + Bφy ) Bφy rocking about Y axes

Cφy = 2ζφy Kφy Jφy
Jψ g 0.5
4 Twisting Bψ = ζψ = , This damping value is valid
ρs rψ5 1 + 2Bψ for twisting about verti-

Cψ = 2ζψ Kψ Jψ cal Z axis.
Y
L
Figure 2.2.6 3D View of the block foundation.
2.2.4.3.1 Vertical motion considering damping of the soil

For vertical direction the equation becomes that of a lumped mass having single degree
of freedom when
mz̈ + Cz ż + Kz z = P0 sin ωm t (2.2.28)

solution is,

Kz (P0 /Kz ) sin ωm t
ωz = and δz = (2.2.29)
m (1 − r2 )2 + (2Dz r)2
ωm
where r = ωn and Dz = damping ratio.
2.2.4.3.2 Coupled horizontal and rocking motion considering damping soil

We have seen that based on Barkan’s formulation the equation of motion in matrix
form is

m 0 ẍ Kx −Kx Zc x P0
+ = sin ωm t (2.2.30)
0 Jxφ φ̈ −Kx Zc Kφ + Kx Zc2 − WZc φ M0
Since the above equation is based on D’Alembert’s equation, the equation are said
to be statically coupled when the stiffness matrix and damping matrix have the same
matrix form (Meirovitch 1975). Thus, based on the above argument the damped
equation of motion in coupled rocking and sliding mode becomes

m 0 ẍ Cx −Cx Zc ẋ
+
0 Jxφ φ̈ −Cx Zc Cφx + Cx Zc2 − WZc φ̇

Kx −Kx Zc x P0
+ = sin ωm t (2.2.31)
−Kx Zc Kφx + Kx Zc2 − WZc φ M0
The above equations constitute the complete equation of motion for coupled sliding
and rocking mode considering the damping effect of the soil.
Actually for all practical calculations for finding out the dynamic response of the
foundation the term −WZc is usually neglected, for it has been observed that unless
the foundation is very massive and deep the term WZc has no significant effect on the
overall response of the system.
Based on the above argument the above equation reduces to

+
0 Jxφ φ̈ −Cx Zc Cφx + Cx Zc2 φ̇

Kx −Kx Zc x P0
+ = sin ωm t (2.2.32)
−Kx Zc Kφx + Kx Zc2 φ M0
The equation above surely looks elegant, but now comes the catch . . . , for this
damping matrix of soil is not proportional to either the mass or the stiffness of the soil,
moreover they are coupled by the term of Zc and W (the weight of the foundation) and

as such do not de-couple on orthogonal transformation6 . This forms a major headache

to the designer as he is not in a position to guess the damping ratio.
As we had already discussed in Chapter 5 (Vol. 1) that the most appropriate tech-
nique in such case is to resort to Time history analysis for the correct answer, many
engineers find time history too intensive in terms of calculation7 and prefer to use
modal response technique as a tool for analysis of the same. Of course the easiest way
out is to neglect the damping and argue that the design is conservative!
But this need not be done for we have already stated in Chapter 5 (Vol. 1) that it is
possible to by pass this problem of orthogonal de-coupling, even when the damping
matrix is non-proportional which though not exact would give still give a designer
a reasonable value to estimate a more realistic amplitude of vibration (it is surely a
better value than no damping considered).
We will consider application of this technique in subsequent section (section 2.2.5).
2.2.4.3.3 Torsional mode

In this mode again the block foundation is again considered as a lumped mass having
single degree of freedom, natural frequency and the torsional rotation, ψ is given by
T sin ωm t
Kψ Kψ
ωψ = ; and, ψ = (2.2.33)
Iψ 2
(2Dr) + (1 − r2 )2
where Kψ = 16Gr3ψ /3, D is the damping ratio in the torsion mode (Table 2.2.4) and
r is the ratio between the natural frequency of the foundation in torsion mode and the
operating frequency of the machine.
2.2.5 Approximate analysis to de-couple equations

with non-proportional damping
We have seen that the equation of motion is given by

+
0 Jxφ φ̈ −Cx Zc Cφx + Cx Zc − WZc φ̇
2
(2.2.34)
Kx −Kx Zc x P0
+ = sin ωm t
−Kx Zc Kφx + Kx Zc2 − WZc φ M0
For finding the natural frequencies we perform the eigen value analysis when the
un-damped equation becomes (neglecting-WZc for reasons as cited earlier)

Kx − mλ −Kx Zc
=0 (2.2.35)
−Kx Zc Kφ + Kx Zc2 − Jφx λ
6 For further explanation on this property refer to Chapter 5 (Vol. 1), the topic of orthogonal trans-
formation for modal response technique.
7 And the theory underlying time history is eluding many is not too an uncommon a fact. . ..

Solving the above equation we find out the eigen value vis-a vis the natural
frequency of the foundation system. Let the eigen values be λ1 and λ2 respectively. Let
corresponding eigen vectors be φxx φφx T and φφx φφφ respectively, when the
T
φ φφx
complete eigen vector matrix is expressed as, xx .
φφx φφφ
Since the eigen vector is known separately for each mode we find out the damping
ratio separately for each mode as follows.
As a first step we perform the operation {φ}T [C]{φ} for each mode.
For the first mode, we have

Cx −Cx Zc φxx
φxx φφx
−Cx Zc Cφx + Cx Zc2 φφx
which gives,

Cx φxx − Cx Zc φφx
φxx φφx
−Cx Zc φxx + (Cφx + Cx Zc2 )φφx
2
= Cx φxx − 2Cx Zc φφx φxx + (Cφx + Cx Zc2 )φφx
2
(2.2.36)
It should be realised that the above is a unique value and we also know that the
operation {φ}T [C]{φ} breaks up the equation to form 2Di ωi where i is the degrees of
freedom of the system.
Now considering,
2
2Di ωi = Cx φxx − 2Cx Zc φφx φxx + (Cφx + Cx Zc2 )φφx
2
,
for the first mode,
Cx φxx
2 − 2C Z φ φ + (C
x c φx xx φx + Cx Zc )φφx
2 2
D1 = (2.2.37)
2ω1
where D1 = damping ratio for the first mode and; ω1 = first natural frequency of the
foundation.
Similarly, for the second mode proceeding in same manner it can be proved that
Cφx φφx
2 − 2C Z φ φ
x c φx φφ + (Cφx + Cx Zc )φφφ
2 2
D2 = (2.2.38)
2ω2
Once the damping ratios are identified we assume, [C] = α[M] + β[K] and
performing the operation
{φ}T [C]{φ} = α{φ}T [M]{φ} + β{φ}T [K]{φ}, (2.2.39)

We have, for two degrees of freedom
2D1 ω1 = α + βω12 and 2D2 ω2 = α + βω22 (2.2.40)
Thus, we have two equation with two unknowns, α and β, and solving the above
two equations we get the value of α and β.
Once these values are known one can obtain an equivalent proportional soil damping
from the operation [C] = α[M] + β[K] which is now quite suitable for modal response
technique.
We now further explain the above method based on a suitable numerical problem.
Example 2.2.1
For a block foundation supporting a centrifugal pump was observed to have the
following design data M = 50 kN sec2 /m, J = 100 kN sec2 · m, Z c = 1.5 m,
K x = 3000 kN/m, K φ = 5000 kN/m, Cx = 200 kN/m, Cφ 350 kN/m.
Find out
• The natural frequencies in coupled horizontal and rocking mode.

• The normalized eigen vectors.
• Find out the approximate damping ratios for each mode.
• Correct the damping matrix based on equivalent Rayleigh damping.
Solution:
The complete equation of motion for the foundation under coupled rocking and
sliding mode is given by

+
0 Jxφ φ̈ −Cx Zc Cφx + Cx Zc2 φ̇

Kx −Kx Zc x P0
+ = sin ωm t
−Kx Zc Kφx + Kx Zc2 φ M0
Based on the above the equation of motion becomes

50 0 ẍ 200 −300 ẋ
+
0 100 φ̈ −300 988 φ̇

3000 −4500 x P0
+ = sin ωm t
−4500 11750 φ M0

• Calculation of the un-damped natural frequency
To find out the natural frequencies we have

3000 − 50λ −4500
=0
−4500 11750 − 100λ
➔ 35.25 × 106 −3 × 105 λ−5.875 × 105 λ+5000λ2 −20.25 × 106 = 0
The above equation on simplification reduces to
λ2 − 177.5λ + 3000 = 0

177.5 + (177.5)2 − 4 × 1 × 3000
λ1 = = 158.5 ➔ ω = 12.58 rad/sec
2

177.5 − (177.5)2 − 4 × 1 × 3000
and λ2 = = 18.92 ➔ ω = 4.35 rad/sec
2
For the first mode we have

3000 − 50 × 18.92 −4500 x
=0
−4500 11750 − 100 × 18.92 φ
Solving the above two homogeneous equations, considering x = 1.00, we

have, x : φ = 1.00 : 0.45644.
Similarly for the second mode we have

3000 − 50 × 158.5 −4500 x
=0
−4500 11750 − 100 × 158.5 φ
Solving the above two homogeneous equations, considering x = 1.00

We have, x : φ = 1.00 : −1.094
Thus, the complete eigen vector matrix becomes

x 1.00 1.00
=
φ i=1,2 0.45644 −1.094
Calculation of normalized eigen vectors

On operation {φ}T [M]{φ} we have
For the first mode

50 0 1.00
1.00 0.45644 = 70.83
0 100 0.45644
√
➔ Mr = 70.83 = 8.416


0.1188173
Thus {φ}i=1 =
0.0542329
For the second mode, on operation {φ}T [M]{φ}, we have

50 0 1.00
1.00 − 1.094 = 130.26
0 100 −1.094
√
➔ Mr = 169.6836 = 13.026

0.07676
Thus {φ}i=2 =
−0.08398
The normalized eigen vectors for the two modes are,

x 0.1188173 0.07676
=
φ i=1,2 0.0542329 −0.08398
Calculation of the modal damping ratios

Now if we perform the orthogonal operation with the complete normalized eigen
vector matrix it will be observed that while {φ}T [M]{φ} diagonalise to → [I] and
{φ}T [K]{φ} → [λ].
But as will be seen now that {φ}T [C]{φ} will NOT de-couple to the form 2Di ωi .

0.1188173 0.0542329 200 −300 0.1188173 0.07676
0.07676 −0.08398 −300 988 0.0542329 −0.08398

0.1188173 0.0542329 7.49359 40.546
=
0.07676 −0.08398 18.130205 −106.00024

1.8736 −0.9311626
=
−0.9473666 12.014236
Since the above matrix has off-diagonal terms will NOT be equal to zero,
hence we conclude that the matrix has not de-coupled due to orthogonal
transformation.
As such we treat the eigen vectors separately for each individual modes,
Thus for the first mode we have

200 −300 0.1188173
0.1188173 0.0542329 = 1.86308
−300 988 0.0542329
➔ 2D1 ω1 = 1.86308.
1.86308
or, D1 = = 0.214
2 × 4.35


200 −300 0.07676
0.07676 − 0.08398 = 12.0142
−300 988 −0.08398
12.0142
➔ D2 = = 0.4775.
2 × 12.58
Thus damping ratio is of the order of 21.4% for the first mode and 47.75%
for the second mode.
Correction to the damping matrix based on Rayleigh damping

Let us assume the damping matrix is of the form [C] = α[M] + β[K], then on
orthogonal transformation we have {φ}T [C]{φ} = α{φ}T [M]{φ} + β{φ}T [K]{φ},
which gives → 2D1 ω1 = α + βω12 and 2D2 ω2 = α + βω22 .
On substituting the respective values, we have
For the first mode, α + 18.92β = 2 × 0.214 × 4.35 → α + 18.92 = 1.8618
For the second mode, α + 158.5β = 2 × 0.4775 × 12.58 → α + 158.5β =
12.0139
Solving the above two equations we have → α = 0.48568 and β = 0.0727
As [C] = α[M] + β[K] we have

50 0 3000 −4500
[C] = 0.48568 + 0.0727
0 100 −4500 11750

242 −327.15
=
−327.15 902.795
is the modified damping matrix which satisfies the orthogonal property.

If we compare the value with the original damping matrix obtained from the
soil property we see each of the term has got slightly modified.
For practical design office calculation this is usually deemed sufficient. It at least
depicts a better result than no damping considered at all.
We will subsequently see how data based on modified damping matrix compare with
time history response which we had stated would be the most appropriate accurate
method that could be adopted with non-proportional damping.
But prior to that let us evaluate another form in which equations of motion for
coupled rocking and sliding motion can be formulated too.

2.2.6 Alternative formulation of coupled equation

of motion for sliding and rocking mode
We had already shown in Chapter 5 (Vol. 1) that other then formulating equation of
motion based on D’Alembert’s equation we can also write them down based on energy
concepts as put forward by Lagrange.
We use here Lagrange’s formulation to derive the equation of motion for coupled
rocking and sliding motion.
We had shown previously in chapter 5 (Vol. 1) that considering a conservative
system having kinetic energy as T and potential energy as U we have
n

d ∂T ∂T ∂U
d(T + U) = 0 and d(T + U) = − + dqi = 0 (2.2.41)
dt ∂ q̇i ∂qi ∂qi
i=1
For the machine foundation subjected to coupled rocking and sliding motion
1 1
Kinetic energy (T) = m(ẋ + Zc φ̇)2 + J φ̇ 2 and
2 2
1 1
The Potential Energy (U) = Kx x2 + Kφ φ 2
2 2
Based on the above equation we have
1 1
T= m(ẋ + Zc φ̇)2 + J φ̇ 2
2 2

∂T d ∂T
= m(ẋ + Zc φ̇) and = m(ẍ + Zc φ̈) = mẍ + mZc φ̈
∂ ẋ dt ∂ ẋ
1 1 ∂U
We have, U= K x x2 + Kφ φ 2 ; = Kx x
2 2 ∂x
1 1 ∂T
Also, T= m(ẋ + Zc φ̇)2 + J φ̇ 2 ; = mZc (ẋ + Zc φ̇) + J φ̇
2 2 ∂ φ̇

d ∂T
∴ = mZc ẍ + mZc2 φ̈ + J φ̈
dt ∂ φ̇
1 1 ∂U
For U= Kx x2 + K φ φ 2 ; = Kφ φ
2 2 ∂φ
Substituting the above values in the Lagrangian equation, we have
mẍ + mZc φ̈ + Kx x = 0; and mZc ẍ + (mZc2 + J)φ̈ + Kφ φ = 0 (2.2.42)

Thus writing in matrix form the free vibration equation becomes

m mZc ẍ K 0 x
+ x =0 (2.2.43)
mZc mZc2 + J φ̈ 0 Kφ φ
Equation (2.2.42) is known as dynamically or inertially coupled equation. You will

observe in contrary to the formulation based on D’Alembert’s equation, where the
mass matrix is diagonal here the stiffness matrix is diagonal while the mass matrix is
non-diagonal but symmetric.
For equation coupled by inertia it has been observed that both the damping and
stiffness matrix remains diagonal (Meirovitch 1975) and the complete equation of
motion thus becomes

m mZc ẍ C 0 ẋ K 0 x P0
+ x + x = sin ωm t
mZc mZc2 + J φ̈ 0 Cφ φ̇ 0 Kφ φ M0
(2.2.44)
Now that we have established the equation the question that obviously crops up
in mind is what is the advantage of this equation over the normal equation that was
derived based on static coupling/Barkan’s equation.
The first thing we will see subsequently that the eigen-values remain invariant with
this formulation.
Moreover it has been observed that damping ratio derived by this method are quanti-
tatively closer to the values derived from classical analysis based on frequency domain
analysis in complex domain. (Wolf 1988).
The reason for the better prediction of damping ratio could be that the damping
matrix derived by this formulation is in uncoupled form.
We now further explain the above based on suitable numerical example.
Example 2.2.2
For a block foundation as described in Example 2.2.1, calculate the following
based on Lagrange’s Formulation.
• The natural frequencies in coupled horizontal and rocking mode

• The normalized eigen vectors
• Approximate damping ratios for each mode.
All design parameters pertaining to design remains same as in Example 2.2.1.
Solution:
The complete equation of motion for the foundation under coupled rocking and
sliding mode based on Lagrange’s formulation is given by


m mZc ẍ C 0 ẋ K 0 x P0
+ x + x = sin ωm t
mZc mZc2 + J φ̈ 0 Cφ φ̇ 0 Kφ φ M0
Based on the above the equation of motion becomes

50 75 ẍ 200 0 ẋ 3000 0 x
+ +
75 212.5 φ̈ 0 350 φ̇ 0 5000 φ

P0
= sin ωm t
M0
To find out the natural frequencies we have

3000 − 50λ −75λ
= 0 which on expansion gives,
−75λ 5000 − 212.5λ
177.5 ± 139.66
5000λ2 − 8.875 × 105 λ + 15 × 106 = 0 → λ = :: λ1 =
2
18.92, ω1 = 4.35 rad/sec; λ2 = 158.5, ω2 = 12.58 rad/sec.
It should be observed that the natural frequencies are identical to the one
obtained in Example 2.2.1 based on static coupling.
Calculation of the eigen vectors

For the first mode we have

3000 − 50 × 18.92 −75 × 18.92 x
=0
−75 × 18.92 5000 − 212.5 × 18.92 φ

2054 −1419 x
➔ =0
−1419 979.5 φ
Solving the above homogenous equations considering, x = 1.00 we have

x : φ = 1.00 : 1.447
Similarly for the second mode

3000 − 50 × 158.5 −75 × 158.5 x
=0
−75 × 158.5 5000 − 212.5 × 158.5 φ

−4925 −11887.5 x
➔ =0
−11887.5 −28681.25 φ
Solving the above homogenous equations considering, x = 1.00 we have

x : φ = 1.00 : −0.4143.

Calculation of the normalized eigen-vectors

For the first mode

T 50 75 1.00
{φ1 } [M]{φ1 } = 1.00 1.447
75 212.5 1.447

= 711.878 ➔ Mr = 26.68

0.03748
Thus, the normalized eigen vector is {φ1N } = .
0.05423
For the second mode we have

50 75 1.00
{φ2 }T [M]{φ2 } = 1.00 − 0.4143
75 212.5 −0.4143

= 24.32945 ➔ Mr = 4.93248

0.20273
Thus, the normalized eigen vector is {φ2N } = .
−0.08399
Calculation of the damping ratios

For the first mode

T 200 0 0.03748
{φ1 } [C]{φ1 } = 0.03748 0.05423
0 350 0.05423
= 1.1632 ➔ 2D1 ω1 = 1.1632
1.1632
i.e. D1 = = 0.133.
2 × 4.35
For the second mode

200 0 0.20273
{φ1 }T [C]{φ1 } = 0.20273 −0.08399
0 350 −0.08399
= 10.688 ➔ 2D2 ω2 = 10.688
10.688
i.e. D2 = = 0.4248.
2 × 12.58
Thus
• for the first mode, the damping ratio is 13.3% and

• for the second mode, the damping ratio is 42.48%.

The table below gives comparative results based static and dynamic coupling
formulations for the examples solved above.
Natural Natural Normalized Normalized Damping Damping

Formulation frequency frequency mode shapes mode shapes ratio 1st ratio 2nd
basis 1st mode 2nd mode 1st mode 2nd mode mode mode
Static
coupling 4.35 12.58 0.1181 : 0.0542 0.0767 : −0.0839 21.4% 47.75%
Dynamic
coupling 4.35 12.58 0.0375 : 0.0542 0.2027 : −0.0839 13.3% 42.48%
Based on the above it would possibly be worthwhile to know how the amplitudes
vary based on the above two methods vis-à-vis the time history response which we
advocated as the most appropriate and correct method for handling responses having
non-proportional damping.
This is what we are going to establish based on suitable numerical example hereafter.
Example 2.2.3
For a block foundation as described in Example 2.2.1. Calculate the following
amplitude of vibration based on
• Static Coupling
• Dynamic coupling
• Time history response
• Discuss the results based on the three answers.
The unbalanced dynamic force is 200 KN having operating frequency of

750 r.p.m acting at a height 600 mm from the top of foundation.
The total height of the concrete block (L) is 1.8 m.
All other design parameters pertaining to design remains same as Example
2.2.1.
Solution:
The operating frequency of the machine is 750 r.p.m. = (750 × 2 × π )/60 =
78.53 rad/sec.
Solution based on static coupling

The equation of motion in matrix form is given by
[M]{Ẍ} + [C]{Ẋ} + [K]{X} = {P}

P0 200
Here, the force matrix is given by = sin 78.53t
M0 180

For static coupling,

The normalized eigen-vector obtained earlier was

0.1188173 0.076746
[φ] =
0.0542333 −0.0839942
Considering the operation, [φ]T {P} we have

0.1188173 0.0542333 200
[φ]T {P} = sin 78.53t
0.076746 −0.0839942 180

33.52
= sin 78.53t
0.2293
On orthogonal transformation we have
{φ}T [M]{φ}{ξ̈ } + {φ}T [C]{φ}{ξ̇ } + {φ}T [K]{φ}{ξ } = {φ}T {P}
This in decoupled form reduces to

{ξ̈1 } + 2D1 ω1 {ξ̇1 } + [ω12 ]{ξ1 } = p0 sin ωm t;
{ξ̈2 } + 2D2 ω2 {ξ̇2 } + [ω22 ]{ξ2 } = m0 sin ωm t
Substituting the different values, we have
{ξ̈1 } + 2 × 0.214 × 4.35{ξ̇1 } + 18.92{ξ1 } = 33.52 sin 78.53t and

{ξ̈2 } + 2 × 0.4776 × 12.58{ξ̇2 } + 158.5{ξ2 } = 0.230 sin 78.53t
33.53 sin 78.53t

∴ ξ1 =
(18.92 − 78.532 )2 − (1.862 × 78.53)2
= 5.45061918 × 10−3 sin 78.53t
0.230 sin78.53t
ξ2 =
(158.5 − 78.532 )2 − (12.016 × 78.53)2
= 3.816235259 × 10−5 sin 78.53t
Thus back transferring on global co-ordinate we have

x 0.1188173 0.076746 5.45061918
= × 10−3 sin 78.53t
φ 0.0542333 −0.08399942 0.03816235259

0.6505566
= × 10−3 sin 78.53t
0.2923994
Solution based on dynamic coupling


0.0374669 0.2027373
[φ] =
0.0542333 −0.0839942

Considering the operation [φ]T {P} we have

T 0.0374669 0.0542333 200
[φ] {P} = sin 75.53t
0.2027373 −0.0839942 180

17.255374
= sin 78.53t
25.428504
On orthogonal transformation we have
[φ]T [M][φ]{ξ̈ } + [φ]T [C][φ]{ξ̇ } + [φ]T [K][φ]{ξ } = [φ]T {P}
{ξ̈1 } + 2D1 ω1 {ξ̇1 } + [ω12 ]{ξ1 } = p0 sin ωm t;

{ξ̈2 } + 2D2 ω2 {ξ̇2 } + [ω22 ]{ξ2 } = m0 sin ωm t
Substituting the different values we have
{ξ̈1 } + 2 × 0.1505967 × 4.35{ξ̇1 } + 18.92{ξ1 } = 17.25537 sin 78.53t and

{ξ̈2 } + 2 × 0.4248 × 12.58{ξ̇2 } + 158.5{ξ2 } = 25.429054 sin 78.53t
17.25537 sin 78.53t

∴ ξ1 =
(18.92 − 78.532 )2 − (1.3102 × 78.53)2
= 2.806253 × 10−3 sin 78.53t
25.428504 sin 78.53t

ξ2 =
(158.5 − 78.532 )2 − (10.869 × 78.53)2
= 4.1900494 × 10−3 sin 78.53t

x 0.0374669 0.2027373 2.806253
= × 10−3 sin 78.53t
φ 0.0542333 −0.08399942 4.1900494

0.9546209
= × 10−3 sin 78.53t
−0.1997474
Figure 2.2.7 gives a comparison of the amplitude value for the static and
dynamic coupling case.
It will be observed that
• Dynamically coupled equation gives slightly higher amplitude than statically

coupled equations.
• This is applicable for both translation and rotation.

1.50E-03
1.00E-03 Translation Based

on Static coupling
Amplitude of vibration
5.00E-04 Rotation based on

static coupling
0.00E+00
Translation Based
1 20 39 58 77 96 11 5 13 4 on Dynamic
-5.00E-04 coupling
Rotation based on
dynamic coupling
-1.00E-03
-1.50E-03
Time steps
Figure 2.2.7 Comparison of response values static versus dynamic coupling.
Next we compare these results with time history response where we treat the
complete mass, damping and stiffness matrix in uncoupled form and integrate
directly the equation
[M]{Ẍ} + [C]{Ẋ} + [K]{X} = {P sin ωM t}8
Figure 2.2.8 shows a very interesting result pertaining to time history vis-à-vis
modal response technique based on dynamically coupled equation.
The time history technique used has been Wilson-θ method having a time step
of 0.0075 seconds and response has been calculated to 500 steps.
It will be observed that in modal response technique we have ignored the
transient response part and have only found out the response based on the steady
state part, while, the step by step integration considers both the transient and
the steady state responses.
In comparison to a steady state response of 1 mm the time history starts with
peak amplitude of approximately 5 mm in step 18 and slowly converges to a
value near to 1 mm at about 295th step and becomes steady after that9 .
8 Refer Chapter 5 (Vol. 1) the topic of Time history/step by step integration.

9 Looking at the displacement value of 1 and 5 mm hardend professionals could well frown or smirk.
But herein the example is deliberately designed like this to give beginners and especially students
a quantitative feel which eludes many when working in micron level. Real life problems would
come surely latter in the chapter.

It will be observed that the values are quite closely matching at the steady
state position with step by step integration giving slightly higher values than
dynamically coupled modal response.
The initial response due to the transient part of the time history analysis is
significant (about 5 mm), as this decays down quickly after some time (here
about 2.0 seconds after the start) will really not have much effect on the over all
behavior of the foundation as such, but for pipes and nozzles rigidly connected
to the machine this initial high amplitude of 5 mm can have significant effect
and if proper care is not taken may induce severe reversal of stress and may even
induce failure.
6.00E-03
5.00E-03
4.00E-03
3.00E-03
Amplitude
2.00E-03 Translation Modal response

Translation based on time history
1.00E-03
0.00E+00
-1.00E-03
1 52 103 154 205 256 307 358 409 460
-2.00E-03
-3.00E-03 Time steps
Figure 2.2.8 Comparison of response time history versus modal response.
2.3 TRICK TO BY PASS DAMPING – MAGNIFICATION

FACTOR, THE KEY TO THE PROBLEM. . .
A perfectionist may not like the methodology proposed regarding approximate esti-
mate of the damping ratio based on individual mode. At the same time he might argue
that for secondary equipment like medium or small capacity pumps doing time history
analysis is too intense and not really called for.
Fair enough, for the argument is not without some sanctity so how do we tackle
this riddle?
The most logical solution to the above problem could be that if we can create a
condition where damping plays a negligible effect compared to un-damped situation
then we can surely ignore damping from our basic equation and arrive at a result
which is as good an answer with damping and the problem is solved.
So the next obvious query will be, what is this condition which will give an invariable
answer irrespective of damping taken or not?
Before we describe this condition it would be worthwhile to understand what the
magnification factor is.

Consider Figure 2.3.1. We had seen earlier that equation of motion for a body
having damped single degree of freedom is given by
P0
Kz sin ωm t
δz = (2.3.1)
(1 − r2 )2 + (2Dz r)2
where r = ωm /ωn (frequency ratio) and Dz = damping ratio

In the above equation the denominator is known as the magnification factor.
1
i.e. M.F. = (2.3.2)
(1 − r2 )2 + (2Dz r)2
Based on the above expression we define magnification factor as a factor which

gives us a measure of how many times the dynamic amplitude gets magnified over the
equivalent static deflection of P0 /K.
Now the above is a very interesting equation for if we plot the above equation for var-
ious values of damping against the frequency ratio we get curve as shown in Fig. 2.3.1.
In the below curve it will be observed that when the frequency ratio (ωm /ωn ) is
about 3.0, all the lines irrespective of whatever is the damping converges nearly to
a single line.
Now what does this signify?
It implies that when the frequency ratio is more than 3.0 irrespective of the damping
ratio the magnification factor do not change and thus the damping ratio plays practi-
cally no part in the overall response of the system. The usual practice is that when the
frequency ratio is more than 3.5 the damping effect of the soil is neglected.
7 Mag. Factor for

5% damping
6
Magnification factor
Mag. Factor for

10% damping
5
Mag. Factor for
4 15% damping
Mag. Factor for
3 25% damping
Mag. Factor for
2 37.5% damping
0
15
95
75
55
35
15
95
75
5
5
55
35
15
95
75
4.5
5.3
0.
0.
0.
1.
1.
2.
2.
2.
3.
3.
4.
4.
5.
Frequency Ratio
Figure 2.3.1 Magnification factor for various damping ratio.

One can immediately draw conclusion from the above fact that when a block
foundation is resting on ground having soft to medium soil supporting machines hav-
ing high operating speed it would possibly be quite justified to neglect the damping
effect of soil.
However, for low tuned machine it has been observed that it is difficult to achieve
this frequency separation for block foundation and for such cases damping cannot be
ignored.
We now further explain the above with a suitable numerical example.
Example 2.3.1
For the block foundation as described in Example 2.2.1. Calculate the following
amplitude of vibration based on undamped equation of motion based on
• Static Coupling
• Dynamic coupling
• Discuss the results based on the answers.
The unbalanced dynamic force is 200 kN having operating frequency of

750 r.p.m acting at a height 600 mm from the top of foundation.
The total height of the concrete block (L) is 1.8 m
All other design parameters pertaining to design remains same as
Example 2.2.1
Solution:
750
The operating frequency of the machine is 750 r.p.m. = ×2×π =
60
78.53 rad/sec
Solution based on static coupling
formis given by [M]{Ẍ} + [C]{Ẋ} + [K]{X} =

The equation of motion in matrix
P0 200
{P}; the force matrix is = sin 78.53t
M0 180
For Static coupling

0.1188173 0.076746
[φ] =
0.0542333 −0.0839942
Considering the operation [φ]T {P}, we have

0.1188173 0.0542333 200
[φ]T {P} = sin 78.53t
0.076746 −0.0839942 180

33.52
= sin 78.53t
0.2293

We had seen earlier that the two natural frequencies are ω1 = 4.35 rad/sec;
ω2 = 12.58 rad/sec
78.53
Thus the frequency separation is r = = 6.24 > 3.5
12.58
Thus we can say that as the frequency ratio is greater than 3.5 damping effect
of the soil can be neglected.
Thus, with orthogonal transformation and neglecting the damping, we have
[φ]T [M][φ]{ξ̈ } + [φ]T [K][φ]{ξ } = [φ]T {P}
{ξ̈1 } + [ω12 ]{ξ1 } = p0 sin ωm t; {ξ̈2 } + [ω22 ]{ξ2 } = m0 sin ωm t
or, {ξ̈1 } + 18.92{ξ1 } = 33.52 sin 78.53t and

{ξ̈2 } + 158.5{ξ2 } = 0.230 sin 78.53t
33.53 sin 78.53t

∴ ξ1 = = 5.4537698 × 10−3 sin 78.53t
(18.92 − 78.532 )2
0.230 sin 78.53t

ξ2 = = 3.8279354 × 10−5 sin 78.53t
(158.5 − 78.532 )2

x 0.1188173 0.076746 5.4537698
= × 10−3 sin 78.53t
φ 0.0542333 −0.08399942 0.038273954

0.6509395
= × 10−3 sin 78.53t
0.2925838
Solution based on dynamic coupling


0.0374669 0.2027373
[φ] =
0.0542333 −0.0839942
Considering the operation [φ]T {P} we have

0.0374669 0.0542333 200
[φ]T {P} = sin 75.53t
0.2027373 −0.0839942 180

17.255374
= sin 78.53t
25.428504

As we have seen earlier that the frequency separation is greater than 3.5, hence
on orthogonal transformation without damping we have
[φ]T [M][φ]{ξ̈ } + [φ]T [K][φ]{ξ } = [φ]T {P}
{ξ̈1 } + [ω12 ]{ξ1 } = p0 sin ωm t; {ξ̈2 } + [ω22 ]{ξ2 } = m0 sin ωm t
{ξ̈1 } + 18.92{ξ1 } = 17.25537 sin 78.53t and
{ξ̈2 } + 158.5{ξ2 } = 25.429054 sin 78.53t
17.25537 sin 78.53t

∴ ξ1 = = 2.8066453 × 10−3 sin 78.53t
(18.92 − 78.532 )2
25.428504 sin 78.53 t

ξ2 = = 4.2321161 × 10−3 sin 78.53t
(158.5 − 78.532 )2

x 0.0374669 0.2027373 2.8066453
= × 10−3 sin 78.53t
φ 0.0542333 −0.08399942 4.2321161

0.963164
= × 10−3 sin 78.53t
−0.2007444
On comparing the amplitudes for the damped and undamped case we have
the following results.
Damped amplitude Undamped amplitude Difference in (%)
Sl. No. Formulation basis X (10−3 ) φ (10−3 ) X (10−3 ) φ (10−3 ) X φ
1 Static coupling 0.650525 0.2924 0.6509 0.29258 0.06 0.06

2 Dynamic coupling 0.9546 −0.1997 0.9631 −0.20074 0.88 0.518
2.4 EFFECT OF EMBEDMENT ON FOUNDATION
The theories described above dominated the scenario of design of machine founda-
tion for quite sometime. But as the machines designed were progressively becoming
heavier and having higher and higher operating frequencies the foundations in turn

were becoming more and more massive in nature10 , and it was realized that when
a foundation is constructed below ground level the surrounding soil in which it is
embedded plays a significant role on the overall response of the foundation and needs
to be carefully evaluated too.
Number of theoretical formulations have been derived and field experiments (Gupta
1972, Erden & Stokoe 1975) have been conducted to study the embedment effect
of soil on the overall response of the foundations, though there exists disagreements
between the theories put forward however the general consensus about the embedment
of soil on the foundation both from theoretical and field observations are as follows:
• The embedment effect increases the natural frequency of the foundation

• It reduces overall amplitude of the foundation.
It is not difficult conceive from the above statements that:
• Embedment effect increases the soil stiffness and

• Also has an incremental effect on the damping of the soil.
The most popular theory which is in practice in design office is given

Tables 2.4.1 and 2 (Whitman 1972).
Here, h = depth of embedment of the foundation in the surrounding soil; υ =
Poisson’s ratio of the soil.
Table 2.4.1 Embedment coefficients for spring constants.
Equivalent
Sl. No. Direction Coefficient radius Remarks

h LB
1 Vertical ηz = 1 + 0.6(1 − υ) rz = This is in vertical Z direction
rz π

h LB
2 Horizontal ηx = 1 + 0.55(2 − υ) rx = This induce sliding in horizon-
rx π tal x or y direction

3
h 4 LB
3 Rocking ηφx = 1 + 1.2(1 − υ) rφx = This produces rocking about
rφx 3π Y axis
3
h
+ 0.2(2 − υ)
rφx

h 4 L3 B
3.1 Rocking ηφy = 1 + 1.2(1 − υ) rϕy = This produces rocking about
rφy 3π X axis
3
h
+ 0.2(2 − υ)
rφy
4 Twisting None available This produces twisting about
vertical Z axis
10 In most of the cases as the plan area of the foundation is dependent on the equipment general arrangement
to increase the mass it was getting deeper and deeper.

Table 2.4.2 Embedment coefficients for soil damping ratio.
Sl. No. Direction Coefficient Equivalent radius Remarks

1 + 1.9(1 − υ) rhz LB
1 Vertical αz = √ rz = ηz is value as obtained
ηz π
as coefficient for
soil spring constant

1 + 1.9(2 − υ) rhx LB
2 Horizontal αz = √ rx = ηx is value as obtained
ηx π
as coefficient for
3
1 + 0.7(1 − υ) rφx
h
+ 0.6(2 − υ) h
rφx 4 LB3
3 Rocking αφx = √ rϕx = ηφx is value as obtained
ηφx 3π
as coefficient for
3
1 + 0.7(1 − υ)
h
+ 0.6(2 − υ) rφy
h
rφy 4 L3 B
3.1 Rocking αφy = √ rφy = ηφy is value as obtained
ηφy 3π
as coefficient for
4 Twisting None available
It is suggested that if we multiply the spring constants available from Richart and
Lysmer formulation vide Tables 2.2.3 and 4 by the above factors we get the modified
spring constants valid for the embedded foundations.
Damping ratio as obtained from Richart and Lysmer’s model when multiplied by
the coefficients as furnished in Table 2.4.2 gives the damping ratio considering the
embedment effect of the soil.
2.4.1 Novak and Beredugo’s model

Novak & Beredugo (1972a) model for embedded foundation has already been worked
out in detail in the chapter 5 (Vol. 1) under soil dynamics and you may refer to
the same for further details. Both vertical and lateral mode coupled with rocking
(Novak & Beredugo 1972b) has been treated therein.
2.4.2 Wolf’s model

Wolf (1985) has devised springs for dynamic foundation where he has con-
sidered additional soil mass vibrating with the foundation effect as shown in
Table 2.4.3.
2.5 FOUNDATION SUPPORTED ON PILES
In many cases due to poor soil condition machine foundations are loaded on piles and
obviously other than static loads they are also subjected to vibrations and dynamic
loading. Dynamic behaviour of piles is still to certain extent not very clearly understood
though theoretical formulas exist to predict their behaviour under time dependent

Table 2.4.3 Soil spring constants as per Wolf (1985).
Mode Spring stiffness γ0 μ0

4Gr 0
Vertical 0.58 0.095
1−υ
8Gr 0
Horizontal 0.85 0.27
1−υ
8Gr 3θ 0.3
Rocking 0.24
3(1 − υ) 1+ 3(1−υ)m
ρ
8r 5θ
16Gr 3ψ 0.433 m
Torsion 0.045
3 1+ 2m
rψ5 ρ
rψ5 ρ
2
In which, C = Vsr kγ0 and m = Vsr kμ0 , where r = equivalent radius and
shall be r0 , rθ , rψ as the case may be; G = Dynamic shear modulus of the soil;
ρ = mass density of the soil; vs = shear wave velocity of the soil; M = mass of
the soil participating in the vibration with the machine and the block foundation,
and C = damping of the soil.
loading; they have been co-related with field observations for only a few simplified
cases.
As such the decision of using piles below machine foundations should be taken
cautiously and not without some understanding of how it would behave under the load
induced from the machine. Though there are very few reports on the field observation
data on dynamic behaviour of piles under machine foundations it is however generally
accepted that under time dependent loads piles,
• Have significant influence on the amplitude, especially near resonance,

• Increases the natural frequency of the system,
• Decrease the geometric damping of the soil foundation system.
Since in some cases particularly in lateral mode, the effect of piles could be adverse
we repeat that it should be used with caution.
For machine foundation on piles three mathematical models are usually in vogue.
1 Piles considered as frequency independent equivalent springs based on Novak’s

(1974) formulation.
2 Piles considered as beam elements connected to soil springs based on Richart’s
formula as mentioned earlier.
3 Considering the underlying soil and the pile as finite elements and executing a
detailed analysis based on appropriate boundary conditions.
We will discuss all the above methods now in some detail but would like to emphasise
at this point is that each one of them has its pros and cons and are not self-sufficient.
As such which would be the most appropriate model for analysis varies from case to
case and one method of analysis may have to be complimented by another model.
While describing the model we start with the most exhaustive one and go the reverse

Figure 2.5.1 Machine foundation supported on piles.
way for we feel this will give you a better insight to the various problems that exists
with dynamic behaviour of piles.
Shown in Figure 2.5.1 is a machine foundation supported by piles.
2.5.1 Pile and soil modelled as f inite element

This would obviously the most exhaustive model one could perceive and is shown in
Figure 2.5.2.
A representative and conceptual 2D model of the pile and soil are shown in
Figure 2.5.2.
Actually, the most appropriate model would be in 3D, where the piles are modelled
as beam elements while the soil can be modelled as eight nodded brick element and a
comprehensive dynamic analysis of the whole system could be performed.
It can however be perceived that the analysis would time consuming and expensive
(both in terms of man-hours and input data generation) and is usually not warranted
except for large multi-shaft gas/steam turbines of high output resting on poor soil
where such analysis could become essential in order to ensure the performance of such
expensive machines under operating conditions.

Rigid body lumped as mass
Pile Cap modeled as beam element
Master Node
Slave Node
Figure 2.5.2 Finite element model of machine foundation with pile and soil.
One of the major disadvantages with this type of model is that the boundary of the
soil has to be extended to substantial distance away both at the sides and from the
pile tip in vertical direction enabling the model to predict correctly the response of the
system. If this boundary limit is inadequate from the pile tip then waves transmitted
to the soil due to the vibration of the machine will get reflected back and result in
spurious responses which could make the analysis completely wrong.
The question as to how far this boundary should extend, no rational basis has been
derived yet and is completely up to the engineer’s judgement11 .
11 One thumb rule is to extend the boundary in vertical direction to 2.5 times the length of the pile.

Other than this there are certain practical problems encountered especially when
the piles are long (say 20/30 m), geotechnical data may not be available to the depth
to which an engineer might like to extend the boundary of the problem and as such
if comprehensive soil data to the desired level is not available it may be difficult to
model the system without the adequate data.
In spite of the above problems the model is not without its advantage and may be
summarised as follows:
• It comprehensively caters to the 3D effect of the pile soil and the foundation
• It can effectively model the soil if layered in nature where each of the layers has
different material property.
• The group interaction effect of soil and pile is automatically catered for.
• Piles having variable geometry (tapered piles) can also be modelled without any
problems.
• If battered piles are provided to counter any lateral thrust can also be modelled
without any difficulty.
2.5.2 Piles modelled as beams supported on elastic springs

In this method the piles are modelled as beam elements connected to springs in hor-
izontal and vertical direction calculated out of the soil material property as shown
Figs. 2.5.3 and 4. Again for a pile cap supporting machines a 3D model will be
developed and dynamic analysis will be carried out.
Shown in Figure 2.5.5 is a 3D model of pile cap, pile with soil springs. In this case
the soil springs may be calculated based on Richart’s formulation or by multiplying
the influence area of each node by the coefficient of uniform compression.
The piles are modelled as beam elements having 6 degrees of freedom at each node.
The pile cap is mathematically modelled as either beam or plate bending element
depending upon the overall aspect ratio of the cap and other design considerations.
Figure 2.5.3 Pile in soil. Figure 2.5.4 Mathematical model.

Figure 2.5.5 Mathematical model of pile cap, piles with soil springs.
It is obvious that with respect to the previous model one of the major advantages is
that it is a relatively less laborious model in terms of input generation and complexity
and many engineers prefer to use this in lieu of a detailed finite element 3D model as
shown previously.
However the above model suffers from one serious lacuna for which it should be
used with caution.
The model in Figure 2.5.5 does not take into cognisance the effect of the soil which
lies between the two piles and treats the soil as only discrete element based on springs.
• This could significantly under rate or even over rate the dynamic response which
depends on the nature of the soil
• It does not take in to cognisance the pile group interaction factor which has been
observed to have significant effect on the dynamic response on the system specially
when the pile spacing is between 2.5D to 3D, where D is the overall diameter of
the pile.
It is recommended that this model may be used when the centre to centre distance
between the piles are at least more than 5D.
2.5.3 Novak’s (1974) model for equivalent spring

stiffness for piles
This is possibly the most popular method used in the design offices to evaluate springs
for piles subjected to dynamic loads and will be discussed in some detail.

Though not without limitations the major advantage with this method is that
• It is simple to use.
• The spring stiffness and damping values are frequency independent.
• The group interaction effect of the piles can be to certain extent taken care of.
• The spring and damping values thus obtained can be very easily implemented as
linear springs in commercially available finite element software.
• Standard Chart and coefficients exists for piles that are quite easy to use.
Novak’s method has found to be in excellent agreement with field observations

specially when the pile group arrangement is not complicated.
2.5.4 Equivalent pile springs in vertical direction

For single end bearing piles undergoing vertical motion, the spring constants are given
by the expression

Ep Ap
kbz = f18,1 (2.5.1)
r0
where, kb z = equivalent spring constant for end bearing piles; Ep = Young’s modulus
of pile material; Ap = cross sectional area of the pile; r0 = equivalent radius of the
pile, and, f18,1 = a factor which depends on pile material (concrete, steel, timber etc.),
ratio of embedded length l to radius (r0 ) and Vs /Vc (shear wave velocity of the soil
above the tip to compression wave velocity in pile).
The damping value in vertical direction is given by

Ep A p
cbz = f18,2 (2.5.2)
vs
where, cbz = damping value of the end bearing

piles; vs = shear wave velocity of
the pile through which the soil is driven ( Gs g/γs ); f18,2 = is a factor as given in
Table 2.5.1.
Table 2.5.1 Values of factor f -as per Novak (1974) for stiffness
and damping factor for single pile. For concrete piles
(γs /γp = 0.7) having /r0 > 25.
Slenderness ratio Stiffness and damping function f for vertical bearing pile
20 f18,1 = 3.75(Vs /Vc )2 − 0.05(Vs /Vc ) + 0.0501

f18,2 = 15.345(Vs /Vc )2.0928
50 f18,1 = 6.25(Vs /Vc )2 + 0.05(Vs /Vc ) + 0.0199
f18,2 = −10(Vs /Vc )2 + 1.5(Vs /Vc ) − 0.012
100 f18,1 = −3.75(Vs /Vc )2 + 0.45(Vs /Vc ) + 0.0061
f18,2 = 1.4(Vs /Vc ) − 0.0083

The values of f18,1 and f18,2 are meant for end bearing piles. However it has been
observed that for friction piles having l/r0 greater than 60 or Vs /Vc greater than 0.03
these values are in small error pertaining to timber and concrete piles.
For steel piles Novak has given a value of f18,1 = 0.030 and f18,2 = 0.045 where
Vs /Vc = 0.033 and l/r0 greater than 80.
For relatively short friction piles the following expression has been suggested by
Novak for calculation of the stiffness and damping

Ep A p Ep Ap
kz1 = f18,1 and cz1 = f18,2 (2.5.3)
r0 vs
Table 2.5.2 Vertical stiffness coefficients for floating piles as per Novak (1983).
f 18,1
L/R Ep /G = 10,000 Ep /G = 2500 Ep /G = 1000 Ep /G = 500 Ep /G = 250
10.8696 0.0021 0.0052 0.0104 0.0187 0.0332

21.7391 0.0031 0.0083 0.0166 0.0301 0.0509
32.6087 0.0042 0.0104 0.0218 0.0364 0.0571
43.4783 0.0042 0.0125 0.0260 0.0405 0.0582
46.7391 0.0052 0.0135 0.0270 0.0416 0.0582
54.3478 0.0052 0.0145 0.0281 0.0416 0.0582
65.2174 0.0062 0.0166 0.0291 0.0416 0.0582
76.0870 0.0062 0.0177 0.0301 0.0416 0.0582
86.9565 0.0073 0.0187 0.0301 0.0416 0.0582
100.0000 0.0083 0.0197 0.0301 0.0416 0.0582
Table 2.5.3 Vertical damping coefficients for floating piles as per Novak (1983).

f18,2
L/R Ep /G = 10,000 Ep /G = 2500 Ep /G = 1000 Ep /G = 500 Ep /G = 250
10.8696 0.0032 0.0126 0.0295 0.0558 0.1032

16.3043 0.0053 0.0179 0.0421 0.0695 0.1137
21.7391 0.0074 0.0232 0.0495 0.0811 0.1126
27.1739 0.0084 0.0263 0.0537 0.0832 0.1095
32.6087 0.0105 0.0305 0.0568 0.0811 0.1053
38.0435 0.0116 0.0326 0.0589 0.0789 0.1021
43.4783 0.0137 0.0347 0.0579 0.0758 0.0989
48.9130 0.0147 0.0368 0.0568 0.0737 0.0979
54.3478 0.0147 0.0379 0.0558 0.0726 0.0979
59.7826 0.0168 0.0379 0.0537 0.0716 0.0979
65.2174 0.0168 0.0379 0.0526 0.0705 0.0979
70.6522 0.0179 0.0379 0.0516 0.0695 0.0979
76.0870 0.0189 0.0379 0.0516 0.0695 0.0979
81.5217 0.0189 0.0368 0.0505 0.0695 0.0979
86.9565 0.0200 0.0358 0.0505 0.0695 0.0979
92.3913 0.0211 0.0358 0.0505 0.0705 0.0989
100.0000 0.0211 0.0337 0.0495 0.0705 0.0989


where, f18,1
and f18,2 are stiffness and damping factors respectively as given
Tables 2.5.2 and 2.5.3 respectively.
It has been suggested that the values given in these charts are most appropriate
when used for the range of a0 = 0.1 to 0.8, where a0 = 2πvfr
s
0
here f = the operating
frequency of the machine. It has also been suggested (Steven 1978) that these values
are even valid for a0 as low as 0.05 which means that reasonably good results may be
expected for even slender piles and low frequencies.
The damping ratio for the pile may be calculated from the expression

Dz1 = cz1 /2 kz1 mc (2.5.4)
where mc is the mass of the cap plus machinery or the portion of structure which
is vibrating in the same phase as the cap. Part of the mass of the pile may also
be included in the above equation but it has been generally found that this ratio
of the pile mass to the mass of the supported weight is very small and is usually
ignored.
2.5.5 The group effect on the vertical spring

and damping value of the piles
Piles in practise usually do not occur as a single pile and usually under a foundation
block there will be a number of piles to form a foundation. For instance under a normal
block foundation supporting some machinery the pile system could be something like
as shown in Figure 2.5.6.
We had shown in Section 2.5.4 as to how to calculate the spring stiffness for indi-
vidual single piles. When we try to find out the equivalent spring stiffness for the
pile group as shown below this does not constitute of the sum of the individual pile
Figure 2.5.6 Plan view of a block foundation supported over piles.

stiffness. It also depends upon the relative distance between the piles itself and the
slenderness ratio of the piles carrying the loads and is expressed as

g

N
N
kz = kz1 αA (2.5.5)
1 1
The equivalent damping for the pile group is given by

g

N
N
cz = cz1 αA (2.5.6)
1 1
where, N = number of piles in a group; αA = displacement interaction factor (axial)

for a typical reference pile in the group relative to itself and to all other piles in the
group assuming the reference pile and all other pile are loaded to same magnitude,
and the factor αA can be evaluated from the expression (Randolph and Poulos 1982);
and also recommended by API 351R.
0.5 ln(lp /s)

αA = for s ≤ lp (2.5.6a)
ln(lp dρA )
Here lp = Pile length, s = spacing of piles, d = diameter of pile ρA = Gav /Gb ;

Gav = Average Shear modulus along pile depth and Gb = Shear modulus at pile base.
Alternatively the value can also be deduced from Poulos’s interaction curve for static
interaction (Poulos and Davis 1980).
2.5.6 Effect of pile cap on the spring

and damping stiffness
If the pile cap is not in contact with the ground the above equations can be directly
used in for the analysis. Pile caps embedded usually have a favourable effect on the
response of the group and should be adapted wherever possible.
It would be realistic to assume that the embedment effect generates only the side
friction between the cap and the soil and that to only when dense granular backfill is
used. For the soil below the pile cap which is likely to be of inferior quality can settle
away from the cap for non-cohesive soil, similarly for cohesive soil this can shrink
away from the sides of the pile cap and can become ineffective.
Table 2.5.4 Values of S̄1 and S̄2 for various Poisson’s ratio.
Poisson’s ratio S̄1 S̄2
0.0 2.7 6.7

0.25 2.7 6.7
0.40 2.7 6.7

The expressions for calculating the stiffness and the damping constant for an
embedded cap of embedd depth h is given by

f f
kz = Gs hS̄1 and cz = hr0 Gs λs /g S̄2 (2.5.7)
Equation (2.5.7) should be added to the pile stiffness and damping of the pile group
as presented in the previous section to arrive at the complete spring and damping
constant of a pile group in vertical direction. Values of S̄1 and S̄2 for various Poisson’s
ratio are given in Table 2.5.4.
2.5.7 Equivalent pile springs and damping

in the horizontal direction
For vibration in horizontal direction the expression for stiffness and damping is as
shown below:

Ep I p Ep I p
kx1 = f11,1 for /r0 ≥ 25, and cx1 = f11,2 for /r0 ≥ 25.
r30 r20 vs
(2.5.8)
Here Ip is the moment of inertia of the pile cross section about the centroidal axis
perpendicular to the direction of the motion. Here x direction depicts the horizontal
motion and f11,1 and f11,2 are factors for fixed headed piles. The values of f11,1 and
f11,2 are furnished in Table 2.5.5.
The group effect is expressed as
N N
g 1 kx1 g cx1
kx = N and cx = 1N (2.5.9)
1 αL 1 αL
Table 2.5.5 Values of factor f -as per Novak (1974) stiffness and damping factors for horizontal and
rocking mode.
Poisson’s ratio Function f
0.25 f11,1 = 7.25(Vs /Vc )2 + 0.38(Vs /Vc ) − 0.0013

f11,2 = 17(Vs /Vc )2 + 0.915(Vs /Vc ) − 0.0032
f7,1 = −55(Vs /Vc )2 + 9.3(Vs /Vc ) + 0.1075
f7,2 = −38.75(Vs /Vc )2 + 6.55(Vs /Vc ) + 0.0734
f9,1 = −1.81(Vs /Vc )
f9,2 = 0.375(Vs /Vc )2 − 2.67(Vs /Vc ) + 0.0005
0.4 f11,1 = 7.875(Vs /Vc )2 + 0.43(Vs /Vc ) − 0.0015
f11,2 = 18.75(Vs /Vc )2 + 1.02(Vs /Vc ) − 0.0037
f7,1 = −57.5(Vs /Vc )2 + 9.65(Vs /Vc ) + 0.1113
f7,2 = −41.25(Vs /Vc )2 + 6.85(Vs /Vc ) + 0.0746
f9,1 = −1.94(Vs /Vc )
f9,2 = 0.75(Vs /Vc )2 − 2.87(Vs /Vc ) + 0.0006

Table 2.5.6 Values of Su1 and Su2 for various Poisson’s ratio.
Poisson’s ratio S̄u1 S̄u2
0.0 3.6 8.2

0.25 4.0 9.1
0.40 4.1 10.6
where, αL = a displacement factor for lateral motion defined in similar way to αA and
is given by12
r
1 0
αLf = 0.6ρc [Ep Gc ] 7 (1 + cos2 βp ) (2.5.9a)
s
r
1 0
αLH = 0.4ρc [Ep Gc ] 7 (1 + cos2 βp ); 2
αθH = αLH 3
, αθM = αLH (2.5.9b)
s
Here ρc = Gz /Gav where Gz = Shear Modulus at depth lc /4

2
lc = 2r0 [Ep /Gc ] 7 and is known as the critical length of the pile where Gc = Average
shear modulus over the critical length of the pile.
αLf = The horizontal interaction factor for fixed headed piles (no head rotation).
αLH = The horizontal interaction factor due to horizontal force (rotation allowed).
αθ H = Interaction factor due to horizontal force for rotation.
αθ M = Interaction factor due to moment for rotation.
βp = Angle subtended by a pile in pile group with respect to the reference pile.
When the calculated
√ interaction factor α exceeds 1/3, its value needs to be replaced
by α = 1 − 2/ 27α, a correction made to avoid α approaching infinity as s tends to
zero.
Alternatively Poulos’s interaction curve for static load case under horizontal load
may also be used.
The stiffness and damping characteristics of the pile cap is expressed as
f f
kx = Gs hS̄u1 and cx = hr0 S̄u2 Gs γs /g (2.5.10)
The factors S̄u1 and S̄u2 are as given in Table 2.5.6.
2.5.8 Equivalent pile springs and damping in rocking motion

The expression for spring stiffness and damping for simple pile has been expressed as

Ep I p Ep Ip
kψ1 = f7,1 ; cψ1 = f7,2 (2.5.11)
r0 vs
12 Example 2.7.3 is a very good conceptual case study for the same.

Here I is the moment of inertia of the pile cross section about the axis of rotation
and f7,1 and f7,2 are factors for rotational direction for fixed head piles, as furnished
in Table 2.5.5.
2.5.9 Group effect for rotational motion

For a pile group the group stiffness, shown in Figure 2.5.7, is expressed as
g

N
kψ = [kψ1 + kz1 Xr2 + kx1 Zc2 − 2Zc kxψ1 ] + kψf (2.5.12)
1
Here Xr and Zc are shown in the Figure 2.5.7 and kz1 and kx1 are stiffness constant
of single piles as described earlier. In addition

Ep I p
kxψ1 = f9,1 and
r20
(2.5.13)
2
δ2 Zc Zc
kψf = Gs r0 hS̄ψ1 + Gs r20 h + −δ S̄u1
3 r0 r0
where δ = h/r0 , here h = embedment depth of pile cap, and Sψ1 is as given in
Table 2.5.7.
The damping matrix for the pile group is expressed by
g

N
cψ = [cψ1 + cz1 Xr2 + cx1 Zc2 − 2Zc cxψ1 ] + cψf (2.5.14)
1
where cz1 and cx1 are damping constant of single piles as described earlier. In addition

Ep Ip
cxψ1 = f9,2 and
r0 v s
Table 2.5.7 Values of Sψ1 and Sψ2 for various Poisson’s ratio.
Poisson’s ratio S̄ψ1 S̄ψ2
0.0 2.5 1.8

0.25 2.5 1.8
0.40 2.5 1.8

Machine block
Soil line
h
Zc
Xr
Figure 2.5.7 Geometric description of the pile group.
2
f δ2 Zc Zc
cψ = δr40 Gs γs /g Sψ2 + + −δ S̄u2 (2.5.15)
3 r0 r0
We now explain the above theory further by a suitable numerical problem.
Example 2.5.1
Find the vertical, horizontal and rocking stiffness of the pile group based on
Novak’s formulation as shown in Fig. 2.5.8 and with the following soil
properties:
Length of the pile = 45.0 m; Diameter of the pile = 950 mm; Grade of concrete
M20 having a dynamic modulus as 300 × 106 kN/m2 . Consider Poisson’s ratio
of soil = 0.4.
Solution:
Since each of the layers has different velocity and thickness we take a weighted
average of the shear wave velocity of the three soil layers as follows
60 × 10 + 110 × 20 + 215 × 15
vs = = 134 m/sec
45

600
1400
10 m vs = 60 m/sec Layer#1
= 18 kN/m3
vs = 110 m/sec Layer#2

20 m = 20 kN/m3
vs = 215 m/sec Layer#3

30 m = 22 kN/m3
2000 2000
Geotechnical profile of soil in which the piles are located
3000 3000
500
(Typ)
3 6 9
2000
2 5 8
2000
1 4 7
Arrangement of the pile group in Plan
Figure 2.5.8 Machine foundation supported on piles.
18 × 10 + 20 × 20 + 22 × 15
Average weight density of soil is = γ s = =
45
20.22 kN/m3
Thus dynamic shear modulus of the soil is taken as
20.22
Gavg = × (134)2 = 37010 kN/m2 ; g, acceleration due to gravity is
9.81
taken as 9.81 m/sec2 .
L×B 7×5
Equivalent Radius of pile cap = = = 3.33 m
π π

Vertical stiffness of piles

π × (0.95)2
Cross-sectional area of each pile = = 0.7088 m2
4
Dynamic modulus of concrete = 300 × 106 kN/m2
Density of concrete (γc ) = 25/kN/m3

Ep × g 300 × 106 × 9.81
vc = = = 10849 m/sec;
γp 25
vs 134
= = 1.235 × 10−2 ≈ 0.01
vc 10849
L 45
= = 94.7 > 25.
r0 0.475
For L/r0 = 95 and vs /vc = 0.01 we have from Novak’s Chart
→ f18,1 = 0.011 and f18,2 = 0.005.
Thus for each pile we have,
Ep A 300 × 106 × 0.7088

→ Kzi = f18,1 = × 0.011
r0 0.475
= 4.924 × 106 kN/m
Ep A 300 × 106 × 0.7088

and Czi = f18,2 → × 0.005
vs 134
= 7934 kN-sec/m.
Calculation of group interaction factor for, λ(L/r0 ) = 95

Thus, for group effect

9
9 × 4.924 × 106
Kzg = = 11480829 kN/m;
3.86
i=1

9
9 × 7934
Czg = = 18498 kN-sec/m.
3.86
i=1

Calculation of group interaction factor (vertical)
Pile No. Spacing (S) L/S α Remarks
1 0.0 0.0 1.0 Pile # 1 is the reference

pile
2 2.0 22.5 0.474
3 4.0 11.25 0.368 α obtained vide
4 3.0 15.0 0.413 Eqn (7.5.6a)
5 3.6 12.5 0.384
6 5.0 9.0 0.334
7 6.0 7.5 0.307
8 6.32 7.12 0.299
9 7.21 6.24 0.278
Total sum 3.86
Effect of the pile cap on overall pile stiffness

Here the effect of layer#1 will be more dominant on the pile cap as such for
finding out the stiffness properties in context of the pile cap, we have
18
Gs = × (60)2 = 6605 kN/m2
9.81
For embedded depth, h = 1.4 m and S1 = 2.7, we have,
f f
Kz = Gs h̄S1 → Kz = 6605 × 1.4 × 2.7 = 24969 kN/m.

f 6605 × 18
Cz = h̄r0 Gs γs /gS2 = 1.4 × 3.33 × 6.7 = 3438.6 kN/m.
9.81
Thus, total vertical stiffness = 11480829 + 24969 = 11505798 kN/m

Total damping for the pile and pile cap = 18498 + 3439 = 21937 kN-sec/m.
Calculation of horizontal stiffness

π × (0.95)4
I= = 0.04 m4 ; with l/r0 = 95, νs = 0.4 and vs /vc = 0.01, from
64
Novak’s chart f11,1 = 0.0036 and f11,2 = 0.0084
Ep I 300 × 106 × 0.04

Kx = × f11,1 = × 0.0036 = 403091 kN/m;
r30 (0.475)3
Ep I 300 × 106 × 0.04

Cx = × f11,2 = × 0.0084 = 3334 kN-sec/m.
r20 vs (0.475)2 × 134

Calculation of interaction factor (lateral)
Pile number Spacing S/r0 β factor in degree αl Remarks
1 0.0 0.0 0 0 Pile#1 is the reference pile

2 2.0 4.21 90 0.246
3 4.0 8.42 90 0.123 αl calculated
4 3.0 6.316 0 0.328 as per eqn (7.5.9a)
5 3.6 7.578 33.6 0.231
6 5.0 10.526 53.13 0.134
7 6.0 12.632 0 0.164
8 6.32 13.306 18.43 0.148
9 7.21 15.178 33.6 0.115
Total sum 1.489 ∼
=1.50
For the pile group we have,

9
403091 × 9
Kx = = 2418546 kN/m;
1.5
i=1

9
3334 × 9
Cx = = 20004 kN-sec/m.
1.50
i=1
Effect of the pile cap on overall pile stiffness

Here the effect of layer#1 will be more dominant on the pile cap as such for
finding out the stiffness properties in context of the pile cap we have
18
Gs = × (60)2 = 6605 kN/m2
9.81
For embedded depth, h = 1.4 m and Su1 = 4.1 and Su2 = 10.6,
f
Kx = Gs h̄Su1 = 6605 × 1.4 × 4.1 = 37913 kN/m;

f 6605 × 18
Cx = h̄r0 Gs γs /gSu2 = 1.4 × 3.33 × 10.6 = 5440 kN/m.
9.81
Thus, the total lateral stiffness = 2418546 + 37913 = 2.456 × 106 kN/m.
Total damping for the pile and pile cap in lateral direction = 20004 + 5440 =
25444 kN-sec/m.

Calculation of rocking stiffness and damping

The rocking stiffness of individual pile, with f7,1 = 0.202, is given by

Ep Ip 300 × 106 × 0.04
kψ1 = f7,1 = × 0.202 = 5103158 kN/m
r0 0.475
The coupled sliding and rocking stiffness with, f9,1 = −0.0194, is given by

Ep I p 300 × 106 × 0.04
kxψ1 = f9,1 = × (−0.0194)
r20 (0.475)2
= −1031801 kN/m
The pile cap stiffness is given by
2
δ2 Zc Zc
kψf = Gs r0 hS̄ψ1 + Gs r20 h + −δ S̄u1
3 r0 r0
Substituting the values, Gs = 6605 kN/m2 ; h = 1.4 m; r0 = 3.33 m; S̄ψ1 =

2.5; δ = rh0 = 0.42; Zc = 1.5, Xr = 2.0 m and S̄u1 = 4.1, we have, kψf =
84417 kN/m.
Thus the total stiffness of the pile group is given by
g
N
kψ = [kψ1 + kz1 Xr2 + kx1 Zc2 − 2Zc kxψ1 ] + kψf
1
g
kψ = 9 × [5103158 + 4.924 × 106 × 4 + 403091 × 2.25
+ 2 × 1.5 × 103801] + 84417
g
kψ = 2.342 × 108 kN/m
Calculation of damping value in rocking mode

The damping value with, f7,2 = 0.139, is given by

Ep Ip 300 × 106 × 0.04
cψ1 = f7,2 = × 0.139 = 12448 kN-sec/m.
vs 134

The coupled sliding and rocking mode damping, with f9,2 = −0.0280, is
given by

Ep Ip 300 × 106 × 0.04
cxψ1 = f9,2 = × (−0.0280) = −5273 kN-sec/m
r0 v s 0.475 × 134
The damping contributed by pile cap is given by

f δ 2 Z c
2
Z c
cψ = δr40 Gs γs /g Sψ2 + + −δ S̄u2
3 r0 r0
where, Gs = 6605 kN/m2 ; h = 1.4 m; r0 = 3.33 m; S̄ψ2 = 1.8; δ = h/r0 = 0.42;

Zc = 1.5 m; S̄u2 = 10.6; γs = 18 kN/m3 ; g = 9.81 m/sec2 ,
Substituting the above values we have, cψf = 14604 kN-sec/m. The damping
value of the pile group is given by
g

N
cψ = [cψ1 + cz1 Xr2 + cx1 Zc2 − 2Zc cxψ1 ] + cψf
1
g
cψ = 9 × [12448 + 7934 × 4 + 3334 × 2.25 + 2 × 1.5 × 5273] + 14604
= 622145 kN-sec/m.
2.5.10 Model for dynamic response of pile

In previous section we had presented the dynamic stiffness of piles as proposed by
Novak. This is a very popular model for dynamic response of machine foundations in
the design offices. Based on the analytical solution of Baranov (1967), Novak (1974)
proposed the method for evaluating the vertical response of piles under dynamic load-
ing. Many researchers like, Wolf and Von Arx (1978), Waas (1981), Kaynia and
Kausel (1982), Banerjee and Sen (1987) have advanced solutions to this problem,
yet Novak’s method remains the most popular due to its sheer simplicity in appli-
cation. Solution of pile and pile-group based on the method proposed by Banerjee
and Sen, which is based on Boundary Element Method gives quite accurate results
but it is computationally too exhaustive to find applications in a day-to-day design
office work.
Applying Finite Element Method, where the pile is modeled as beam elements
and the soil as Winkler springs, has yielded good results. But, they are found to
be valid only when piles are single or when the distance between piles is signifi-
cant (≥5d, d being the diameter of the pile), when the pile-soil interaction can be
neglected.
Novak’s solution is mostly based on charts, and it furnishes stiffness and damping
of a pile and the solution is addressed to the fundamental degree of freedom.

In spite of its immense popularity, the model do have a few limitations as

summarized below:
a The solution does not take into account inertial effect of the pile;
b Extrapolation is required when design data are out of range of the chart;
c Charts are available only for RCC or timber piles, whether these charts are
applicable to cases of steel piles13 , there is no clear-cut guideline;
d The charts do not address to the case where a pile is partially embedded;
e The formulation do not cater to dynamic axial load, moments or shears induced in
pile due to dynamic loads.
In certain cases when piles are supporting reciprocating compressors, it becomes

essential to check the design for higher frequencies of the foundations to ensure that
they are not matching with the second or third harmonics, when higher forces may be
induced by the machine at harmonic other than the first. In such cases one has no other
options but to resort to an elaborate and expensive three-dimensional Finite Element
based soil-pile foundation model to arrive at an answer to this problem and in number
of cases uncertainties present in such results are many. However, it was shown by
Novak that results obtained for higher harmonics are not significantly different for
the type of problem that is normally encountered in pile vibration studies.
We present now a model (Chowdhury and Dasgupta 2006) that overcomes many of
the limitations cited above. The solution is simple (yet realistic) and does not require
elaborate software to be developed for the analysis. A simple spreadsheet would suffice
for the problem considering the solution is basically analytic.
2.5.10.1 Proposed method

We had stated at the outset that most of the work relating to dynamic stiffness of pile is
based on Baranov’s (1967) theory on the response of a soil embedded foundation. The
present formulation is based on Novak and Beredugo’s (1972) approach on embedded
foundation.
2.5.10.2 Vibration of friction piles

Let us consider a pile as shown in Figure 2.5.9. The pile is assumed to provide resistance
both through bearing as well as friction.
Let Kf represents the frictional stiffness of the pile and the pile tip bearing stiffness
is taken as Kb . The longitudinal vibration of such beams having only the frictional
stiffness may be represented by the expression
∂ 2u ∂ 2u
EA + K f u = m(z) (2.5.16)
∂z2 ∂t 2
13 This is an important issue for many real life projects specially in Arctic condition (like North Siberia)
or very arid region (like Sudan, Algeria) due to extreme low temperature or absence of water makes
concreting hazardous and almost all the structures and foundations are built on steel piles.

Kf
dz
Kv
Kb
Figure 2.5.9 Pile embedded in ground up to a depth L and its mathematical model.
in which, E = Young’s modulus of pile; A = area of pile; Kf = dynamic frictional

stiffness of soil having dimension (F/L) and u(z, t) = dynamic amplitude of pile =
φ(z) q(t); and m(z) = mass of element dz.
One of the solutions of equation (2.5.16) is given by
q(t) = C3 sin ωt + C4 cos ωt (2.5.17)
With the definition of u and using Equation (2.5.17), Equation (2.5.16) may be
written as
d 2 φ(z)
EA + Kf φ(z) = −m(z)ω2 φ(z) (2.5.18)
dz2
The above equation can further be simplified to
d 2 φ(z)
+ p2 φ(z) = 0 (2.5.19)
dz2
where p2 = (mω2 + Kf ).
If you observe Equation (2.5.19) carefully, you should realize that it suggests that
the presence of frictional stiffness Kf does not affect the basic shape function of the
pile and would remain same for the case had the pile would not have been embed-
ded. However, the bearing stiffness Kb connected at the end of pile would affect the

shape function depending on the appropriate boundary condition. For computing the
correct shape function of the system, one has to start with the model as shown in the
Figure 2.5.9. The general solution for Equation (2.5.19) is given by Humar (1990)
φ(z) = (C1 cos pz + C2 sin pz)(C3 sin ωt + C4 cos ωt) (2.5.20)
in which, C1 , C2 , C3 and C4 are the integration constants to be determined from the

appropriate boundary conditions.
The pile has at the free head, z = 0, EA du
dz
= 0, which gives
EAp[−C1 sin pz + C2 cos pz][C3 sin ωt + C4 cos ωt] = 0 ➔ C2 = 0. (2.5.21)
and at the tip, z = L, EA du

dz
= −Kb u(z)z=L , which gives
EAp[−C1 sin pL] = −Kb C1 cos pL ➔ pL tan pL = Kb L/(EA) (2.5.22)
in which Kb = Gb r0 Cb (2.5.23)
where, Gb = dynamic shear modulus of the soil at pile tip; r0 = radius of the pile;
Cb = a frequency independent dimensionless constant as suggested by Novak and
Beredugo (1972) and is given in Table 2.5.8.
Combining Eqns. (7.5.22) and (7.5.23), one can have
Gb C b L
pL tan pL = (2.5.24)
Eπr0
It will be observed that the right hand side of Equation (2.5.24) is a dimensionless
quantity.
If η = GEπ
b Cb L
r0 = Gb
E
Cb
π λ; where λ = slenderness ratio (L/r0 ) of the pile, Equation
(2.5.24) can be represented as
pL tan pL − η = 0. (2.5.25)
Equation (2.5.25) is a transcendental equation in pL and can be solved numerically.

The values of pL for various values of η for the first mode are shown in Table 2.5.9.
Writing, pL = β, the arbitrary shape function of the problem is given by
z
φ(z) = cos β (2.5.26)
L
The potential energy d of an element of depth dz, shown in Figure 2.5.9, is given
by Shames and Dym (1995)

EA du 2 Kv 2
d = + u (2.5.27)
2 dz 2

where, E = Young’s modulus of pile; A = area of pile; Kv = dynamic stiffness of soil

having dimension kN/m; w = displacement of pile in the z direction and may be written
as φ(z)q(t).
In Equation (2.5.27), Kv consists of two parts, namely,
1 the bearing stiffness at pile tip, and
2 the friction stiffness along the shaft.
For a rigid circular embedded footing with embedment Df , the stiffness of the footing
may be expressed as per Novak & Beredugo as
Kv = Gb r0 Cb + GDf S1 (2.5.28)
where, Kv = foundation stiffness in the vertical direction; G = dynamic shear modulus

of the soil along the embedment length; Gb = dynamic shear modulus of the soil at
the base; r0 = radius of the foundation; Cb and S1 = dimensionless constant which
are basically frequency dependent.
However, it has been shown by Novak and Beredugo that considering Cb and S1
as frequency independent, no accuracy is lost for practical design problems and the
analysis becomes quite simplified for rigid circular embedded footing.
The frequency independent values of Cb and S1 are as given below in Table 2.5.8.
However, it should be remembered that an embedded circular footing is usually
considered to be rigid having infinite structural stiffness. On the contrary, a pile will
be far more flexible member whose structural stiffness will be much lower, thus the
above recommended value may be valid for certain pile geometry but may not be
valid for others. Comparing the stiffness data of piles obtained by Novak, Dobry and
Gazetas (1988) it is proposed that following value of S1 be used for dynamic analysis
of piles in vertical direction.
9.553(1 + ν)
S1 = (2.5.29)
λ0.333
where ν = Poisson’s ratio of the soil; and λ = slenderness ratio of the pile.
This value of S1 is derived based on similar technique used earlier by Lysmer and
Richart (1966) for deriving equivalent stiffness and damping of circular footings for
Lysmer’s analog from the solutions of a similar elasto-dynamic analysis as proposed
by Bycroft (1956). The value Cb may be taken as suggested in Table 2.5.8 for it has no
bearing on the flexibility of pile and is a function of the base area only. Considering pile
base area is much smaller in comparison to a footing, its contribution is only marginal.
Table 2.5.8 Suggested frequency independent values suggested

by Novak and Beredugo (1972) for embedded
footing.
Poisson’s ratio Cb S1
0.0 3.9 2.7

0.25 5.2 2.7
0.5 7.5 2.7

Table 2.5.9 Roots of equation pL tan(pL)-η = 0 for the first or fundamental mode.
η pL η pL η pL η pL
0 0.02 0.7 0.75 2 1.077 15 1.473

0.1 0.322 0.8 0.791 2.5 1.142 20 1.496
0.2 0.433 0.9 0.828 3 1.192 25 1.51
0.3 0.522 1 0.86 3.5 1.232 30 1.52
0.4 0.593 1.25 0.931 4 1.265 35 1.527
0.5 0.653 1.5 0.988 5 1.314 40 1.533
0.6 0.705 1.75 1.036 10 1.429 50 1.54
Moreover in most of the practical cases its effect does not come into consideration (as
will be shown subsequently) for analysis of such piles are either considered as bearing
pile i.e. having infinite base stiffness or floating having no base effects. The first term in
Equation (2.5.28) represents the contribution of base resistance, while the second term,
the embedment effect of the foundation. Substituting Equation (2.5.28) in Equation
(2.5.27) for an element dz, d may be written as

EA du 2 Gb r0 Cb 2 GS1 dz 2
d = + u + u (2.5.30)
2 dz 2 2
and the total potential energy over the total length of the pile (L) is given by
L 2 L
EA du GS1 G b r0 C b 2
= dz + u2 dz + u (2.5.31)
2 dz 2 2
0 0
Considering u(z, t) = φ(z)q(t), it can be proved (Hurty and Rubenstein 1967), that
L L
Kij = EA φi (z)φj (z)dz + GS1 φi (z)φj (z)dz + Gb r0 Cb φi (L)φj (L) (2.5.32)
0 0
where the shape function of the problem is given by Equation (2.5.26).

The first derivative of the above with respect to z is given by
β z
φ (z) = − sin β (2.5.33)
L L
Using z/L = ξ implying dz = Ldξ , and converting the shape function as furnished
in Equation (2.5.26) from local to generalized co-ordinates, the limits of the problem
get converted to 1 to zero.

Now, if one assumes
F(ξ ) = cos βξ (2.5.34)

β
Fi (ξ ) = − (sin βξ ), and (2.5.35)
L
+1 1
EAβi βj
Kij = Fi (ξ )Fj (ξ )dξ + GS1 L Fi (ξ )Fj (ξ )dξ + Gb r0 Cb Fi (1)Fj (1).
L
0 0
(2.5.36)
For the fundamental mode i = j = 1 and Equation (2.5.36) reduces to
+1 1
EAβ 2
K1 = F1 (ξ ) dξ + GS1 L F1 (ξ )2 dξ + Gb r0 Cb F1 (1)2
2
(2.5.37)
L
0 0
Equation (2.5.37) can be rewritten as
+1 1
EAβ 2
K1 = (sin βξ ) dξ + GS1 L (cos βξ )2 dξ + Gb r0 Cb (cos β)2
2
(2.5.38)
L
0 0
Equation (2.5.38) on integration and after some simplification may be expressed as

K1 = I1 + I2 + I3 in which,

EAβ 2 1 sin 2β 1 sin 2β
I1 = − ; I2 = GS1 L + ;
L 2 4β 2 4β
Gb r 0 C b
I3 = (1 + cos 2β) (2.5.39)
2
Finally, K1 can be written as

EAβ 2 GS1 L Gb r0 Cb GS1 L EAβ Gb r 0 Cb
K1 = + + + − sin 2β + cos 2β
2L 2 2 4β 4L 2
(2.5.40)
which may be further simplified to
K1 = X1 + X2 sin 2β + X3 cos 2β (2.5.41)
in which

EAβ 2 GS1 L Gb r0 Cb GS1 L EAβ Gb r0 Cb
X1 = + + ; X2 = − ; X3 =
2L 2 2 4β 4L 2
(2.5.42)

Equation (2.5.42) gives the stiffness of the pile for the vertical mode, without any
limitation to slenderness ratio, E/G or the material type.
2.5.10.2.1 Mass of the pile

For a conservative system, if T is the kinetic energy of the system then at any time t,
the energy equations may be written as
H 2
1 ∂u(z, t)
T(t) = m(z) dz (2.5.43)
2 ∂t
0

n
Using, u(z, t) = φi (z)qi (t) (2.5.44)
i=1
where u(z, t) = displacement function; φi (z) = shape function; qi (t) = generalized

co-ordinate; m(z) = mass of element dz and substituting Equation (2.5.44) in Equation
(2.5.43), the energy equation may be written as
⎡ ⎤⎡ ⎤
H
n
n
1
T (t) = m(z)⎣ φi (z)q̇i (t)⎦ ⎣ φj (z)q̇j (t)⎦dz
2
0 j=1 j=1
(2.5.45)
⎡H ⎤

n
n
1
= q̇i (t)q̇j (t)⎣ m(z)φi (z)φj (z)dz⎦
2
i=1 j=1 0
from which the mass matrix may be written as
⎡H ⎤

mij = ⎣ m(z)φi (z)φj (z)dz⎦ for i, j = 1, 2, 3 . . . n (2.5.46)
0
Similarly the stiffness value with transformation from local to natural co-ordinate,
the mass contribution of the pile may be obtained as
1
γp AL
mij = Fi (ξ )Fj (ξ )dξ (2.5.47)
g
0
were γp = bulk density of pile material; A = area of pile cross section; L = pile length
embedded in soil, and g = acceleration due to gravity.

For the fundamental mode, i, j = 1, and one can have
1
γp AL
m1 = F1 (ξ )2 dξ (2.5.48)
g
0
The above on expansion results in
1
γp AL
m1 = (cos βξ )2 dξ (2.5.49)
g
0
Integration of Equation (2.5.49) gives

γp AL sin 2β
m1 = 1+ (2.5.50)
2g 2β
which is the contributory mass of the pile for the fundamental mode in the vertical
direction.
2.5.10.2.2 Damping of the pile

The damping of the pile embedded in soil will constitute of two parts:
• Material damping of the pile itself;

• Radiation damping of the soil-pile system.
It is obvious that the material damping of the pile will be much lower than that of
the soil radiation damping. As the first step for calculating the soil damping one may
ignore the material damping of the pile for the time being. Material damping of soil
also is part of the system vibration. However, it has been found that for translational
vibration their effect is insignificant and may be neglected without any significant effect.
Else, if one wishes, their values may be obtained from resonant column test from the
laboratory when damping may be obtained from ratio of successive amplitudes.
For a rigid footing embedded in soil for a depth Df , Novak and Beredugo has
proposed an expression

Cz = r0 ρb Gb C̄b + r0 ρG S̄2 Df (2.5.51)
where, r0 = radius of the foundation; Gb = dynamic shear modulus at foundation

base; G = dynamic shear modulus of soil in which the foundation is embedded; Df =
depth of embedment; C̄b and S̄2 = frequency independent constants as defined by
Novak and furnished in Table 2.5.10.

Table 2.5.10 Values of damping coefficients based on Novak and Beredugo (1972).
Poisson’s ratio C̄b S̄2

0.7022a0
0.0 3.438a0 + 0.5742a02 − 1.154a03 + 0.7433a04 6.059a0 +
a0 + 0.01616
0.25 5.06a0 Do
0.5 7.414a0 − 2.986a02 + 4.324a03 − 1.782a04 Do
where a0 = ωr/vs in which, ω operating frequency of the system in rad/sec; r = radius of the pile; vs = shear wave
velocity of the soil.
With reference to Figure 2.5.1 for a pile element of length dz, embedded in the soil,
the above equation may expressed as

Cz = r0 ρb Gb C̄b + r0 ρG S̄2 dz (2.5.52)
For systems having continuous function, the damping is usually expressed as

(Mario 1987):

Cz = c(z) φi (z)φj (z)dz (2.5.53)
For the present case, Equation (2.5.53) can be expressed as
L
Cz = r0 ρG S̄2 φi (z)φj (z)dz + r0 ρb Gb C̄b φi (L)φj (L) (2.5.54)
0
Considering φ(z) = cos β Lz , for the fundamental mode, one can have
L
z
Cz = r0 ρGS̄2 cos2 β dz + r0 ρb Gb C̄b cos2 β (2.5.55)
L
0
and hence
1
Cz = r0 ρGS̄2 L cos2 βξ dξ + r0 ρb Gb C̄b cos2 β (2.5.56)
0
Equation (2.5.56), on integration simplifies to

√
1 1 r0 ρGS̄2 L r0 ρb Gb C̄b
Cz = r0 ρG S̄2 L + r0 ρb Gb C̄b + sin 2β + cos 2β
2 2 4β 2
(2.5.57)
Equation (2.5.57) expresses the soil damping for a single pile under vertical mode
of vibration. Here the Factor S̄2 and C̄b are damping coefficients which are frequency

dependent. Fortunately the damping factor is required for calculation of the amplitude
when the eigen solution of the problem is already done vis-a-vis, the dimensionless
frequency number a0 = ωr/vs term is known14 . Polynomial fit curve for S̄2 and C̄b are
available in terms of a0 which can be used to arrive at these parameters.
The damping constants are given in Table 2.5.10.
2.5.10.2.3 Consideration of material damping of pile

The structural stiffness contribution of the pile is given by Equation (2.5.40), while
that of the mass is given in Equation (2.5.50).
Thus, if Cc be the critical damping of the
pile then it can be expressed as Cc = 2 Kmp , where K (equals I1 in Equation 2.5.39)
and mp are the stiffness and mass matrices of the pile. Depending on the material
used for pile like (RCC, steel etc.) a suitable damping ratio (D) can be assumed. The
damping (C) for the pile can expressed as
Cp = DCc (2.5.58)
This, when added to the radiation damping, calculated earlier, gives the complete
damping quantity for the soil-pile system. It should be noted that for perfectly floating
piles structural contribution of pile vanishes, and the material damping of the pile
mentioned in the preceding need not be considered.
2.5.10.3 Vibration of bearing piles

The expressions derived so far give a general case when the load is transferred from
the pile to the soil, through both friction and bearing. There will be cases when the
pile is pre-dominantly bearing in load transfer. Using the above formulation when
lim η → ∞ (i.e. Gb is very large compared to E), pL(tan pL) = ∞, when β → π/2,
the pile reduces to a perfectly bearing pile (i.e. fixed at the base), however for practical
case when η → 50, it will not be too erroneous to assume β → π/2, when the stiffness
of the pile reduces to

EAπ 2 GS1 L
K1 = + and (2.5.59)
8L 2
the damping may be expressed as
1
C1 = r0 ρGS̄2 L (2.5.60)
2
and the mass is
γp AL
m1 = . (2.5.61)
2g
14 For calculation of damping it is considered ω = ωn for it is most critical at resonance.

2.5.10.4 Vibration of friction piles

When Gb is very small the load is transferred mainly through pile friction. In the
above formulation when lim η → 0, pL tan pL = 0, when, β → 0, the pile becomes a
perfectly friction pile.
Thus, for β → 0, the stiffness of pile is given by
K1 = (GS1 L)/2 (2.5.62)
The damping matrix may be expressed as
1
C1 = r0 ρG S2 L + r0 ρb Gb C̄b (2.5.63)
2
From Equation (2.5.63) it should be noted that for a friction pile, the damping factor
increases, while the stiffness term in Equation (2.5.62) is less than the bearing case in
Equation (2.5.59). A similar observation has also been made by Novak (1974) in his
investigation. For very poor soil, the term Gb in Equation (2.5.63) may be ignored.
However for cases when piles located in medium to stiff homogenous clayey soil where
G = Gb and yet the load is basically transferred through friction, the last term cannot
be ignored and would further enhance the radiation damping. The mass matrix shall
be same as stated in Equation (2.5.50).
2.5.10.5 Vertical vibration of partially embedded piles

In many instances, especially in the arctic condition, due to environmental reasons,
the steel piles are driven into the ground when they protrude about 2 to 3 m above the
ground over which the pile cap and vibrating equipments as placed (Figure 2.5.10). In
such cases Novak’s (1976, 1983) chart cannot be used readily.
Rotating Machine
Pile Cap
Partially embedded piles
G.L.
L
L1
Figure 2.5.10 Schematic diagram of partially embedded piles.

To evaluate the pile stiffness for such cases, the stiffness Equation (2.5.36) is to be
modified as
L L1
Kij = EA φi (z)φj (z)dz + GS1 φi (z)φj (z)dz + Gb r0 Cb φi (L)φj (L) (2.5.64)
0 0
where L1 = partial depth of embedment of pile and L = total length of pile.

It is apparent from Equation (2.5.64) that the first and last term remains unchanged
and the second term based on depth of embedment gets modified, where the integra-
tion limits changes to (L1 − 0) and the stiffness expression for the fundamental mode
reduces to

EAβ 2 GS1 L1 G b r0 C b GS1 L1 EAβ G r0 C b
K1 = + + + − sin 2β + b cos 2β
2L 2 2 4β 4L 2
(2.5.65)
The damping of the pile-soil system is given by

√
1 1 r0 ρGS2 L1 r0 ρb Gb C̄b
Cz = r0 ρGS̄2 L1 + r0 ρb Gb C̄b + sin 2β + cos 2β
2 2 4β 2
(2.5.66)
The mass matrix remains the same as stated in Equation (2.5.50).

It should be noted that for this case while calculating the value of S1 [Equation
(2.5.29)], the slenderness ratio is to be calculated based on the embedded length
of the pile.
2.5.10.5.1 Stiffness of the pile for soils with varying elastic property
In the previous section, the calculation of stiffness as well as the damping of soil was
based on the dynamic shear modulus of soil invariant with depth. While this could be
possible for clayey soils, there are many cases when the dynamic shear modulus of the
soil has been found to vary with depth. Generically this can be expressed as
G = G (z/H)α (2.5.67)
where α = a number varying from 0–2 [considered 0 when G is constant with depth,
assumed 1 for linear variation and 2 for parabolic distribution].
For instance for the soil with variable elastic property, Equation (2.5.67) may be
modified to
G = Gξ α (2.5.68)
where ξ = z/H.

For the cases mentioned above, Novak’s (1976) chart is possibly not valid. To
accommodate the above variation, the stiffness equation can be modified to
+1 1
EAβ 2
Kij = Fi (ξ )Fj (ξ )dξ + GS1 L ξ α Fi (ξ )Fj (ξ )dξ + Gb r0 Cb Fi (L)Fj (L)
L2
0 0
(2.5.69)
2.5.10.5.2 Shear modulus having a linear variation

When the soil has linear distribution with depth, the stiffness Equation (2.5.69) may
be expressed as
+1 1
EAβ 2
K1 = (sin βξ ) dξ + GS1 L ξ (cos βξ )2 dξ + Gb r0 Cb (cos β)2 (2.5.70)
2
L
0 0
which, on integration and subsequent simplification, gives rise to

1 EAβ 2 GS1 L 1 G b r0 C b 1 GS1 L EAβ
K1 = + 1− 2 + + − sin 2β
2 L 4 β 2 2 β L

GS1 L Gb r0 Cb
+ + cos 2β (2.5.71)
4β 2 2
It may be noted that while for bearing pile β = π /2, for friction pile (unlike constant
G case), β = 0 is an inadmissable function in this case. For the fundamental mode the
admissible function is β = π , which is the next higher mode. This is logical also for
the soil having stiffness increasing with depth and the pile will have a natural tendency
to wobble about its centre rather than moving en-mass.
The damping matrix in this case can be expressed as
1
Cz = r0 ρGS̄2 L ξ cos2 βξ dξ + r0 ρb Gb C̄b cos2 β (2.5.72)
0
The integration of the first term in Equation (2.5.72) being cyclic in nature and can
be solved approximately by expanding the cosine function in series. On integration,
Equation (2.5.72) reduces to

2 1 β2 2 4
Cz = r0 ρGS̄2 L − 2β 2 − + β + r0 ρb Gb C̄b cos2 β (2.5.73)
3 7 33 675

2.5.10.5.3 Shear modulus having a parabolic variation

When the soil modulus has a parabolic distribution with depth, the stiffness equation
may be expressed as
+1 1
EAβ 2
K1 = (sin βξ ) dξ + GS1 L ξ 2 (cos βξ )2 dξ + Gb r0 Cb (cos β)2
2
L
0 0
(2.5.74)
which on integration and subsequent simplification reduces to

EAβ 2 GS1 L Gb r0 Cb GS1 L 1 1 EAβ
K1 = + + + + − sin 2β
2L 6 2 2 2β 3β 3 4L

GS1 L Gb r0 Cb
+ + cos 2β (2.5.75)
β 2
In this case, the first admissible function will be β = π for a friction pile and β = π/2
for a bearing pile.
The mass matrix for both the cases remains same as stated in Equation (2.5.50)
while the damping matrix can be obtained from the expression
1
Cz = r0 ρGS̄2 L ξ cos2 βξ dξ + r0 ρb Gb C̄b cos2 β (2.5.76)
0
which on integration and simplification reduces to

√ √
r0 ρb Gb Cb r0 ρGS2 L r0 ρGS2 L
Cz = − + sin 2β
2 4 8β
√
r0 ρGS2 L r0 ρb Gb Cb
+ + cos 2β. (2.5.77)
4β 2
2.5.10.6 Group effect of pile

This has already been explained in detail in section 2.5.5 and may be used for the
present case also.
2.5.10.7 Effect of pile cap on pile stiffness

The pile cap has been found to affect the response of footing significantly. Before
considering its effect within the proposed framework, it would be worthwhile to
recapitulate the practice in vogue.

Df
Figure 2.5.11 Schematic diagram of pile and pile-cap with embedment.
The sketch given in Figure 2.5.11 can represent the pile group with pile cap.
In such case usually the embedment stiffness GSf Df is directly added to the pile
group stiffness and the system is considered as a lumped mass single degree freedom
system where

Kgroup + Gf Sf Df
ω= (2.5.78)
M
where Gf = dynamic shear modulus of the soil surrounding the pile cap; Df = depth
of embedment; Sf = constant as suggested by Novak furnished in Table 2.5.4 (as S1 );
M = mass of pile cap and machine placed on it.
It may be noted that contributing effect of the pile mass is ignored in the above which
could be significant for a pile group having large number of piles. To overcome the
above limitation and also to derive a better response we propose a two mass lumped
model has been proposed and shown in Figure 2.5.12.
The mass and stiffness matrices for the above model may be written as

Kgroup + Gf Sf Df −Gf Sf Df
[K] = (2.5.79)
−Gf Sf Df G f Sf D f

M = Mass of pile cap + machine
K2 = GfSfDf
mp = Mass of piles in group
K1 = Kgroup
Figure 2.5.12 Proposed two mass lumped model for the pile and pile cap.
⎡ ⎤
nγ p AL sin 2β
1+ 0
and [M] = ⎣ 2g 2β ⎦ (2.5.80)
0 M
where n = number of piles in the pile group.

Since Equation (2.5.79) is statically coupled, the damping matrix is given by

Cgroup + Cf −Cf
C= (2.5.81)
−Cf Cf

where Cz = r0 ρGS̄2 Df and Df is the embedment depth of pile cap. (2.5.82)
Once the stiffness, mass and damping matrices are established, the natural frequency
of the system may be obtained from the standard expression
[K] − [M] ω2 = 0 (2.5.83)
leading to

(mp B + MA) ± [(mp B + MA)2 − 4mp MAB]
λ1,2 = (2.5.84)
2mp M
!
nγp AL sin 2β
in which mp = 2g 1+ 2β and A = Kgroup + Gf Sf Df ; B = Gf Sf Df .

√ √
where, ω1 = λ1 and ω2 = λ2 here ω1 and ω2 are the natural frequency of the
structure.
The damping matrix generated here is non-classical in nature and will not be
de-coupled on orthogonal transformation. However, since the degrees of freedom
considered here is two, the same can also be converted into an equivalent Rayleigh
damping (refer section 2.2.5 in this chapter where we have solved this) when the matrix
will decouple and standard modal solution can be applied.
2.5.10.8 Solutions for higher modes

This case is usually not considered in design office practices and neither any guidelines
presently exists for the same except treating the pile as a beam and the soil as Winkler
springs and solving the same based on finite element method. Using the proposed
methodology, the stiffness, damping and mass matrices can be computed for the higher
modes.
Referring to Equation (2.5.37), the stiffness matrix can be stated as
EA
[Kij] =
L
⎡ ⎤
#1 #1 #1 #1
⎢ β1
2
F 1 (ξ )2
β 1 β2 F 1 (ξ )F2 (ξ ) β1 β3 F 1 (ξ )F 3 (ξ ) ......... β1 βn F 1 (ξ )Fn (ξ ) ⎥
⎢ 0 0 0 0 ⎥
⎢ # # # ⎥
⎢ 1 1 1 ⎥
⎢β2 β1 F2 (ξ )F1 (ξ ) β2 F2 (ξ )
2 2
......... ........ β2 βn F2 (ξ )Fn (ξ )⎥
⎢ ⎥
⎢ 0 0 0 ⎥
×⎢⎢ #1 # 1 #1 # 1 ⎥
β β (ξ )F (ξ ) β β (ξ )F (ξ ) β 2
(ξ )2
β β (ξ )F (ξ ) ⎥
⎢ 3 1 F 3 1 3 2 F 3 2 3 F 3 .......... 3 n F 3 n ⎥
⎢ ⎥
⎢ 0 0 0 0
⎥
⎢ ............. ........ ........ ....... ........... ⎥
⎢ ⎥
⎣ #1 #1 ⎦
βn β1 Fn (ξ )F1 (ξ ) βn Fn (ξ )
2 2
0 0
⎡ ⎤
#1 #1 #1 #1
⎢ F1 (ξ ) 2
F1 (ξ )F2 (ξ ) F1 (ξ )F3 (ξ ) ......... F1 (ξ )Fn (ξ )⎥
⎢ 0 0 0 0 ⎥
⎢#1 #1 #1 ⎥
⎢ ⎥
⎢ F2 (ξ )F1 (ξ ) F2 (ξ )2 ......... ........ F2 (ξ )Fn (ξ )⎥
⎢0 ⎥
GS1 L ⎢
⎢#1
0 0 ⎥
⎥
× dξ + #1 #1 #1
2 ⎢ ⎢ F3 (ξ )F1 (ξ ) F3 (ξ )F2 (ξ ) F3 (ξ )2 .......... F3 (ξ )Fn (ξ )⎥
⎥
⎢0 ⎥
⎢ 0 0 0
⎥
⎢ .............. ........ ........ ....... ............. ⎥
⎢1 ⎥
⎣# #1 ⎦
Fn (ξ )F1 (ξ ) Fn (ξ )2
0 0
× dξ + Gbr0 Cb F1 (0)Fj (0) etc. (2.5.85)
For first three modes this can simply be presented as
⎡ ⎤
K11 K12 K13
[K]i=1,3 j=1,3 = ⎣K21 K22 K23 ⎦ (2.5.86)
K31 K32 K33

where for i = j

EAβi2 GS1 L GS1 L EAβi G r0 C b
Kii = + + G b r0 C b + − sin 2βi + b cos 2βi
2L 2 4βi 4L 2
(2.5.87)
For i = j we have

EAβi βj GS1 L sin(βi − βj ) EAβi βj GS1 L sin(βi + βj )
Kij = − + −
2L 2 βi − β j 2L 2 βi + β j
+ Gb r0 Cb cos βi βj (2.5.88)
It should be noted at this point that there are no suggestive values available for S1
and Cb for higher modes either by Novak or any other research. However, it may
be reasonably stated that for higher modes the dimensionless frequency a0 would
be ≥1.0 (or near 1.0 at worse) when the curve for S1 becomes almost constant
(Novak 1974) and the values furnished in Table 2.5.6 may be used without much
error.
The value of β for the fundamental mode is already furnished in Table 2.5.9 for the
next two modes the values of β are furnished in Table 2.5.11 and Table 2.5.12.
Table 2.5.11 Roots of equation pL tan(pL) − η = 0 for second mode.
0 3.141 0.7 3.348 2 3.644 10 4.425

0.1 3.173 0.8 3.374 2.25 3.689 20 4.491
0.2 3.204 0.9 3.4 2.5 3.732 25 4.533
0.3 3.234 1 3.426 3.0 3.809 30 4.561
0.4 3.264 1.25 3.486 3.5 3.876 35 4.582
0.5 3.292 1.5 3.542 4 3.935 40 4.598
0.6 3.320 1.75 3.595 5 4.034 50 ∼
=3π/2
Table 2.5.12 Roots of equation pL tan(pL) − η = 0 for third mode.
0 6.28 0.7 6.392 2 6.578 15 7.316

0.1 6.299 0.8 6.407 2.25 6.611 20 7.495
0.2 6.315 0.9 6.422 2.5 6.643 25 7.56
0.3 6.331 1 6.437 3.0 6.704 30 7.606
0.4 6.346 1.25 6.474 3.5 6.761 35 7.639
0.5 6.362 1.5 6.510 4 6.814 40 7.665
0.6 6.377 1.75 6.544 5 6.910 50 ∼
=5π/2

Mass matrix is similarly given by

For i = j

γp AL sin 2βi
mii = 1+ + M cos2 βi (2.5.89)
2g 2βi
where M = Mass of machine plus pile cap

For i = j

γp AL sin(βi + βj ) sin(βi − βj ) M
mij = − + cos(βi + βj ) − cos(βi − βj )
2g βi + β j βi − β j 2
(2.5.90)
The damping matrix can be obtained as

For i = j
√
1 1 r0 ρGS2 L r0 ρb Gb C̄b
Cii = r0 ρGS̄2 L + r0 ρb Gb C̄b + sin 2βi + cos 2βi
2 2 4β 2
(2.5.91)
For i = j

sin(βi + βj ) sin(βi − βj )
Cij = r0 ρGS̄2 L −
βi + β j βi − β j
(2.5.92)
r2 ρGb C̄b
+ 0 cos(βi + βj ) − cos(βi − βj )
2
It is apparent that the dynamic analyses of piles with pile cap are standard and
the validity of the same would depend on how correctly the pile stiffness values have
been obtained. For this, it would worthwhile to evaluate how the present formulation
matches with other established methods. To this end, the pile stiffness as obtained by
Equations (2.5.59) and (2.5.62) has been compared with Novak’s chart (1974) and
equation based on rigorous analysis as proposed by Dobry and Gazetas (1988).
It should be noted that their expression is valid for floating piles of length say, L and
embedded in an elastic half space of length 2L. The results have been compared for a
single pile of various slenderness ratio λ varying from 20 to 100 and Ep /Gs value of soil
varying from 250 to 10,000 for an RCC pile of diameter of 600 mm and having Ep =
30 GPa. Poisson’s ratio value for soil considered is 0.4. Here Ep = Young’s modulus
of pile material; Gs = dynamic shear modulus of soil. The results for Kpile(bearing) and
Kpile(friction) are shown in Figure 2.5.13 through 20 for various slenderness ratios.
Finally, the natural frequency of a real life centrifugal compressor foundation
supported on 9 RCC piles, 45 meter long having diameter of 950 mm, have been
compared. The piles are spaced at 3.0 m c/c. The size of pile cap is 7 m × 5 m × 2.0 m,
embedded to depth of 1.4 meter. The weight of the generator supported on it weighs
400 kN. The frequencies are again compared for a range of Ep /Gs varying from 250
to 10000.

Pile stiffness L/r=20

4.00E+06
3.50E+06
Pile stiffness (kN/m)

3.00E+06
2.50E+06 Kpile(bearing)
2.00E+06 Novak(bearing)
1.50E+06 Gazetas
1.00E+06
5.00E+05
0.00E+00
250 500 1000 2500 10000
Ep/Gs
Figure 2.5.13 Comparison of bearing pile stiffness for slenderness ratio = 20.
Stiffness of pile for L/R=20

2.00E+06
Stiffness (kN/m)
1.50E+06
Kpile(friction)
1.00E+06 Novak(friction)
5.00E+05 Gazetas
0.00E+00
250 500 1000 2500 10000
Ep/Gs
Figure 2.5.14 Comparison of friction pile stiffness for slenderness ratio = 20.
Vertical stiffness for L/r=40

Stiffness (kN/m)
5.0000E+06
4.0000E+06 Kpile(bearing)
2.0000E+06
1.0000E+06 Gazetas
0.0000E+00
0
00
00
25
50
0
00
10
50
10
Ep/Gs
The results based on Kp(bearing) and Kp(friction) has been compared to Dobry and
Gazetas’ results and presented in Table 2.5.13. The results have not been compared
with Novak in this case for the charts are too crude especially in the range when
the ratio of Ep /Gs = 2500–10000 and significant variation can occur based on eye
estimate of stiffness function. Results have been found to be excellently matching
particularly for friction piles.

Stiffness of pile for L/r=40

3.00E+06
Stiffness (kN/m)
2.50E+06
2.00E+06
Kpile(friction)
1.50E+06
Novak(friction)
1.00E+06
Gazetas
5.00E+05
0.00E+00
250 500 1000 5000 10000
Ep/Gs
Pile stiffness L/r=80

6.0000E+06
5.0000E+06
4.0000E+06
Kpile(bearing)
3.0000E+06
Novak(bearing)
2.0000E+06 Gazetas
1.0000E+06
0.0000E+00
250 500 1000 5000 10000
Ep/Gs
Pile stiffness for L/r=80

5.00E+06
Stiffness (kN/m)
4.00E+06
3.00E+06
Kpile(friction)
2.00E+06 Novak(friction)
1.00E+06 Gazetas
0.00E+00
250 500 1000 5000 10000
Ep/Gs
As stated earlier, the results from Equation (2.5.41) (with appropriate boundary con-
dition for bearing and friction) have been compared with Novak’s chart and Dobry and
Gazetas’ expression. The results have been studied against both the bearing and friction
pile coefficients as suggested by Novak and El-Sharnouby (1983). It will be observed
in Figures 2.5.13 through 2.5.18 that the frictional stiffness values obtained are very

Pile stiffness (kN/m) Pile stiffness L/r=100

6.0000E+06
5.0000E+06
4.0000E+06
Kpile(bearing)
3.0000E+06
1.0000E+06 Gazetas
0.0000E+00
250 500 1000 5000 10000
Ep/Gs
Pile stiffness for L/r=100

6.00E+06
5.00E+06
4.00E+06
Kpile(friction)
3.00E+06
Novak(friction)
2.00E+06
Gazetas
1.00E+06
0.00E+00
250 500 1000 5000 10000
Ep/Gs
Table 2.5.13 Variation of vertical frequency for compressor foundation.
Freq (rad/sec) Freq (rad/sec) As per Dobry and

Sl. No. Ep /Gs for Kpile (bearing) for Kpile (friction) Gazetas (rad/sec)
1 250 196 195 197

2 500 139 138 139
3 1000 99 98 98
4 2500 64 62 62
5 5000 47 44 44
6 7500 39 36 36
7 10000 35 31 31
close to Dobry and Gazetas’ results in all the cases for various L/r and E/Gs values.
For the bearing piles, the values obtained are slightly higher than Dobry and Gazetas’
values but matching very closely to Novak’s data from Ep /Gs = 500 onwards. This
is expected. It was pointed out by Novak and others that bearing stiffness for a pile is
slightly more than that of friction stiffness.
At L/r = 20 the bearing values obtained are higher than that of Dobry and Gazetas
(which is logical considering his case is that of a floating pile) as well as from Novak
but the difference reduces considerably from Ep /Gs = 1000 onwards, and this is

the range in which piles are commonly used in practice. The values, where Ep /Gs
is ≤1000 are actually far too stiff for any piles to be bored or driven. Moreover,
a pile with L/r = 20 is actually a fictitious values. For instance a standard pile of
length 30 meter, the diameter becomes 3.0 m, which is actually a cassion and not a
pile. It is possibly in such cases, the axial stiffness is far too high and this shows a
significant higher stiffness in bearing compared to friction piles for such an unrealistic
L/r ratio. For real life problems, the values of L/r is around 50–100 and Ep /Gs >
1000. It will be observed that the values obtained by the proposed method are quite
close to the reported results useful for practical ranges of application. As for the
frequencies obtained for various Ep /Gs values the results in Table 2.5.10 are extremely
encouraging.
2.5.10.9 What is the major advantage of this model?

The major advantage with the proposed method is that instead of solving the differ-
ential equation (especially when the boundary condition gets complicated with cases
like partial embedment or variable soil) the stiffness, damping and mass matrices
are directly derived from energy principles and the subsequent derivation gets quite
simplified.
Finally, the formulation have been derived for a general case when pile can act
both as bearing and friction pile for which no direct solutions are available-and
this could be the reality in many cases when the pile is neither in full bearing or
full floating. Comparing the results it can be well inferred that the method can
be used for practical design office work without the limitations as stated at the
outset.
2.5.10.10 Design steps

Based on the derivations presented, the design steps may be summarized as follows:
• Determine the soil properties like G, Gb , Gf and ν (Poisson’s ratio of the soil);
• Determine the pile properties like Length of pile L and diameter of pile (2r0 ) and
also the Young’s Modulus E of the pile material;
• Determine the pile cap property like its mass and depth of embedment Df ;
• Determine the weight of machine supported on the pile cap;
• Obtain Novak’s stiffness and damping coefficients Cb , S1 , C b , S2 from Table 2.5.9
and Table 2.5.10, Equation (2.5.29) etc.;
• Establish the dimensionless parameter η = (Gb /E) (Cb /π ) λ;
• For the given value η determine the value of pL from Table 2.5.9;
• If the pile is bearing (known priori) β = π /2;
• Consider β = pL;
• Determine K1 and mp from Equations (2.5.40) and (2.5.50) respectively;
• Determine the embedment stiffness matrix from the Equation (2.5.79);
• Form the mass, stiffness;
• Perform eigen solution;
• Find the value of the frequency and obtain the dimensionless frequency number a0 ;
• Find the value of S2 from Beredugo’s expression as given in Table 2.5.10;

• Determine the damping of the system based on Equations (2.5.57), (2.5.60) &
(2.5.63);
• Perform Modal or time history analysis to obtain the amplitude of vibration.
2.5.11 Dynamic analysis of laterally loaded piles
2.5.11.1 Piles under dynamic lateral loading

Having presented the mathematical model of vibration in vertical direction, we now
present a model of vibration of piles under lateral or horizontal load. This is an impor-
tant study for the pile supporting rotating machines under centrifugal and reciprocating
loads.
In majority of cases it has been found that of all modes (like vertical, rocking,
yawning, twisting etc.); lateral vibration (coupled with rocking) is most critical and
often governs the design. Thus a study of such motion is of paramount importance
for piles supporting important installations and also for facilities, which are valuable
to the community under earthquake threats. Recall Novak’s method described earlier
for lateral pile you will realize on retrospection that the model has got the following
limitations
• The method is coefficient based [function of the ratio of Young’s modulus of pile
(Ep ), and dynamic shear modulus of soil (Gs ), as such for intermediate values one
has to interpolate which may not be always very accurate.
• The values are given for Poisson’s ratio of 0.25 and 0.40 only. Thus for any
value between 0.25 and 0.4, or beyond 0.4 another set of linear interpolation/
extrapolation is necessary.
• Novak and El Sharnouby (1983) has given stiffness and damping coefficients for
soil having parabolic profile but in many cases the variation is linear and no
coefficients are available for this case.
• The method does not have a solution for partially embedded piles, which is of great
practical importance for piles driven in arctic condition (especially in Northern
Siberia which constitute of a large number of Oil and Gas facilities).
• The dynamic bending moment and shear force induced on pile cannot be evaluated.
• Finally the formulation is valid for long piles (i.e. the failure takes place in the pile
body before soil yields) and do not cater to piles, which are short.
The simplified formulas given by Dobry and Gazetas (1988) is based on more rigor-
ous analysis, however it also does not address the issues of partial embedment, dynamic
bending moment and shear, or the issue- if the pile is short (i.e. L/r < 25) etc.
We now present herein (Chowdhury and Dasgupta 2008) a mathematical model for
analysis of such piles under lateral load that overcomes many of the bottle necks cited
above.
Similar to the vertical vibration model presented earlier the present formulation is
based Novak and Beredugo’s (1972) formulation for a rigid cylinder embedded in
elastic half space. Shown in Figure 2.5.21 is a pile embedded in homogeneous elastic
medium and considered under plane strain condition. The pile is considered long and
slender, to start with. Under static conditions, the equation of equilibrium in the

Z
dz
Soil Stiffness = GSx1
Figure 2.5.21 Conceptual model of pile under lateral load.
x-direction [similar to beams on elastic foundation] is given by Timoshenko (1956) as
d4x
E p Ip = −ks x (2.5.93)
dz4
where, Ep = Young’s modulus of the pile; Ip = moment of inertia of the pile cross
section; ks = elastic stiffness of the soil and is expressed as GSx1 ; G = dynamic
shear modulus of the soil; Sx1 = Beredugo’s constant which are basically frequency
dependent.
However, it has been shown by Novak and Beredugo (1972) that considering this
term frequency independent, no accuracy is lost for practical design problems and
the analysis becomes quite simplified for rigid circular embedded footing. Elaboration
about this parameter, in terms of piles, will be made later.
The general solution of Equation (2.5.93) may be written as
x = e−pz (C0 cos pz + C1 sin pz) + epz (C2 cos pz + C3 sin pz) (2.5.94)

GSx1
where p = 4
.
E p Ip
For long piles under load or moment at its head, it is reasonable to assume that
at significant distance from the pile head (where the load is applied), the curvature
vanishes. This condition can only be satisfied when C2 and C3 in Equation (2.5.94) is

considered insignificant. Hence, the deflection equation can be taken as
x = e−pz (C0 cos pz + C1 sin pz) (2.5.95)
Considering the pile head undergoing specified deflection and rotation as well as
its head is fixed to the pile cap (same boundary condition as considered by Novak
(1974)), one can have [Figure 2.5.21],
At z = 0, let x = x0 ⇒ C0 = x0 , which gives
x = e−pz (x0 cos pz + C1 sin pz) (2.5.96)
Again, at z = 0, dx
dz
= θ0 one can have
θ0
C1 = x0 + (2.5.97)
p
Thus Equation (2.5.98) can now be represented as

−pz θ0
x=e x0 cos pz + x0 + sin pz (2.5.98)
p
For magnitude of rotation being small θ0 ∼

= x0 /L, x may be written as

x0
x = e−pz x0 cos pz + x0 + sin pz (2.5.99)
pL

x −pz 1
=e cos pz + 1 + sin pz (2.5.100)
x0 pL
Now considering β = pl and using Equation (2.5.100), for any arbitrary loading,
the generic shape function in dimensionless form can be represented as

−βz βz 1 βz
φ(z) = e L cos + 1+ sin (2.5.101)
L β L
in which

GSx1 L4
β= 4
, L being the length of the pile. (2.5.102)
Ep Ip
Equation (2.5.101) can be further reduced to

−βz βz βz
φ(z) = e L cos + η sin (2.5.103)
L L
1
where η = 1 + (2.5.104)
β

The generic shape function of the pile for the fundamental mode as in Equation
(2.5.103) is shown in Figure 2.5.22 for Ep /G = 5000.
The potential energy d of an element of depth dz, shown in Figure 2.5.21 is then
given (Shames and Dym 1995) by
2
Ep Ip d2v Kh 2
d = + v (2.5.105)
2 dz2 2
where, Ep = Young’s modulus of pile; Ip = moment of inertia of pile; Kh = lateral

dynamic stiffness of soil; v = displacement of the pile in the x direction and may be
written as [φ(z)q(t)].
For a rigid circular embedded footing of embedment Df , the stiffness of the footing
may be expressed (Beredugo and Novak (1972)) as
Kh = Gb r0 Cb + GDf Sx1 (2.5.106)
where, Kh = foundation stiffness in horizontal direction; G = dynamic shear modu-

lus of the soil along foundation surface; Gb = dynamic shear modulus of soil at the
foundation base; r0 = radius of the foundation; Cb and Sx1 = constants which are
basically frequency dependent.
Ignoring the first term in Equation (2.5.106), which represents the contribution
of base resistance, and substituting the same in Equation (2.5.105), for a cylindrical
element of depth dz, embedded in soil, the potential energy d may be expressed as
2
Ep Ip d2v GSx1 dz 2
d = + v . (2.5.107)
2 dz2 2
1.2
1
Shape function
0.8
0.6
0.4
0.2
0
-0.2
9
15
45
75
0
0.
0.
0.
0.
0.
0.
z/L
Figure 2.5.22 Generic shape function long pile in the horizontal mode for Ep /G = 5000.

The total potential energy over the length of the pile (L) is then given by
L 2 L
Ep Ip d2v GSx1
= dz + v2 dz (2.5.108)
2 dz2 2
0 0
Considering v (z, t) = φ(z)q(t), it can be proved (Hurty and Rubenstein 1967) that
L L
Kij = Ep Ip φi (z)φj (z)dz + GSx1 φi (z)φj (z)dz (2.5.109)
0 0

For the fundamental mode, stiffness of the pile is then given by
L L
K = Ep Ip φi (z)2 dz + GSx1 φi (z)2 dz (2.5.110)
0 0
On double differentiation, Equation (2.5.103) reduces to

2β 2 − βz βz βz
φ (z) = e L sin − η cos and (2.5.111)
L2 L L

4β 4 2βz X Y 2βz 2βz
φ (z) = 4 e− L
2
− cos − η sin (2.5.112)
L 2 2 L L
where, X = 1 + η2 ; Y = 1 − η2 and η is given in Equation (2.5.104).

Again from Equation (2.5.103)

2 − 2βz X Y 2βz 2βz
φ(z) = e L + cos + η sin (2.5.113)
2 2 L L
Substituting Equations (2.5.112) and (2.5.113) in Equation (2.5.110), the stiffness

reduces to
L
4Ep Ip β 4 2βz X Y 2βz 2βz
K= e− L − cos − η sin dz
L4 2 2 L L
0
L
2βz X Y 2βz 2βz
+ GSx1 e− L + cos + η sin dz (2.5.114)
2 2 L L
0

Equation (2.5.114) on integration by parts and on simplification may be expressed as

!
4Ep Ip β 4 X L Y L −2β
K= (1 − e−2β ) − e (sin 2β − cos 2β) + 1
L4 2 2β 2 4β

4Ep Ip β 4 ηL −2β
− (1 − e (sin 2β + cos 2β))
L4 4β

X L Y L −2β
+ GSx1 (1 − e−2β ) + (e (sin 2β − cos 2β) + 1)
2 2β 2 4β

ηL
+ GSx1 (1 − e−2β (sin 2β + cos 2β)) (2.5.115)
4β
In Equation (2.5.115), e−2β (sin 2β + cos 2β) and e−2β (sin 2β − cos 2β) may be
ignored as their values are exceedingly small (highest is of the order 10−3 and the
lowest is 10−30 for Ep /G value varying from 250 to 10,000) and has practically no
effect on the stiffness value and this also considerably simplifies the expression.
Based on the above simplification, Equation (2.5.115) may be rewritten as

4Ep Ip β 4 X L Y L ηL
K= (1 − e−2β ) − −
L4 2 2β 2 4β 4β

X L Y L ηL
+ GSx1 (1 − e−2β ) + + (2.5.116)
2 2β 2 4β 4β

Ep Ip β 3 −2β Y
➔ K= X(1 − e ) − − η
L3 2

GSx1 L Y
+ X(1 − e−2β ) + + η (2.5.117)
4β 2
Taking Ep Ip β 3 /L3 as common in Equation (2.5.117) and substituting the value of

β from Equation (2.5.102a), Equation (2.5.117) reduces to

Ep Ip β 3 5X 3Y 3η
K= (1 − e−2β ) − − which can be further simplified to
L3 4 8 4

5X −2β ) − 3Y − 34 η
Ep Ip 4 (1 − e 8
K= 3 3
(2.5.118)
L (η − 1)

The accuracy of Equation (2.5.118) will be dependent on the correct selection of

Sx1 . For instance for rigid circular footing Beredugo and Novak (1972) has furnished
a frequency independent value of Sx1 = 4.0 to 4.1 (depending on Poisson’s ratio). This
has been found to give adequate accuracy for practical engineering design. Comparing
the stiffness data with Novak (1974), Dobry and Gazetas (1988), it is proposed that the
following values of Sx1 as furnished in Tables 2.5.14 to 16 be used for the calculation
of dynamic response of the pile in the lateral direction.
For a particular pile having specific slenderness ratio and Poisson’s ratio of the
soil, the value of Sx1 can be selected from Tables 2.5.14, 2.5.15 and 2.5.16 and on
substitution of the same in Equation (2.5.102), Equation (2.5.118), gives the solution
of pile stiffness in the lateral direction.
Table 2.5.14 Suggested value of Sx1 for Poisson’s ratio of soil = 0.25.
L/r0
Poisson’s (Slenderness
ratio ratio) Sx1 (250) Sx1 (500) Sx1 (1000) Sx1 (2500) Sx1 (5000) Sx1 (10000)
0.25 25 2.00 1.83 1.66 1.43 1.25 1.07

40 2.19 2.05 1.90 1.70 1.55 1.39
60 2.30 2.17 2.05 1.87 1.74 1.60
80 2.36 2.24 2.12 1.96 1.84 1.71
100 2.39 2.28 2.17 2.01 1.90 1.78
Note: The value in parenthesis after Sx1 indicates Ep /Gs value of the soil.
L/r0
0.40 25 2.27 2.08 1.89 1.63 1.43 1.23

40 2.48 2.32 2.16 1.94 1.76 1.59
60 2.60 2.46 2.31 2.12 1.97 1.82
80 2.66 2.53 2.40 2.22 2.08 1.94
100 2.70 2.57 2.45 2.28 2.15 2.02
L/r0
0.50 25 2.45 2.25 2.05 1.77 1.55 1.34

40 2.67 2.50 2.33 2.09 1.91 1.72
60 2.80 2.65 2.50 2.29 2.13 1.96
80 2.87 2.72 2.58 2.39 2.24 2.10
100 2.91 2.77 2.63 2.45 2.32 2.18

2.5.11.1.1 Estimation of contribution of pile mass

The mass matrix of the pile may be expressed as (Meirovitch 1967)

Mx = m(x) φi (z)φj (z)dz (2.5.119)
For the present case of the pile of length L, Equation (2.5.118), can be expressed as
L
γp Ap
Mx = φ(z)2 dz (2.5.120)
g
0
where, γp = unit weight of the pile material; Ap = cross sectional area of the pile;
g = acceleration due to gravity.
L
γp A p − 2βz X Y 2βz 2βz
or, Mx = e L + cos + η sin (2.5.121)
g 2 2 L L
0
Equation (2.5.121) on integration and after simplification gives

γp Ap L X(1 − e−2β ) + Y
2 +η
Mx = (2.5.122)
4g β
Equation (2.5.122) is the inertial contribution of the pile material for the fundamen-
tal mode. Incidentally, the inertial effect is usually ignored in design but could have
significant effect if the number of piles is large in a pile group.
2.5.11.2 Radiation damping for pile under lateral load

For a rigid footing embedded in soil for a depth Df , Beredugo and Novak (1972) have
proposed the expression

Cz = r0 ρb Gb C̄b + r0 ρG S̄2 Df (2.5.123)
where, r0 = radius of the foundation; Gb = dynamic shear modulus at the foundation

base; G = dynamic shear modulus of the soil in which the foundation is embedded;
Df = depth of embedment; C̄b and S̄2 = frequency independent constants as defined
by Novak and Beredugo (1972).

With reference to Figure 2.5.21 for a pile element of length dz embedded in the soil,
and ignoring the bearing effect, Equation (2.5.123) may be expressed as

c(x) = r0 ρGSx2 dz (2.5.124)
For systems having continuous response function, the damping may be expressed as

Cx = c(x) φi (z)φj (z)dz (2.5.125)
For the pile of length L, Equation (2.5.125) may be expressed as
L
Cx = r0 ρGSx2 φ(z)2 dz (2.5.126)
0
L
− 2βz X Y 2βz 2βz
or, Cx = r0 ρGSx2 e L + cos + η sin (2.5.127)
2 2 L L
0
On integration and after simplification Equation (2.5.127) reduces to

X(1 − e−2β ) + Y
+η
2
Cx = r0 ρGSx2 L (2.5.128)
4β
Equation (2.5.128) expresses the soil damping for a single pile under horizontal
mode of vibration. The factor Sx2 is a frequency dependent damping coefficient. The
damping factor is required for calculating the amplitude only after the eigen solution of
the problem is already done vis-a-vis, the dimensionless frequency number a0 = ωr0 /vs
term is known a priori. Polynomial fit curve for Sx2 are available in terms of a0 which
can be used directly to obtain these parameters. Sx2 for different Poisson’s ratios are
given in Table 2.5.17.
Table 2.5.17 Values of Sx2 (Beredugo & Novak 1972).
Poisson’s ratio Sx2

0.8652a0
0.0 7.334a0 +
a0 + 0.00874
41.59a0
0.25 0.83a0 +
a0 + 3.90
56.559a0
0.5 0.96a0 +
a0 + 4.68

2.5.11.3 Material damping of pile

The structural stiffness contribution of the pile is given in the first part of Equation
(2.5.117), while that of the mass is given in Equation (2.5.122).
Thus, if Cc be the
critical damping of the pile then it can be expressed as Cc = 2 Kmp ; K (the first term
in Equation (2.5.117)) and mp being the stiffness and mass of the pile.
Depending on the material used for pile (like RCC, steel etc.) a suitable damping
ratio (D) can be assumed. The damping (Cp ) for the pile can then be expressed as
Cp = DCc (2.5.129)
This, when added to the radiation damping, calculated through Equation (2.5.128)
gives the complete damping quantity for the soil-pile system.
2.5.11.4 Piles with other boundary conditions

Having established the stiffness, mass and damping of the pile in lateral direction
based on minimization of the potential energy of the system, the above method can be
extended for the piles with other boundary conditions for which there are no standard
solutions available.
2.5.11.5 Partially embedded piles

In Arctic and North Siberian condition, due to environmental reasons, the steel piles are
driven into the ground when they protrude about 2–3 m above the ground over which
the pile cap and vibrating equipments are placed. Piling configuration has already been
shown earlier while explaining the vertical vibration of pile. In such cases the existing
solutions cannot be used. However, a solution of the same is proposed hereunder. Let
L be the full length of the pile and the length of the embedment in soil be L1 (refer
Figure 2.5.10).
For this case, one may write

4
4 GSx1 L1
βe = (2.5.130)
E p Ip
Here subscript “e” represents embedment of the pile.
The shape function can thus be represented by

− βLe z βe z βe z
φ(z) = e 1 cos + η sin (2.5.131)
L1 L1

2βe2 − βLe z βe z βe z
and φ (z) = e 1 sin − ηe cos (2.5.132)
L21 L1 L1

4βe4 − 2βLe z Xe Ye 2βe z 2βe z
and hence φ (z)2 = e 1 − cos − ηe sin (2.5.133)
L41 2 2 L1 L1
1
where, Xe = 1 + ηe2 ; Ye = 1 − ηe2 and ηe = 1 + .
βe

Now, considering the fact that the embedment of a pile does not have any effect on
the shape function of the system, the stiffness of the pile for the fundamental mode
may be written as
L L1

K = Ep Ip φi (z) dz + GSx1 φi (z)2 dz
2
(2.5.134)
0 0
%
Considering, α = L L1 , Equation (2.5.134) may be rewritten as

α L1
4Ep Ip βe4 − 2βLe z Xe Ye 2βe z 2βe z
K= e 1 − cos − ηe sin dz
L41 2 2 L1 L1
0
L1
− 2βLe z Xe Ye 2βe z 2βe z
+ GSx1 e + cos + ηe sin dz (2.5.135)
2 2 L1 L1
0
Equation (2.5.135) on integration by parts and after simplification, may be

expressed

Ep Ip βe3 1 1 α 1 −2βe α −2βe (1−α)
K= Xe +α +Ye − +ηe − α −Xe e e +1
L31 4 8 2 4 4
(2.5.136)
this can be further simplified to

!
Ep Ip Xe 1
4 + α + Ye 18 − α2 + ηe 14 − α − Xe e−2βe α4 e−2βe (1−α) + 1
K=
L31 (ηe − 1)3
(2.5.137)
Equation (2.5.137) gives the solution for stiffness of a partially embedded pile in the
ground. The correctness of the equation can be back checked by the fact that when
the pile becomes fully embedded i.e. for L1 = L α → 1, βe = β, Xe = X etc., when
Equation (2.5.137) degenerates to Equation (2.5.118).
Proceeding in an identical manner as done before, the mass and damping terms may
be computed as

γp Ap L1 Xe α(1 − e−2βe ) + Y2e α + ηe α
Mx = (2.5.138)
4g 1/(ηe − 1)

Xe (1 − e−2βe ) + Y2e + ηe
Cx = r0 ρGSx2 L1 (2.5.139)
4/(ηe − 1)

2.5.11.6 Pile embedded in soils with varying elastic property

We present now the effect of variation of shear modulus with respect to depth. In the
previous section, the calculation of stiffness as well as the damping of soil was based on
constant dynamic shear modulus of the soil. For varying shear modulus, the variation
with depth can be expressed as
G = G(z/L)m (2.5.140)
where m = a number varying from 0–2 [considered 0 when G is constant with depth,
assumed 1 for linear variation and 2 for parabolic distribution].
For a linearly varying soil the stiffness matrix can be written as
L
4Ep Ip β 4 2βz X Y 2βz 2βz
K= e− L − cos − η sin dz
L4 2 2 L L
0
L
z − 2βz X Y 2βz 2βz
+ GSx1 e L + cos + η sin dz (2.5.141)
L 2 2 L L
0
Integration of above and ignoring the terms containing the factor, βe−2β · cos 2β,
β · e−2βsin 2β etc., having extremely small contributions, Equation (2.5.141) reduces to

Ep Ip β 3 −2β Y GSx1 L −2β 3Y η
K= X(1−e )− −η + X[1 − e (1 + β)] + +
L3 2 4β 2 4 2
(2.5.142)
and can be further simplified to

Ep Ip β 3 −2β 1 1 1 3 1
K= X 1 − e 1 + + − Y − β − η 1 −
L3 4 4β 2 16 8β
(2.5.143)
The damping matrix for this case, proceeding in same manner as outlined earlier,
can be represented by

r0 ρG Sx2 L −2β 3Y η
Cx = X[1 − e (1 + β)] + + (2.5.144)
4β 2 4 2
The mass coefficient remains the same as expressed in Equation (2.5.122).

When the dynamic shear modulus variation is parabolic with depth, the stiffness
equation of the pile can be expressed as
L
4Ep Ip β 4 − 2βz X Y 2βz 2βz
K= e L − cos − η sin dz
L4 2 2 L L
0
L z 2
2βz X Y 2βz 2βz
+ GSx1 e− L + cos + η sin dz (2.5.145)
L 2 2 L L
0
Equation (2.5.145) on integration and on subsequent simplification reduces to

Ep Ip β 3 −2β Y
K= X(1 − e )− −η
L3 2

GSx1 L 1 −2β 1 2
+ X − e 2 + − (2.5.146)
4β 4β 2 β β2
which can be further simplified to

Ep Ip β 3 1 −2β 3 1 1 Y
K= X 1 + − e + − − − η (2.5.147)
L3 16β 2 2 4β 8β 2 2
Equation (2.5.147) gives the stiffness expression of pile under parabolic variation
of G along the length of pile.
Proceeding in same manner as stated above the damping matrix may be expressed as
√
r0 ρGSx2 L 1 −2β 1 2
Cx = X −e 2+ − 2 (2.5.148)
4β 4β 2 β β
The mass coefficient remains the same as expressed in Equation (2.5.122).
2.5.11.7 Computation of bending moment and shear force

For machine foundation subjected to a lateral load of P0 sin ωm t, the amplitude of
vibration is given by
P0
K sin ωm t
v(t) = (2.5.149)
(1 − r2 )2 + (2Dr)2
where, ωm = operating frequency of the machine; P0 = unbalanced dynamic force;

r = ωm /ωn = the ratio of operating and natural frequency; D = damping ratio of the
system.

Thus the peak amplitude is given by
P0
K
v(t) = (2.5.150)
(1 − r )2 + (2Dr)2
2
The complete displacement function is then given by
P0
K
v(z, t) = φ(z) (2.5.151)
(1 − r2 )2 + (2Dr)2
P0
βz βz βz
or v(z, t) = K
e− L cos + η sin (2.5.152)
(1 − r ) + (2Dr)
2 2 2 L L
The bending moment is given by
Ep Ip P0
Ep Ip v = −M(z) = − K
e L sin − η cos
(1 − r )2 + (2Dr)2
2 L2 L L
(2.5.153)
E p I p P0
K 2β 2 − βz βz βz
Mz = e L sin − η cos (2.5.154)
(1 − r2 )2 + (2Dr)2 L2 L L
The maximum moment will be at the head i.e. at z = 0, and it can be expressed as
2Ep Ip P0
K β(β + 1)
Mmax = (2.5.155)
L2
(1 − r2 )2 + (2Dr)2
The shear force is given by
Ep Ip P0
2β 3 βz βz
Ep Ip v = − V(z) = K
3
(η − 1) sin + (η + 1) cos or
2 L L L
(1 − r2 )2 + (2Dr)
Ep Ip P0
K 2β 3 βz βz
V(z) = − 3
(η − 1) sin + (η + 1) cos (2.5.156)
2 L L L
(1 − r2 )2 + (2Dr)
2.5.11.8 Dynamic response of short piles

in the horizontal mode
There are no solutions till date for this type of piles. Existing solutions are based on
long piles with the implicit assumption that under ultimate load piles fail before the

soil. However there are number of areas (e.g. Bonny River Delta in Nigeria, where the
topsoil constitute of very weak clay underlain by dense sand) where the soil will yield
much before the pile. Broms (1965) has shown that the displacement curvatures for
such piles are completely different than that of long piles.
While a long pile embedded in soil behaves as a semi-infinite beam on elastic foun-
dation, a short pile behaves as a beam of finite length on elastic foundation. Bojtsov
et al. (1982) has given solution to the generic displacement curves of such short beams
on elastic foundation that is given by
x = C0 cos hpz cos pz + C1 cos hpz sin pz + C2 sin hpz sin pz + C3 sin hpz cos pz
(2.5.157)
where p is same as expressed in Equation (2.5.96).

Expressing in terms of Puzrevsky function (Karnovsky and Lebed 2001), Equation
(2.5.157) can be expressed as
x = C0 V0 (pz) + C1 V1 (pz) + C2 V2 (pz) + C3 V3 (pz) (2.5.158)
where, V0 (pz) = cosh pz cos pz (2.5.159)
1
V1 (pz) = √ (cosh pz sin pz + sinh pz cos pz) (2.5.160)
2
V2 (pz) = sinh pz sin pz (2.5.161)
1
V3 (pz) = √ (cosh pz sin pz − sinh pz cos pz) (2.5.162)
2
Puzrevsky’s functions, defined below, have some unique functional properties,

which will be used for subsequent analysis for derivation of the stiffness, damping
and mass of the piles.
V0 (0) = 1; V0 (0) = 0; V0 (0) = 0; V0 (0) = 0 (2.5.163)

√
V1 (0) = 0; V1 (0) = p 2, V1 (0) = 0; V1 (0) = 0 (2.5.164)
V2 (0) = 0; V2 (0) = 0; V2 (0) = 2p2 , V2 (0) = 0 (2.5.165)

√
V3 (0) = 0; V3 (0) = 0; V3 (0) = 0; V3 (0) = 2 2p3 (2.5.166)
√ √
V3 (pz) = p 2V2 (pz); V2 (pz) = p 2V1 (pz) (2.5.167)
√ √
V1 (pz) = p 2V0 (pz); V0 (pz) = p 2V3 (pz) (2.5.168)
For a solution of the short pile one may use the model shown in Figure 2.5.23.
For the analysis (similar to long piles) the pile may be assumed as fixed at base and
can undergo deflection and rotation at the pile head.

M
P
X
Z
dz
Soil Stiffness=GSx1
Figure 2.5.23 Conceptual model of short pile under lateral load.
Considering base of pile at z = 0, shown in Figure 2.5.23, one may write
At z = 0, x = 0 ⇒ C0 = 0
At z = 0, x = 0 ⇒ C1 = 0
which gives, x = C2 V2 (pz) + C3 V3 (pz) (2.5.169)
At the pile head, i.e. at z = L x = 1 yielding,
C2 V2 (pL) + C3 V3 (pL) = 1 (2.5.170)
Again at z = L x = 1/L which gives,
C2 V2 (pL) + C3 V3 (pL) = 1/L. (2.5.171)
Using Equations (2.5.167) and (2.5.170), one may write
1
C2 V1 (pL) + C3 V2 (pL) = √ (2.5.172)
pL 2
The above may be expressed in matrix form as

& '
[V] {C} = p (2.5.173)
which can be further reduced to
{C} = [V]−1 {p} (2.5.174)

Performing the above operation gives

C2 1 V2 (pL) −V3 (pL) 1√
= (2.5.175)
C3 −V1 (pL) V2 (pL) 1/pL 2
where = V22 (pL) − V1 (pL)V3 (pL) which implies

1 V3 (pL) 1 V2 (pL)
C2 = V2 (pL) − √ and C3 = √ − V1 (pL) (2.5.176)
pL 2 pL 2
Thus, the displacement for the given boundary condition is then expressed as

1 V3 (pL) 1 V2 (pL)
x= V2 (pL) − √ V2 (pz) + √ − V1 (pL) V3 (pz) (2.5.177)
pL 2 pL 2
Based on above, the generic shape function in dimensionless form is given by

1 V3 (β) βz 1 V2 (β) βz
φ(z) = V2 (β) − √ V2 + √ − V1 (β) V3
β 2 L β 2 L
(2.5.178)
where the determinant gets modified to = V22 (β) − V1 (β)V3 (β).

Considering A = C2 / and B = C3 / the shape function can now be expressed as

βz βz
φ(z) = AV2 + BV3 (2.5.179)
L L
A typical shape function for the short piles Ep /Gs = 2500 is shown in Figure 2.5.24.
0.2
0
0
1
1
9
0.
0.
0.
0.
0.
0.
0.
0.
0.
Shape Function
-0.2
-0.4
-0.6
-0.8
-1
z/L
Figure 2.5.24 Generic shape function of short pile for Ep /G = 2500.

Differentiating Equation (2.5.179) and using properties mentioned earlier one

could have

2β 2 βz βz
φ (z) = 2
AV 0 + BV 1 (2.5.180)
L L L
Substituting the above functions in Equation (2.5.110), the stiffness can be

expressed as
L 2
4Ep Ip β 4 βz βz
K= AV 0 + BV 1 dz
L4 L L
0
L 2
βz βz
+ GSx1 AV 2 + BV 3 dz (2.5.181)
L L
0
Equation (2.5.181) is too complicated to solve in closed form and a numerical

quadrature scheme may be used to obtain K.
Considering ξ = z/L we have L · dξ = dz and as z → L; ξ → 1; as z → 0 ξ → 0;
which gives
1 L
4Ep Ip β 4
K= [AV0 (βξ ) + BV1 (βξ )] Ldξ + GSx1 [AV2 (βξ ) + BV3 (βξ )]2 Ldξ
2
L4
0 0
(2.5.182)
Substituting the value of β [Equation (2.5.102)] in Equation (2.5.182), the stiffness

may be written as
⎡ ⎤
1 1
K = GSx1 L ⎣4 [AV0 (βξ ) + BV1 (βξ )] dξ + 2
[AV2 (βξ ) + BV3 (βξ )] dξ ⎦
2
0 0
(2.5.183)
➔ K = GSx1 L[4I1 + I2 ] (2.5.184)
1 1
2
where I1 = [AV0 (βξ ) + BV1 (βξ )] dξ and I2 = [AV2 (βξ ) + BV3 (βξ )]2 dξ
0 0
(2.5.185)
The integrals I1 and I2 can very easily be solved by using Simpson’s 1/3rd rule
between limits 0–1 and can be back substituted in Equation (2.5.184) to compute the
stiffness for the short pile.

However, one should note that there is no theoretical or experimental benchmarking

against which the stiffness values can be checked or compared. So, use of the expression
must always be backed up by dynamic field test of the piles to adjust the data (especially
Sx1 or Ep /G) to match with the field observed values. In absence of comparative
benchmarks the design may be initiated with the suggestive values of Sx1 for various
Ep /Gs given in Table 2.5.18.
The values mentioned in Table 2.5.18, are based on the formulation for long pile
(with L/r < 25) but may be used as a starting point for the iteration based on field
observed data.
The mass of pile for the fundamental mode is given by
L
γp Ap
Mx = φ(z)2 dz
g
0
1
γp Ap L
or Mx = [AV2 (βξ ) + BV3 (βξ )]2 dξ (2.5.186)
g
0
γp Ap L
Mx = I2 (2.5.187)
g
To start the design a value of Sx1 is selected for specific Ep /Gs from Table 2.5.18

and find out the value of the frequency ( K/M) based on Equations (2.5.184)
and (2.5.187). Let this be defined as ωc where the subscript c stands for the word
“computed”. Let the field-tested natural frequency of the pile be ωf , where, ωf = ωc .
In most of cases it has been seen (Jadi 1999) that the field observed frequency value
deviates from the computed ones and usually varies by about 30–40%. This is logical,
for when the pile is bored or driven the soil gets displaced and clayey soil may loose a
part of its shear strength thus resulting in reduced dynamic shear modulus compared
to the value observed during geo-technical investigation. There could be cases where
the field observed values might be more than the computed ones, especially in sandy
soil where the soil gets densified due to pile driving. The bottom line is that in rare
cases the computed and observed values would match.
Table 2.5.18 Suggested for Sx1 for short piles (L/r ≤ 20) for field data
iteration.
Ep /Gs Sx1 (ν = 0.25) Sx1 (ν = 0.4) Sx1 (ν = 0.5)
250 1.53 1.75 1.89

500 1.35 1.54 1.68
1000 1.17 1.34 1.46
2500 0.95 1.09 1.46
5000 0.95 1.09 1.46
10000 0.95 1.09 1.46

Based on the above argument the error (ε) in the analysis is then given by
ε = ωc − ωf
For ε → 0 we have ωc = ωf → ωc2 = ωf2 .

K
Considering ωc2 = Mx and using Equations (2.5.184) and (2.5.187), one can have

GSx1 g I1
4 + 1 − ωf2 = 0 (2.5.188)
γp A p I2
It will be observed that all the factors β, I 1 , I2 in Equation (2.5.188) is a function

of Ep /G. The difference (= the error ε) can now be set to zero or minimum by varying
the value of Ep /G for which lim ε → 0.
This can very easily be done by using the standard solver or goal-seek in a spreadsheet
with boundary constraint that Sx1 > 0.
The solver basically uses an algorithm called generalized reduced gradient technique
(GRG2) used for constrained optimization (Lasdon et al. 1978). The procedure begins
with the nonlinear optimization technique with equality constraints. The necessary
slack and surplus variables are added as xs or x2s to any inequality constraints, and the
problem is to
Optimize: y(x)
Subject to: fi (x) = 0: for i = 1, 2, . . . , j
where j is the number of constrained equations and n is the number of independent
variables where n > m.
This is a very standard technique used in all nonlinear programming and is used
routinely as a mathematical tool in many standard commercially available software
like MS excel, MATLAB etc. having varied applications in engineering, science and
economics modeling.
Use of the above will automatically revise the value of Ep /G and upgrade the values
of I2 and I1 (dimensionless but a function of Ep /G), which may then be used to calculate
the revised and exact stiffness and mass contribution of the pile which would closely
simulate the field condition.
The steps are furnished in detail in Figure 2.5.27 as to how the data are updated
and corrected for the example cited in example mentioned below.
Having established the mass and stiffness coefficients of the pile correctly based on
field data the damping may now be established as

Cx = r0 ρGSx2 LI2 (2.5.189)
where I2 is the corrected upgraded value and Sx2 is as obtained from Table 2.5.17.
2.5.11.8.1 Comparison of results

A comparison of results against established methods to ensure that the method is not
an utopian exercise with differential equations and it does have applications.

6.00E+05
5.00E+05
Stiffness (kN/m)
4.00E+05 Kxx
3.00E+05 Novak
2.00E+05 Gazetas
1.00E+05
0.00E+00
0
0
00
00
00
0
00
50
25
10
25
50
10
Ep/Gs
Figure 2.5.25 Comparison of stiffness values for, r = 0.3 m and length = 30 m.
1.20E+06
Stiffness (kN/m)
1.00E+06
8.00E+05 Kxx
6.00E+05 Novak
4.00E+05
Gazetas
2.00E+05
0.00E+00
00
00
0
00
0
00
25
50
25
50
10
10
Ep/Gs
Figure 2.5.26 Comparison of stiffness values for, r = 0.6 m and length = 30 m.
For this two RCC piles of radius 0.3 m, 0.6 m of length 30 m has been has been
checked with the reported results for comparison. The values Kxx [Equation (2.5.106)]
is shown in Figures 2.5.25 and 2.5.26 for comparison.
Next, the results of uncoupled horizontal frequency of a real time compressor foun-
dation weighing 400 kN supported on 9 RCC piles of length 36 m and diameter 1.8 m.
The pile cap size is 7 m × 5 m × 2 m. The piles are spaced at distance of 3.0 m.
The natural frequencies of the foundation are compared for Ep /G value varying from
250–10,000. Weight of the compressor is 400 kN.
Table 2.5.19 clearly shows that the values are in very good agreement for the base
case and thus can well be used for other cases as mentioned above for which there are
no direct solutions.
Finally, the stiffness of a short pile has been computed. This is based on the field
observed data having the following properties:
Length of pile = 10 m, diameter of pile = 1.2 m. Material of pile = RCC.
• method of installation-bored pile.

• based on soil test, observed Ep /G = 5000.
• Ep considered = 3 × 107 kN/m2 .
• unit weight of pile material = 25 kN/m3 .
• field observed natural frequency of the pile is = 58 rad/sec (9 Hz).
• Poisson’s ratio of soil considered = 0.4.

Table 2.5.19 Comparison of frequency for a compressor foundation proposed versus Novak and
Gazzetas.
Frequency
Frequency (rad/sec) Frequency (rad/sec) (rad/sec) with
Ep/G with Kproposed with KNovak KGazzetas
250 252.64 251.44 252.57

500 192.10 194.85 192.07
1000 146.14 150.76 146.07
2500 101.79 107.21 101.71
5000 77.30 94.66∗∗ 77.35
10000 58.87 63.98 58.82
∗∗ The stiffness value was linearly interpolated from Novak (1983) Table for Ep/G = 500.
For the above conditions:

Selected value of Sx1 from Table 6 = 1.09
Ep /Gs = 5000 (given),
β = 2.1512 Equation (2.5.102)
A = 0.50135; B = 0.02705 Equation (2.5.179)
I1 = 0.23802, I2 = 0.9035 Equation
(2.5.185)

Computed natural frequency Kp /Mp = 68.26 rad/sec (11 Hz)
➔ Error (ε) = 10.26
Setting the error (ε) = 0 and running the solver function in a spread sheet for
changing Ep /Gs for boundary constraint Sx1 > 0, the following upgraded data have
been obtained:
Sx1 = 1.09; Ep /Gs = 7246; β =1.96064; A = 0.65984; B = −0.04832; I1 =
0.27266 and I2 = 0.949504.
Computed natural frequency based on above data = 58 rad/sec (9 Hz).
Revised error (ε) = −2.79 × 10−7 .
Thus based on the above data as per Equation (2.5.184), the correct stiffness of the
pile is deduced as Kpile = 9.206 × 104 kN/m.
It is to be noted here that the Ep /G value has increased from 5000 to 7246 meaning
thereby that the soil had lost some of its initial strength due to boring of the pile-which
is quite logical.
2.5.11.8.2 Computer run steps for short pile based on f ield observed data
The following section shows the computer run for evaluation of the stiffness of the
pile in lateral direction in three steps.
1 Stiffness and frequency calculation of pile based on theoretical data and calculating
the error based on field observed data.
2 The data screen just prior to run of the solver with command to change Ep /G
value keeping the Sx value > 0.
3 Final value of the stiffness and frequency of the pile after solver has optimized the
data.

Step-1: Shows the initial calculation of frequency and stiffness of pile including the error with
respect to field observed frequency.
Figure 2.5.27 Steps of computation.
Steps of calculations are given in Figure 2.5.27.

A comprehensive analytical solution for dynamic analysis of long piles has been
presented and is in good agreement with the existing solution. Based on this, piles
with boundary conditions like partial embedment and soils with varying G can also
be analyzed.
Considering the fact that the dynamic bending moment and shear force can also be
obtained by this method, the standard practice of restricting the pile capacity to 50%
of its capacity will not be necessary. It will be observed from Equations (2.5.153) and
(2.5.156) that the moment and shear takes care of the dynamic magnification factor
of the load at the same time gives a complete distribution of its magnitudes along the
depth of the pile. This when combined with static load would give the design moment
for the pile. Considering that there is no uncertainty with this formulation, one can
perhaps restricts the pile load limit to 80% of its capacity in lieu of 50% as in vogue
presently and this would bring significant economy in design and for large project
savings could quite significant.
Short piles, for which no established method exists, also can be solved by the present
method.

Step-2: Showing solver on the verge of optimizing by changing Ep/G value by setting the error
to zero.
Figure 2.5.27 (continued).
2.5.11.9 Dynamic analysis of piles under rocking or rotational mode

We present herein the mathematical model for rocking or rotational mode. This mode
generally comes coupled with translational mode.
Shown in Figure 2.5.21 is a pile embedded in ground considered in a homogeneous
elastic medium under plane strain condition. The pile is considered to be long and
slender.
Under static loading, the equation of equilibrium in the x-direction for such beam
on elastic foundation is given by Timoshenko (1956)
d4x
E p Ip = −ks x (2.5.190)
dz4
where Ep = Young’s modulus of the pile; Ip = moment of inertia of the pile cross
section; ks = elastic stiffness of the soil and is expressed as GSθ1 ; Gs = dynamic shear
modulus of the soil; Sθ1 = Berdugo’s rotational constant which are basically frequency
dependent,

Step-3: Final value of stiffness of piles after the solver has optimized the error.
Figure 2.5.27 (continued).
The general solution of Equation (2.5.190) is given by
x = e−qz (C0 cos qz + C1 sin qz) + eqz (C2 cos qz + C3 sin qz) (2.5.191)

GSθ1
where q = 4 (2.5.192)
Ep Ip
Similar to lateral load case the deflection equation can be considered as
x = e−qz (C0 cos qz + C1 sin qz) (2.5.193)
Considering the pile head undergoing specified deflection and rotation as well as
it’s head is fixed to the pile cap (same boundary condition as considered by Novak
(1974)), we have
At z = 0, let x = x0 ⇒ C0 = x0 , which gives
x = e−qz (x0 cos qz + C1 sin qz) (2.5.194)

dx
Again considering at z = 0, = θ0 , we have
dz
θ0
C1 = x0 + (2.5.195)
q
Thus Equation (2.5.194) can now be represented as

−qz θ0
x=e x0 cos qz + x0 + sin qz (2.5.196)
q
Dividing each of the above term by L we have

x x0 x0 θ0
= e−qz cos qz + + sin qz (2.5.197)
L L L qL
x0 x
For magnitude of rotation being small θ0 ∼
= and θz ∼
= when we have
L L

−qz 1
θz = θ0 e cos qz + 1 + sin qz (2.5.197a)
qL
Now considering β = qL and looking at Equation (2.5.197) we can say that

for any arbitrary loading, the generic shape function in dimensionless form can be
represented as

−βz βz 1 βz
ϕ(z) = e L cos + 1+ sin (2.5.198)
L β L
where

GSθ 1 L4
β= 4
; L = Length of the pile. (2.5.199)
Ep Ip
Equation (2.5.198) can thus be written as

−βz βz βz
ϕ(z) = e L cos + η sin (2.5.200)
L L
Thus it is observed that shape function for rotational mode remains invariant with
respect to the lateral motion of pile for the given boundary condition.

Generic Shape Function of Pile in Rotational Mode

1.5
Shape function
1
0.5 F(z)
9
15
45
75
0
0.
0.
0.
0.
0.
0.
-0.5
z/L
Figure 2.5.28 Generic shape function of pile for Ep /G = 5000.
Differentiating Equation (2.5.200) we have

β −βz βz βz
ϕ (z) = e L (η − 1) cos − (1 + η) sin (2.5.201)
L L L
1
when, η = 1 + (2.5.202)
β
The generic shape function of the pile in fundamental mode as per Equation
(2.5.200) is as shown in Figure 2.5.28 for Ep /G = 5000.
The potential energy d of an element of depth dz, shown in Figure 2.5.25, under
rotational mode is then given by (Craig 1981)
2
Ep Ip dθ Kθ 2
d = + θ (2.5.203)
2 dz 2
where, Ep = Young’s modulus of pile; Ip = moment of inertia of pile; Kθ = rotational

stiffness of soil having dimension kN/m; θ = rotational displacement of pile in x
direction and may be written as (z)q(t).
For a rigid circular embedded footing with embedment Df , the stiffness of the footing
in rotational mode may be expressed as

Gs Df D2f
Kθ = Gb r30 Cθ1 + Sθ1 + Sx1 (2.5.204)
G b r0 3r20
where, Kθ = foundation stiffness in horizontal direction; Gs = dynamic shear mod-

ulus of the soil along foundation surface; Gb = dynamic shear modulus of the soil
at foundation base; r0 = Radius of the foundation; Cθ1 and Sθ1 Sx1 = Beredugo’s
Constant which are basically frequency dependent.
Ignoring the first term within bracket in Equation (2.5.204) which represents the
contribution of base resistance, and substituting the same in Equations (2.5.203) for a

cylindrical element of depth dz, embedded in soil, and also ignoring the term containing
dz2 which is exceedingly small the potential energy d may be written as
2
E p Ip dθ Gr20 Sθ1 dz 2
d = + θ (2.5.205)
2 dz 2
the total potential energy over the whole length of the pile (L) is then given by
L 2 L
Ep Ip dθ Gr20 Sθ1
= dz + θ 2 dz (2.5.206)
2 dz 2
0 0
Considering v(z, t) = (z)q(t), it can be proved that
L L
Kij = Ep Ip ϕi (z)ϕj (z)dz + Gr20 Sθ1 ϕi (z)ϕj (z)dz (2.5.207)
0 0

Thus for fundamental mode the rotational stiffness of the pile is then given by
L L
2
Kθ = Ep Ip ϕ (z) dz + Gr20 Sθ1 ϕ(z)2 dz (2.5.208)
0 0
where, (z) is as expressed in Equation (2.5.201).

Substituting the value of φ(z) and φ (z) in Equation 2.5.208, and carrying out
integration by parts and some simplification, we finally get the rotational stiffness as
⎡ ⎤
−2β ) + Y 1 + ψ ψ
Ep Ip X (1 + ψ) (1 − e 2 4 −η 1− 2
Kθ = ⎣ ⎦ (2.5.209)
L 2 (η − 1)
2
where ψ = 4Gλ Sθ 1
π Ep β 2
and λ = L/r0 the slenderness ratio of the pile. It is to be noted that
ψ is a dimensionless quantity, X, Y, η etc. are same as derived for lateral stiffness case.
The accuracy of Equation (2.5.209) will be dependent on the correct selection of Sθ1 .
For instance for rigid circular footing Novak and Beredugo (1972) has furnished
a frequency independent value of Sθ1 = 2.5 (for any value Poisson’s ratio) which has
been found to give adequate accuracy for practical engineering design.
Comparing the stiffness data with Novak (1974) and Gazetas (1988) data it is pro-
posed that the following values [Tables 2.5.20 to 22] of Sθ1 be used for the calculation
of dynamic response of pile under rocking mode.

Table 2.5.20 Suggested value of Sθ 1 for Poisson’s ratio of soil = 0.25.
L/r0
Poisson’s (slenderness
ratio ratio) Sθ1 (250) Sθ1 (500) Sθ1 (1000) Sθ1 (2500) Sθ1 (5000) Sθ1 (10000)
0.25 25 16.968 23.089 30.776 43.412 54.647 66.877

40 17.358 23.656 31.586 44.678 56.390 69.253
60 17.567 23.961 32.016 45.333 57.272 70.418
80 17.674 24.110 32.225 45.648 57.688 70.958
100 17.736 24.199 32.348 45.833 57.930 71.267
Note: The value in Parenthesis after Sθ1 depicts the value of Ep /Gs value of the soil.
L/r0
ratio ratio) Sθ1 (250) Sθ1 (500) Sθ1 (1000) Sθ1 (2500) Sθ1 (5000) Sθ1 (10000)
0.40 25 18.037 24.623 32.937 46.707 59.054 72.614

40 18.448 25.221 33.794 48.05 60.909 75.145
60 18.671 25.543 34.249 48.748 61.851 76.393
80 18.781 25.702 34.471 49.084 62.298 76.974
100 18.847 25.795 34.603 49.281 62.557 77.307
L/r0
ratio ratio) Sθ 1 (250) Sθ 1 (500) Sθ 1 (1000) Sθ1 (2500) Sθ1 (5000) Sθ1 (10000)
0.50 25 18.717 25.599 34.316 48.817 61.888 76.316

40 19.141 26.217 35.202 50.21 63.813 78.946
60 19.37 26.55 35.674 50.936 64.794 80.247
80 19.484 26.714 35.905 51.285 65.259 80.853
100 19.552 26.811 36.041 51.49 65.531 81.203
For a particular pile having specific slenderness ratio and Poisson’s ratio of the
soil we select the value of Sθ1 from the above table and on substitution of the same
in Equation (2.5.199) and Equation (2.5.209) gives the solution of pile stiffness in
rocking mode.
2.5.11.9.1 Estimation of mass contribution of pile

The mass matrix of the pile may be expressed as

Mx = m(z) φi (z)φj (z)dz (2.5.210)

For the present case of pile of length L, mass moment of inertia Jx is represented by
L
Mx r20
Jx = dz + z2 dz (2.5.211)
L 4
0
Substituting Equation (2.5.210), we may now write
L L 2
γp Ap r20 2 γ p A p L2 z
Jx = ϕ(z) dz + ϕ(z)2 dz (2.5.212)
4g g L
0 0
where γp = weight density of the pile material; Ap = cross sectional area of pile;
g = acceleration due to gravity.
Equation (2.5.212) on integration by parts and simplification finally gives

γp Ap r20 L Y
Jx = XF(λ) + + η (2.5.213)
16βg 2

−2β λ2 2 −2β 1 2
where F(λ) 1−e + 2 − 4λ e 2+ − 2
β β β
and λ = L/r0 the slenderness ratio of the pile.

Equation (2.5.213) gives the inertial contribution of pile in the fundamental mode.
Incidentally the effect of this is usually ignored in design but could have significant
effect if the number of piles is large in a pile group.
2.5.11.9.2 Radiation damping factor for pile under rocking mode

For a rigid footing embedded in soil for a depth Df , Novak and Beredugo (1972) has
proposed an expression

G s Df
D2f
Cθ = r40 ρG Cθ2 + Sθ2 + 2 Sx2 (2.5.214)
G r0 3r0
where, r0 = radius of the foundation; G = dynamic shear modulus at foundation

base; Gs = dynamic shear modulus of soil in which the foundation is embedded; Df =
depth of embedment; Cθ2 , Sθ2 and Sx2 = frequency independent constants as defined
by Novak and Beredugo (1972).
Ignoring the first term in Equation (2.5.214) which represents the contribution of
base damping for a cylindrical element of depth dz, embedded in soil, and ignoring

the term, containing dz2 which is again exceedingly small, we have

c(θ ) = r30 ρGs Sθ2 dz (2.5.215)
For systems having continuous function, the damping is usually expressed as

Cθ = c(θ ) φi (z)φj (z)dz (2.5.216)
For the present case of pile of length L, Equation (2.5.216) can be expressed as
L
Cθ = r30 ρGs Sθ2 φ(z)2 dz (2.5.217)
0
Equation (2.5.217) on integration by parts and simplification, we have

X(1 − e−2β ) + Y
+η
Cθ = r30 ρGSθ2 L 2
(2.5.218)
4β
Equation (2.5.218) expresses the soil damping for a single pile under horizontal
mode of vibration. Here the Factor Sθ2 is damping coefficient which is frequency
dependent. Fortunately the damping factor is required for calculation of the amplitude
when the eigen solution of the problem is already done vis a vis, the dimensionless fre-
quency number a0 = ωr0 /vs term is known. Polynomial fit curve for Sθ2 are available
in terms of a0 which can be used directly to arrive at these parameters.
The value of Sθ2 is as given hereafter as per Novak and Beredugo (1972)
Sθ 2 = 0.0144a0 + 5.263a20 − 4.177a30 + 1.643a40 − 0.2542a50 (2.5.218a)
This value unlike other Beredugo’s constant is independent of Poisson’s ratio.
2.5.11.9.3 Consideration of material damping of pile

The structural stiffness contribution of the pile is given in the first part of Equation
(2.5.208), while that of the mass moment of inertia is given in Equation (2.5.212).

Thus, if Cc is the critical damping of the pile then it can be expressed as Cc = 2 KJx ,
where K and Jx are the stiffness and mass moment of inertia of the pile.
Depending on the material used for pile like (RCC, steel etc.) a suitable damping
ratio (D) can be assumed. The damping (Cp ) for the pile can be expressed as
Cp = D Cc (2.5.219)
This, when added to the radiation damping, calculated in Equation (2.5.218) gives
the complete damping quantity for the soil-pile system.

2.5.11.9.4 Piles with other boundary conditions

Having established the stiffness, inertial and damping contribution of the pile in rock-
ing mode based on minimization of the potential energy of the system we extend the
above method for piles with other boundary conditions for which there are no standard
solutions.
2.5.12 Partially embedded piles under rocking mode

As stated earlier, this is a very common practice in Arctic and North Siberian condition,
where due to environmental reasons; the steel piles are driven into the ground when
they protrude about 2–3 m above the ground over which the pile cap and vibrating
equipments are placed. In such cases Novak’s (1974, 1983) chart cannot be used, nor
is Gazetas’ formulation valid.
We provide the solution of the same as hereafter.
Let L be the full length of the pile and let the length of the embedment is soil be L1 .
For this case we have

GSθ 1 L41
βe =
4
(2.5.220)
E p Ip
Here subscript “e” represents embedment of the pile.

The shape function can thus be represented by

− βLe z βe z βe z
φ(z) = e 1 cos + η sin (2.5.221)
L1 L1
The stiffness function can thus be represented as

βe −βL e z βe z βe z
ϕ (z) = e 1 (ηe − 1) cos − (1 + ηe ) sin (2.5.222)
L1 L1 L1
Square of the above is given by

βe2 2βe z Xe 2βe z 2βe z
φ (z)2 = 2
e− L − 2ηe cos + Ye sin (2.5.223)
L1 2 L1 L1
Here X = 1 + ηe2 ; Y = 1 − ηe2 and ηe = 1 + β1e .

Now considering the fact that embedment of a beam does not have any effect on
the shape function of the system, the stiffness of the pile is expressed as
L L1

Kθ = Ep Ip φi (z) dz + GSθ 1 [φi (z)]2 dz
2
(2.5.224)
0 0

Equation (2.5.224) on integration by parts and simplification may be expressed as

Ep Ip βe
Kθ = Xe (α + ψ) − Xe {αe−2βe α + ψe−2βe }
2L1

α ψ ψ
+ Ye + − ηe α − (2.5.225)
2 4 2
which can further be expressed as (here α = L/L1 )

!
−2βe α − ψe−2βe ) + Y α ψ ψ
Ep Ip Xe (α + ψ − αe e 2 + 4 − ηe α − 2
Kθ =
L1 2 (ηe − 1)
(2.5.226)
Equation (2.5.226) gives the solution for stiffness of partially embedded piles in the
ground. The correctness of the equation can be back checked by the fact that when the
pile becomes fully embedded i.e. L1 = L we have α → 1, βe = β, Xe = X etc. when
Equation (2.5.226) degenerates to Equation (2.5.209), the stiffness for fully embedded
pile.
Proceeding in identical manner as done before, the mass and damping terms can be
obtained as given earlier.
The mass moment of inertia of pile remains same as stated in Equation (2.5.213).
The damping matrix is given by the expression

Xe (1 − e−2βe ) + Y2e + ηe
Cθ = r30 ρGSθ2 L1 (2.5.227)
4/(ηe − 1)
2.5.12.1 Stiffness of the pile for soils with varying elastic property
Considering the variation of shear modulus with depth as
G = G(z/L)m (2.5.228)
where m = a number varying from 0–2 [considered 0 when G is constant with depth,
assumed 1 for linear variation and 2 for parabolic distribution] we derive the pile
stiffness and other parameters as hereafter.
Thus for linearly varying soil the stiffness matrix can be written as
L
Ep Ip β 2 2βz 2βz 2βz
Kθ = e− L X − 2η cos − Y sin dz
L2 L L
0
L
+ Gr20 Sθ 1 e L + cos + η sin dz (2.5.229)
L 2 2 L L
0

Equation (2.5.229) can be further simplified to

Ep Ip β ψ ψ
Kθ = X 1+ − e−2β 1 + (1 + β)
2L 2β 2β

3ψ 1 ψ
+Y + −η 1− (2.5.230)
8β 2 4β
The damping matrix for this case can thus be represented by
√
r30 ρGSθ2 L −2β 3Y η
Cx = X[1 − e (1 + β)] + + (2.5.231)
4β 2 4 2
The mass coefficient remains same as expressed in Equation (2.5.213).

When the dynamic shear modulus variation is parabolic with depth the stiffness
equation of the pile is expressed as
L
Ep Ip β 2 − 2βz 2βz 2βz
Kθ = e L X − 2η cos + Y sin dz
L2 L L
0
L 2
+ Gr20 Sθ 1 e L + cos + η sin dz (2.5.232)
L 2 2 L L
0
which can be further simplified and expressed as

Ep Ip β ψ −2β ψ 1 2 Y
Kθ = X 1+ − e 1 + 2 + − + − η
2L 8β 2 2 β β2 2
(2.5.233)
Equation (2.5.233) gives the stiffness expression of pile under parabolic variation
of G along the length of pile.
Proceeding in same manner as stated above the damping matrix is expressed as
√
r30 ρGSθ2 L 1 −2β 1 2
Cθ = X −e 2+ − 2 (2.5.234)
4β 4β 2 β β
The mass coefficient remains same as expressed in Equation (2.5.213).
2.5.12.1.1 Calculation of dynamic bending moment and shear force in pile

Neither Novak nor Gazetas’ method can be used for this purpose. For machine foun-
dation subjected to a dynamic moment of M0 sin ωm t, the amplitude of vibration is

given by
(M0 /Kθ ) sin ωm t

θ (t) = (2.5.235)
(1 − r2 )2 + (2Dr)2
where, ωm = operating frequency of the machine; M0 = unbalanced dynamic moment;

ρ = ωm /ωn, the ratio of operating and natural frequency; D = damping ratio of the
system.
Thus the peak amplitude is given by
(M0 /Kθ )
θ (t) = (2.5.236)
(1 − r2 )2 + (2Dr)2
The complete displacement function is then given by

(M0 /Kθ ) βz βz βz
θ (z, t) = e− L cos + η sin (2.5.237)
(1 − r2 )2 + (2Dr)2 L L
Thus bending moment is given by
Ep Ip θ = −M(x)

(Ep Ip M0 )/Kθ
β − βz βz βz
or, M(x) = e L (1 + η) sin − (η − 1) cos
(1 − r2 )2 + (2Dr)2 L L L
(2.5.238)
The dynamic shear force is given by

(Ep Ip M0 )/Kθ
Ep Ip θ = −V(z) = − e L sin − η cos
(1 − r2 )2 + (2Dr)2 L2 L L
(2.5.239)

(Ep Ip M0 )/Kθ
➔ V(z) = e L sin − η cos . (2.5.240)
(1 − r2 )2 + (2Dr)2 L2 L L
2.5.12.2 Dynamic response of short piles under rotational mode

As mentioned earlier no solution exists till date for this type of piles. Bojtsov (1982)
has given solution to the generic displacement curvature of such short beams on elastic
foundation which is given by
x = C0 cos h pz cos pz + C1 cos h pz sin pz + C2 sin h pz sin pz + C3 sin h pz cos pz

(2.5.241)
where p = q is same as expressed in Equation (2.5.192).

Expressing the above in terms of Puzrevsky function (Karnovsky 2001) Equation

(2.5.257) can be expressed as
x = C0 V0 (pz) + C1 V1 (pz) + C2 V2 (pz) + C3 V3 (pz) (2.5.242)
where,
V0 (pz) = cos h pz cos pz (2.5.243)

1
V1 (pz) = √ (cos h pz sin pz + sin h pz cos pz) (2.5.244)
2
V2 (pz) = sin h pz sin pz (2.5.245)
1
V3 (pz) = √ (cos h pz sin pz − sin h pz cos pz) (2.5.246)
2
For analysis similar to previous case we assume the pile as fixed at base and is fixed
also at pile cap level and can undergo deflection and rotation at pile head. Considering
base of pile as z = 0 and applying the Puzrevsky’s functional properties as elaborated
in case of piles under lateral load we have
At z = 0, x = 0 ⇒ C0 = 0
At z = 0, x = 0 ⇒ C1 = 0 which gives
x = C2 V2 (pz) + C3 V3 (pz) (2.5.247)
At the pile head we have at z = L x = 1 which gives
C2 V2 ( pL) + C3 V3 ( pL) = 1 (2.5.248)
Again at z = L x = 1/L which gives
C2 V2 (pL) + C3 V3 (pL) = 1/L (2.5.249)
Using the derivative properties as shown above we have
1
C2 V1 (pL) + C3 V2 (pL) = √ (2.5.250)
pL 2
Expressing the above in matrix form we have
[V] {C} = {p} (2.5.251)
which can be further expressed as
{C} = [V]−1 {p} (2.5.252)

Performing the above operation gives
⎧ ⎫

1 V2 (pL) −V3 (pL) ⎨
1 ⎬
C2 1
= (2.5.253)
C3 −V1 (pL) V2 (pL) ⎩ √ ⎭
pL 2
where = V22 (pL) − V1 (pL)V3 (pL) which gives

1 V3 (pL) 1 V2 (pL)
C2 = V2 (pL) − √ and C3 = √ − V1 (pL) (2.5.254)
pL 2 pL 2
Thus the displacement for the given boundary condition is then expressed as

1 V3 (pL) 1 V2 (pL)
x= V2 (pL) − √ V2 (pz) + √ − V1 (pL) V3 (pz) (2.5.255)
pL 2 pL 2
Considering the fact that for long piles the shape function remains invariant for
rocking mode with respect to lateral motion, for same boundary condition it may
be concluded that for short piles also the same condition would hold good thus the
generic shape function in dimensionless form in rocking mode is given by

1 V3 (β) βz 1 V2 (β) βz
φ(z) = V2 (β) − √ V2 + √ − V1 (β) V3
β 2 L β 2 L
(2.5.256)
where the determinant gets modified to = V22 (β) − V1 (β)V3 (β).

Considering A = C2 / and B = C3 / the shape can now be expressed as

βz βz
φ(z) = AV2 + BV3 (2.5.257)
L L
Typical generic shape function for the short piles Ep /Gs = 2500 is as shown in
Figure 2.5.29.
Differentiation of above and using the differential properties as mentioned earlier
we have
√
β 2 βz βz
φ (z) = AV1 + BV2 (2.5.258)
L L L

0.2
9
0
1
0.
0.
0.
0.
0.
0.
0.
0.
0.
Shape Function
-0.2
-0.4
-0.6
-0.8
-1
z/L
Figure 2.5.29 Generic shape function of short pile for Ep /G = 2500.
Substituting the above functions, we have
L 2
2Ep Ip β 2 βz βz
K= AV1 + BV2
L2 L L
0
L 2
βz βz
+ Gr20 Sθ1 AV2 + BV3 (2.5.259)
L L
0
The above is too complicated to solve in closed form as such numerical integration
may be used to arrive at the stiffness value.
Considering ξ = Lz we have L · dξ = dz and as z → L; ξ → 1; as z → 0 ξ → 0;
which gives
1
2Ep Ip β 2
K= [AV1 (βξ ) + BV2 (βξ )]2 Ldξ
L2
0
L
+ Gr20 Sθ1 [AV2 (βξ ) + BV3 (βξ )]2 Ldξ (2.5.260)
0
Substituting the value of (from Equation 2.5.209) in Equation (2.5.260), we have

⎡ ⎤
1 1
2
K = Gr20 Sθ 1 L ⎣ [AV1 (βξ ) + BV2 (βξ )]2 dξ + [AV2 (βξ ) + BV3 (βξ )]2 dξ ⎦
ψ
0 0
(2.5.261)

2
K = Gr20 Sθ 1 L I1 + I 2 (2.5.262)
ψ

Table 2.5.23 Suggested for Sθ 1 for short piles (L/r ≤ 20) for field data iteration.
Ep /Gs Sθ 1 (ν = 0.25) Sθ1 (ν = 0.4) Sθ1 (ν = 0.5)
250 15.563 16.561 17.197

500 21.046 22.468 23.372
1000 27.873 29.860 31.135
2500 39.05 42.041 43.976
5000 49.07 53.014 55.576
10000 60.187 65.311 68.598
Here
1 1
I1 = [AV1 (βξ ) + BV2 (βξ )]2 dξ and I2 = [AV2 (βξ ) + BV3 (βξ )]2 dξ
0 0
(2.5.263)
The integrals I1 and I2 can very easily be solved by using Simpson’s 1/3rd rule
between limits 0–1 and can be back substituted in Equation (2.5.261) to arrive at the
stiffness for the short pile.
As there is no theoretical or experimental benchmarking against which the stiffness
values can be checked or compared. So use of this expression must always be backed
up by dynamic field test of the piles to adjust the data (especially Sθ1 or Ep /G) to match
the field observed value.
In the absence of comparative benchmarks we may start the design with the fol-
lowing suggestive values of Sθ1 for various Ep /Gs values given in Table 2.5.23. These
values as mentioned above, is based on formulation for long pile (with L/r < 25) but
may be used as a starting point for the iteration based on field observed data.
The mass moment of inertia of the pile for fundamental mode is given by
L L
γp Ap r20 2 γ p A p L 2 z 2
Jx = ϕ(z) dz + ϕ(z)2 dz (2.5.264)
4g g L
0 0
γp Ap r20 L γ p Ap L3
→ Jx = I1 + I3 (2.5.265)
4g g
1
Here I3 = ξ 2 [AV2 (ξ ) + BV3 (ξ )]dξ (2.5.266)
0
Mp r20
or Jx = I 1 + M p L 2 I3 (2.5.267)
4
γp Ap L Mp r20
where, Mp = ➔ Jx = [I1 + 4λ2 I2 ] (2.5.268)
g 4

To start the design we select a value of Sθ1 for a specified Ep /Gs from Table 2.5.23
and find out the value of the frequency based on Equation (2.5.262) and (2.5.267).
Let this be defined as ωc where the subscript c stands for the word “computed”.
Let the field tested natural frequency of the pile be ωf where ωf = ωc .
Based on the above argument the error(ε) in the analysis is then given by
ε = ωc − ωf
For ε → 0 we have, ωc = ωf or ωc2 = ωf2
2

K 4GSθ1 L ψ I1 + I2
Considering ωc2 = , we have − ωf2 = 0 (2.5.269)
Mx Mp I1 + 4λ2 I2
It will be observed that all the factors β, I1 , I2 in Equation (2.5.269) is a function

of Ep /Gs . The difference (which is the error ε) can now be set to zero or minimum by
varying the value of Ep /Gs for which, lim ε →0.
This can very easily be done by using the standard solver or goal seek in a spread
sheet with boundary constraint that Sθ1 > 0.
The above will automatically revise the value of E/Gs and upgrade the values of I3 ,
I2 and I1 (which are dimensionless functions), which may then be used to calculate
the revised and exact stiffness and mass contribution of the pile which would closely
simulate the field condition.
Having established the mass and stiffness coefficients of the pile correctly based on
field data the damping may now be established as

Cθ = r30 ρGSθ2 LI2 (2.5.270)
2.5.13 Group effect of pile

Refer Section 2.5.7 where this has been dealt in detail and may well be used for this
case too.
2.5.13.1 Effect of pile cap on pile stiffness

The sketch given in Figure 2.5.30 represent the pile group with pile cap. In such case
usually the embedment stiffness GSf Df is added to the pile group stiffness and the
system is considered as a lumped mass single degree freedom system, the details of
which are furnished in Novak (1974) and Prakash and Puri (1988).
In conventional formulation as the stiffness matrix is statically coupled another set
of stiffness Kxθ needs to be derived in addition to what has been derived above. To
circumvent this issue we propose to use the following model as shown in Figure 2.5.31.

To derive the equations we use the Lagrange’s equation from the energy principle
as derived earlier when we finally get the stiffness and mass matrix as
⎡ ⎤⎧ ⎫ ⎡ ⎤⎧ ⎫
Mf Mf M f Zc ⎨ẍ⎬ Kf 0 0 ⎨x⎬
⎣ Mf Mf + M x Mf Zc ⎦ ü + ⎣ 0 Kx 0⎦ u = 0 (2.5.271)
⎩⎭ ⎩⎭
Mf Zc Mf Zc Jx + Mf Zc2 θ̈ 0 0 Kθ θ
The above gives the complete free vibration equation of motion for pile plus pile
cap with machine considering pile springs in translation and rocking mode.
Considering the equation to be dynamically coupled the damping matrix can now
be expressed as
⎡ ⎤
Cf 0 0
⎣
[C] = 0 Cx 0⎦ (2.5.272)
0 0 Cθ
Zc Df
Figure 2.5.30 Schematic diagram of pile and pile-cap with embedment.
M = Mass of (Pile cap + Machine)
Kf = Embedded stiffness of soil @ GSfDf

Zc
mp = Mass of pile group
K Kx
u
Jx = Moment of inertia
of Pile group
Figure 2.5.31 Mathematical model of pile group and pile cap under coupled sliding and rocking mode.

where, Mf = mass of pile cap plus mass of machine; Mx = mass of pile group; Jx =
mass moment of inertia of piles; Zc = center of mass of foundation plus machine
along vertical axes; Kf = lateral embedded stiffness of pile cap @ G Sfx Df ; G =
dynamic shear modulus of soil; Sfx = Berdugo’s constant @ 3.6, 4, 4.1 for ν = 0.0,
0.25, 0.4 respectively; Df = depth of embedment; Kx = Group lateral stiffness of pile
group based on Equation (2.5.118) where Kθ = rotational stiffness of pile group; U =
potential energy of the system, and T = kinetic energy of the system.
It is to be noted that for pile group for calculation of mass and mass moment of
inertia the mass and inertia of single pile has to be multiplied by the number of piles
in the group. While for stiffness and damping the group stiffness and damping has to
be derived according to Equation in section 2.5.7.
2.5.14 Comparison of results

The method proposed herein is now compared withy Novak and Gazetas’ values to
check their accuracy.
For this two RCC piles of radius 0.4 m, 1.0 m of length 40 m has been has been
checked with the reported results for comparison. The values Kθ [Equation (2.5.209)]
is shown in Figures 2.5.32 and 33 for comparison.
The results clearly shows that the values are in very good agreement for the base
case and thus can well be used for other cases as mentioned above for which there are
no direct solution.
We finally calculate the stiffness of a short pile based on field observed data having
the following properties.
Length of pile = 10 m, Diameter of pile = 1.2 meter. Material of pile RCC.
• Method of installation-bored pile.

• Based on soil test, observed Ep /G =5000.
• Ep considered @ 3 × 107 kN/m2 .
• Density of pile material = 25 kN/m3 .
• Field observed natural frequency of the pile is = 28 rad/sec (4 Hz).
• Poisson’s ratio of soil considered = 0.4.
For the above conditions
Selected value of Sθ1 from Table 2.5.23 = 53.014.

Ep /Gs = 5000 (given); β = 5.681 : vide Equation (2.5.198); A = −0.000912 : vide
Equation (2.5.271); B = −0.003447 do; I1 = 0.0277902, I2 = 0.201259,I3 =
0.16886 : vide Equation (2.5.263).
Computed Stiffness = 7.78 × 10 5
kN/m.
Computed natural frequency ( Kθ /Jx ) = 39.98 rad/sec (6 Hz).
➔ Error (ε) = 11.98
Setting the error (ε) = 0 and running the goal seek function in a spread sheet for
changing Ep /Gs for boundary constraint Sθ1 > 0, we have the following upgraded
data:

1.00E+06
Stiffness (kN/m)
8.00E+05
6.00E+05 Kxx
4.00E+05 Novak
Gazetas
2.00E+05
0.00E+00
00
00
00
0
25
50
00
10
25
50
10
Ep/Gs
Figure 2.5.32 Comparison of stiffness values for r = 0.4 m and length = 40 m.
Comparison of Rocking Stiffness for piles

Stiffness (kN/m)
2.00E+06
1.50E+06
Kxx
1.00E+06
5.00E+05 Novak
0.00E+00 Gazetas
00
00
00
0
0
00
25
50
25
10
50
10
Ep/Gs
Figure 2.5.33 Comparison of stiffness values for r = 1.0 m and length = 40 m.
Sx1 = 53.014; Ep /Gs = 3354; β = 6.2777; A = −0.000889; B = −0.00528; I1 =

0.00111, I2 = 0.138856, I3 = 0.11672.
Computed natural frequency based on above data = 28 rad/sec (4 Hz).
➔ Revised error (ε) = −0.00092
Thus based on the above data as per Equation (2.5.261) the correct stiffness of the
pile is given by Kpile = 2.64 × 105 kN/m.
In case the above correction is already done for lateral pile stiffness and E/G value
has been already modified to suite the field observed data, the same can directly be
used without carrying out the above mentioned modification again.
Referring to Figures 2.5.32 and 2.5.33, it is observed that the results are in excellent
agreement with both Novak (1983) and Gazetas (1988) stiffness. Considering the base
case being in such agreement formulations for other cases like partial embedment,
varying shear modulus etc., can now be very easily adapted for which there are no
standard solutions.
The short pile case is basically a theoretical solution and needs significant field test
and lab testing to arrive at a predefined Sθ1 values which would make the method
more powerful.
However in absence of such data the present algorithm as mentioned herein could
become a very powerful tool for dynamic analysis of such piles for which no solution
is available till date and yet remains a serious practical problem.

B
X Z X
L
Y
Foundation resting on ground Foundation on piles/springs
Figure 2.5.34
2.5.15 Practical aspects of design of machine

foundations
Enough gazing at the moon and theoretical contemplation, from the hallowed domain
of academics let us now digress into the real world of a professional engineer and see
based on the above theories how he goes about to design the foundation
• What are the input data he looks for?

• What are the assumptions he considers in his process of design.
There are three aspects to be considered at the start of the design
1 Environmental and economic impact.

2 Machine data.
3 Soil data.
2.5.15.1 Environmental and economic impact

This is the first point that a designer should assess, but unfortunately the effect of
environmental impact on the machine foundation is often overlooked.
There could be a situation, where other than the vibration of the machine itself
there are external source of vibration affecting the foundation and this could be in the
form of
• Blasting in the vicinity of the foundation

• Pile driving
• Waves transmitted by other machines operating in the vicinity of the machine
foundation in question.
Our experience shows that young engineers while doing their design of machine
foundations are more focussed on the quantitative magnitude of the natural frequency
and the amplitude and often overlooks this point.

To asses the effect of environmental impact on the foundation, if need be, seek help
of a vibration specialist and try to assess what could be the cascading effect of this
secondary source of disturbance.
If it is felt that this may possibly have some effect on the foundation isolate the
foundation by providing pockets/cut outs all-round the foundation and leaving this
space void or feeling it up with suitable dampers like cork boards, felt sheets etc.
Next try to assess how important role does the machine play in the overall process
system.
In other words, “What would be the economic impact of the machine on the overall
process vis-à-vis its performance”?
For instance if a minor chemical pump stops during an engineering process the over-
all cost impact on the process could vary from a few hundreds of dollars to thousand
dollar.
While for a major generator or a compressor foundation if the performance is not
up to the acceptable standard the client could stand to loose millions of dollars in
terms of production output and man-hours lost.
If required talk to your process engineering or mechanical engineering colleagues to
asses the criticality of the machine.
More important is the machine be more conservative in your design approach.
Do not try to economise on the material. The money that could be saved by cutting
down on a few cubic meter of concrete or hundred Kilogram of reinforcement, could
be well be offset by manifolds if your company stands to pay liquidated damages due
to malfunctioning of the foundation15 .
For machine foundations economy lies more on the smooth performance of the
machine rather than any other factors.
2.5.15.2 Machine data

• Once you have assessed the above aspect, as a next step, you should have the
machine data at your disposal in the form of a General Arrangement Drawing of
the machine.
On study of the drawing see if the following check list is satisfied as a minimum
1 Do the drawing furnishes the overall dimension of the machine/skid on which it

is mounted?
2 Are the anchor bolt locations, size of the bolts (both diameter and length) and
details of how it should be anchored to the foundation furnished by the vendor?
3 Do the drawing supply the height at which the centre line of the shaft of the
machine is located from the bottom of the machine frame (which will be the top
of concrete or top of grout for you)?
4 Is it clear to you what type of machine it is i.e. if it is centrifugal or reciprocating
in nature?
15 And this we are sure will not have a very positive outcome on your annual performance
appraisal. . . . . .

5 Does the drawing supply you with the operating speed of the machine or the range
which should be cleared during the design of the machine foundation?
6 Do the foundations need to support any pipes or valves on it other than the
machine itself?
7 If so, are all the loads and locations of these valves and pipes are mentioned in
the drawing?
8 Does the drawing clearly mention the unbalanced mass, eccentricity or the
dynamic loads generated during the operation of the machine including any
specific direction?
9 Is it clear to you what would be the level of the top of concrete of the foundation?
This is very important for the top of foundation usually fixed from the process
engineering group and if there is any mismatch in the level in the field could create
problems in terms of alignment of pipe flanges or variation in the net positive
suction head (NPSH) for the pump.
10 Is the location of the equipment in terms of co-ordinates with respect to the overall
plant available with you?
11 Finally has the equipment supplier defined any performance criterion which needs
to be met in terms of amplitude, frequency etc.
The above are very vital points both from performance and contractual point of
view. For if the equipment supplier has furnished this information then it should be
strictly adhered to, for once this is complied with the supplier alone stands guarantee
for the performance of the machine.
On the contrary if this is violated, even if the equipment supplied is faulty, the
vendor can always wriggle out of the situation by saying that his specifications were
violated and as such he cannot stand guarantee for the performance of the machine16 .
If the vendor has not specified such conditions the usual de-fault is the local code
stipulation.
But do not presume this, ask him specifically to define his performance criteria and
if he is unable to do so, make it clear to him (in writing) as to what performance
criterion you are using based on which code (could be IS, DIN, BS, ASTM special
publications etc). If possible seek his written compliance that the code-norms that is
being followed by you is acceptable to him.
Remember for important machines you are fiddling with millions of dollars so play
safe. Guard yourself both technically as well as contractually.
2.5.15.3 My Machine is perfectly balanced. . . you don’t need

to worry about the dynamic force!
A standard sales talk you will hear time and again from the equipment sales
engineer.
Rookie engineers often get carried away by this and fall prey to this over sales
strategy.
16 Refer to case history 2 at the outset of this chapter and retrospect a bit.

Many equipment suppliers do not supply any unbalanced dynamic load claiming
their machines to be perfectly balanced!
This often leaves an inexperienced engineer with the option of doing only a resonance
check and leaves it at that for he has no other data as a guideline to perform any further
check.
What should be realised at this point is that it is possible perhaps to achieve
a perfect balance in the manufacturing unit under a controlled condition at the
outset.
But when such machines are performing under a much gruelling conditions of oper-
ating day in day out and often left exposed to the vagaries of nature, due to normal
wear and tear some imbalance will invariably be generated in the system which will
induce dynamic loads on the foundation.
So do not get carried away by the claims of the vendor, for you as designer alone
remain responsible for the performance of the foundation.
In absence of such data from the vendor you may use the following guidelines
(Arya et al. 1979).
2.5.15.3.1 Design eccentricities of centrifugal machines

Ecentricities of machines under varying speeds are given in the table below.
Eccentricity in double
Sl. No. Operating speed amplitude(inch)
1 750 0.014–0.032
2 1500 0.008
3 3000 0.002
Here unbalanced dynamic force for centrifugal machine is given by
2
Fdyn = meωm
where, m = mass of the rotating shaft; e = eccentricity developed in the shaft, and
ωm = operating speed of the machine.
2.5.15.3.2 For Centrifugal compressors

12,000
e(mil) = α ≤ 1.0(mil)
r.p.m.
where, α = 0.5 at installation time = 1.0 after several years of operation

r.p.m. = Operating frequency of the machine.
1 mil = 0.001 inch.

2.5.15.3.3 For electrical motors
Peak to peak
displacement
Sl. No. Motor type Speed (RPM) amplitude (inch)
1 Integral horsepower electric motor 3000–4000 0.0010

1500–2999 0.0015
1000–1499 0.0020
999 and below 0.0025
2 Large induction motor 3000 and above 0.0010
1500–2999 0.0020
1000–1499 0.0025
999 and below 0.0030
2.5.15.4 Soil data

These data are furnished in the geo-technical report and should supply you with the
following parameters17
• Ground water table prevalent at the site

• Atterberg’s limits
• Poisson’s ratio of the soil, ν
• Unit weight of the soil, γ
• Dynamic shear modulus of the soil, G
• The foundation depth and bearing capacity of the soil at which the above
parameters are valid
• All other information, regarding the static design of the foundation.
The knowledge of ground water table is essential for all block foundations and
should preferably have the bottom of foundation above the ground water table for
waves passing through water attenuates the dynamic response.
A check on the Atterberg’s limit can give a very good indication qualitatively about
the fundamental property of the soil as to how it will behave. But unfortunately very
little attention is paid to this aspect in design offices.
The various Atterberg’s limits like liquid limit, plastic limit etc not only give a clear
indication of how the soil would behave but also holds key to the fact that if the soil
is sensitive to shocks induced by vibration or not.
We do not discuss the details of Atterberg’s limit and its interpretations but make
you aware of one criterion which is quite important in context of machine foundation
design.
Generically when the natural moisture content of the soil is closer to the liquid limit
the soil is deemed soft and when the natural moisture content is close to the plastic
17 Here we assume the reader has some knowledge about the static design procedure of a foundation.

limit it is considered as stiff. However there are certain types of soils whose natural
moisture content is greater than the liquid limit. If you ever encounter such case you
should immediately be on the alert.
For such soils generally belong to the montmorillonite group and constitutes a brittle
structure. This type of soil, when disturbed by vibration, flows like a liquid. If this soil
is allowed to remain in place it can be very dangerous for the foundation which may
undergo sudden settlement without any notice.
The liquidity index values of such soils are greater than unity. If such of soils are
encountered at a level where foundation would be resting, the complete layer should
be replaced by PCC or removed and back-filled with hydraulically compacted sand
fill compacted to a Procter Density as specified by the soil consultant.
If this strata is quite deep possibilities to be investigated to provide piles (driven/
bored) to a substantial depth below this strata and ignoring the stiffness effect of this
montmorillonite clay strata while calculating the equivalent springs for the piles.
The Poisson’s ratio of the soil is usually supplied in the soil report. This is required
for calculation of the soil springs used for dynamic analysis of the foundation. In
absence of such data υ = 0.4 would suffice for most of the cases.
The weight density of soil is usually furnished in the soil report this needs to be
divided by acceleration due to gravity (g) to arrive at the mass density.
or, ρ = γ /g here, ρ = mass density of soil; γ =unit weight of the soil, and g =
acceleration due to gravity @ 9.81m/sec2 or 32.2 ft/sec2 .
The Dynamic shear modulus plays a key role in evaluation of the spring data.
Though co-relation exists for theoretical evaluation of G from other engineering
parameters of the soil18 for important foundations we still advocate that you insist on
field test to get the field observed value of G.
Try to convince the client19 , it is worth spending a few thousand dollars now rather
than to pay through your nose in terms of performance compensations and could lead
to a classic case of being penny wise and pound foolish.
Designing a foundation with improper G value will completely waste the design
effort for the said foundation.
2.5.15.5 Trial sizing of the block foundation

Based on the above input data the next step for the designer is to do a trial sizing of
the block foundation with which he starts his first check for resonance and amplitude.
The basic guideline for the same could be summarised as follows:
• The rigid type block foundation should be so proportioned that it should have
following mass ratio with respect to the machine
◦ For centrifugal machine it should be 2 to 3 times the weight of the machine.

◦ For Reciprocating type it should be 3 to 5 times the weight of the machine.
18 Refer Chapter 1 (Vol. 2) for these theoretical co-relations.

19 Even your boss at times. . . . . .

• The top of foundation is usually kept about 300 mm above the finished grade
elevation to prevent damage due to surface water run-off.
However this should be back checked with process department to ensure that
NPSH of the pump or piping connections will not be affected adversely.
• The vertical thickness of the foundation should be selected based on maximum
value of the following:
◦ Maximum embedded length of the anchor bolts plus 250 mm

◦ One fifth the width (least dimension) of the foundation in plan
◦ One tenth of largest dimension in plan
◦ A depth of 600 mm.
• The width of the foundation is selected based on the maximum value of the
following:
◦ Centre to centre distance of the anchor bolts plus 150 mm on both the side of
the foundation
◦ Length to the edge of the machine plus 300 mm at the both the ends of the
foundation
◦ 1 to 1.5 times the vertical distance from the bottom of foundation to the
machine centre line
◦ Once the width and height of foundation is selected the length can be
calculated based on the mass criteria as stated above.
• The plan dimension of the machine should be so adjusted that c.g. of the machine
assembly matches with c.g. of the foundation.
• For foundation resting on soil, eccentricity in c.g. of the machine and the
foundation shall not be more than 5%.
• For large reciprocating machines the embedded depth to be so adjusted that at
least 60 to 80% of the depth of the foundation is embedded in the soil. This will
increase the lateral restraint and damping ratio for modes of vibration.
We now give below some useful data and mathematical expressions which could
effective in day to day design office practise for design of block foundations.
2.5.15.6 Centre of gravity of the machine foundation

Here the whole machine foundation is broken into different segments having mass
as mi having co-ordinates as xi , yi , zi respectively then the c.g. of the foundation is
given by

mi x i m i yi m i zi
x̄ = i , ȳ = i and z = i
i mi i mi , i mi
Second moment of inertia of standard geometric shapes (Refer to Figures 2.5.35

and 36).

For foundations resting on soil the moment of inertia is calculated by
1 1
Ixx = LB3 ; Iyy = BL3 , and Izz = Ixx + Iyy
12 12
For foundations resting on piles or springs moment of inertia is calculated by

Ixx = yi2 ; Iyy = x2i , and Izz = x2i + yi2 e
i i i
lx
ly lz
Figure 2.5.35 A solid rectangular prism.
Figure 2.5.36 A solid circular cylinders.

2.5.15.7 Mass of inertia of geometrical shapes

For solid rectangular prisms (Refer to Figure 2.5.36)
Jx = m/12(ly2 + lz2 )
Jy = m/12(lx2 + lz2 )
Jz = m/12(lx2 + ly2 )
For solid circular cylinders

The second moment of inertia is given by

m 3 2 2 m m 3 2
Jx = D +l ; Jy = D2 ; Jz = D + l2
12 4 8 12 4
Here D = diameter of the cylinder; L = length of the cylinder.
2.6 SPECIAL PROVISIONS OF IS-CODE
We now give below some salient provisions and recommendations of IS-2974 for
rotary and reciprocating types of machines that constitute the normal design office
practice in India.
2.6.1 Recommendations on vibration isolation

To avoid transmission of vibration to adjoining parts of the buildings or other founda-
tions, it is necessary to provide a suitable isolation between the equipment foundation
and the adjoining structures.
This may be achieved by providing sand trench around the foundation block, the
thickness and depth of which shall be determined for each individual case.
As a rule the equipment foundation shall not be allowed to serve as a support for
other structures or for machines not related to the particular equipment.
In case it becomes necessary to support unimportant parts of other structures on the
machine foundation itself, measure shall be taken to make the connections resilient by
introducing gaskets made of rubber, cork, felt or other resilient materials.
2.6.2 Frequency separation

The natural frequency of the foundation system shall be such as to avoid resonance with
operating frequency of the machine and the amplitudes be kept below the permissible
limit.
Foundations for low frequency machine shall preferably be designed as such that
the natural frequency of the foundation is higher then the operating frequency of the
machine.
The natural frequency of any foundation should not preferably be within 20% of
the operating speed of the machine.

2.6.3 Permissible amplitudes

Normally the recommendation of the vendor supplying the equipment shall guide the
design, however in absence of such data the code recommends that if no resonance
is to occur in adjoining structure the amplitude of vibrations of a foundation at the
upper edge shall not exceed 0.20 mm in both directions.
When several foundations for similar machines are erected on a common mat the
computation for vibration shall proceed assuming that each machine foundation is
independent of others by breaking up the raft into sections corresponding to separate
foundations.
The design value for the permissible amplitude of vibrations may be increased
by 30%.
2.6.4 Permissible stresses

Concrete of grade M15 or higher shall be used for foundations. Concrete and steel
stresses are as specified in IS: 456-2000 shall be used for considering the dynamic loads
separately in detailed design. The following elastic moduli of concrete may be used in
design.
Grade of concrete Edyn (kN/m2 )
M15 250 × 106

M20 300 × 106
M25 340 × 106
M30 370 × 106
2.6.5 Concrete and its placing

The concrete used shall be controlled concrete conforming to design requirements.
The grade of concrete should generally be M15 to M20 for block foundation and M20
for frame foundation. The concrete shall be placed and designed in accordance with
IS: 456-2000.
The concrete used shall be of plastic consistency having an allowable slump, which
may vary between 50 to 80 mm. The water cement ratio shall not exceed 0.45. The
same consistency shall be maintained throughout the foundation.
2.6.6 Reinforcements
All foundation units of foundation shall be provided with top and bottom reinforce-
ment in two directions. Reinforcement shall be provided along the surface only in case
of block foundation.
The reinforcement in block foundation shall not be less than 25 kg/m3 .
The minimum diameter of bars shall be 12 mm with a maximum spacing of 200 mm
in order to care of the shrinkage.

16100
500 1 4 6 15 17
1442.5 9 12
1442.5 6770
2 A 7 10 B 13 C
D E
1442.5 11 14
1442.5
3 5 8 16 18
500 O
385 3285 2405 425 800 2400 800 800 4000 800
PLAN VIEW OF THE BLOCK FOUNDATION
C/L of Shaft of the machine

800 800 2000
1600
(typ.)
600
3600
ELEVATION OF THE BLOCK FOUNDATION
Figure 2.6.1 Plan and elevation of a gas turbine foundation.
2.6.7 Cover to concrete

For block foundation the concrete cover for protection of reinforcement shall be
75 mm at the bottom, 50 mm on both sides and 40 mm at top.
We now solve a practical design problem for a Gas Turbine resting on a block
foundation for your perusal and we hope that this will give you a better insight to
the aspect of how to apply the previously mentioned theories to the day-to-day design
office work of design of machine foundation (Figure 2.6.1).
Example 2.6.1
Design the gas turbine foundation shown in Fig. 2.6.1.
Design data
1 Bearing capacity of soil = 200 kN/m2

2 Shear wave velocity of soil = 125 m/sec
3 Density of soil = 20 kN/m3
4 Poisson’s ratio of soil = 0.25
5 C/L of shaft of the machine = 2.0 above T.O.C.
6 Operating frequency of turbine = 2250 r.p.m

7 Grade of concrete = M25

8 Grade of steel = Fe415
9 Allowable amplitude = 0.2 mm
10 Load at various anchor locations are as shown in the table hereafter.
Equipment load data at various Anchor Bolt locations
Static load Dynamic load

Vertical load Vertical Horizontal
Anchor bolt # (kN) load (kN) load (kN) Remarks
1 −311
2 −42 ±7 ±7
3 −311
4 −517
5 −517
6 −311
7 −50 ±6.76 ±6.76
8 −311
9 −200
10 −200 ±51 ±51 All horizontal force
is along global Y axes
11 −200
12 −350
13 −350 ±23 ±23
14 −350
15 −185 Anchor Bolt for generator
Total −4760
Calculate the natural frequency and amplitude based on Figure 2.6.1 and
using
• Richart and Lysmer model

• Richart and Lysmer model with embedment
• Wolf’s model
• Time history analysis based on Newmark-Beta method.
Solution:
Geometric property of the foundation
Area of foundation = 16.1 × 6.77 = 108.9 m2
1 1
Second Moment of Inertia = LB3 = 16.1 × 6.773 = 416.30 m4
12 12

Equivalent radius in vertical and horizontal mode:

108.9
r0 = = 5.89 m
π
Equivalent radius in rocking mode about the X-axis

4 × 416.30
rθ = = 4.80 m
4
Calculation of soil springs

Shear Wave velocity = 125 m/sec; Poisson’s ratio = 0.25.
Unit weight of soil = 20 kN/m3
20
Dynamic shear modulus (G) = ρVs2 = × (125)2 = 31855.25 kN/m2
9.81
Springs based on Richart’s model
4Gr0 4 × 31855.25 × 5.89

Kz = = = 1000679.6 kN/m
(1 − ν) (1 − 0.25)
32Gr0 (1 − υ) 32(1 − 0.25) × 31855.25 × 5.89
Ky = = = 900611.63 kN/m
(7 − 8υ) (7 − 8 × 0.25)
8Gr3θ 8 × 31855.25 × (4.8)3
Kφy = = = 12511907.52 kN/m.
3 (1 − υ) 3(1 − 0.25)
Table for calculation of c.g. and second moment of inertia of m/c & fdn.
(explained in next page)
Centre of gravity
11291.04 4997.63
x̄ = = 7.65 m from the point O; ȳ = = 3.39 m from the
1476.4 1476.4
3722.95
point O; and z̄ = = 2.52 m from the bottom of the foundation.
1476.4
8.05 − 7.65
Eccentricity in x direction = × 100 = 2.5% < 5% hence OK.
16.1
3.399 − 3.385
Eccentricity in y direction = × 100 = 0.07% < 5% hence OK.
6.77
m 2
Total mass moment of inertia = ly + lz2 + m(yoi
2 2
+ zoi ) + mz̄2
12
= 4703.00 + 4509.65 + 1476.4 × (2.52)2
= 18588.4 kN-m-sec2 .

Loading Lx Ly Lx Weigt m/l2(Ly 2 + m(yoi2 +
location (m) (m) (m) ht(kN) Mass Xi Yi Zi mi xi mi yi mi zi Lz2 ) yoi zoi zoi )
1 311 31.70 0.385 6.27 3.6 12.21 198.77 114.13 0.00 −2.89 −1.078 300.73
2 42 4.28 0.385 3.385 3.6 1.65 14.49 15.41 0.00 0 −1.078 4.98
3 311 31.70 0.385 0.5 3.6 12.21 15.85 114.13 0.00 2.885 −1.078 300.73
4 517 52.70 3.67 6.27 3.6 193.41 330.44 189.72 0.00 −2.89 −1.078 499.93
5 517 52.70 3.65 0.5 3.6 192.36 26.35 189.72 0.00 2.885 −1.078 499.93
6 311 31.70 6.075 6.27 3.6 192.59 198.77 114.13 0.00 −2.89 −1.078 300.73
7 50 5.10 6.075 3.385 3.6 30.96 17.25 18.35 0.00 0 −1.078 5.93
8 311 31.70 6.075 0.5 3.6 192.59 15.85 114.13 0.00 2.885 −1.078 300.73
9 200 20.39 7.3 4.828 4.6 148.83 98.42 93.78 0.00 −1.44 −2.078 130.49
10 200 20.39 7.3 3.385 4.6 148.83 69.01 93.78 0.00 0 −2.078 88.07
11 200 20.39 7.3 1.943 4.6 148.83 39.60 93.78 0.00 1.443 −2.078 130.49
12 350 35.68 9.7 4.828 4.6 346.08 172.23 164.12 0.00 −1.44 −2.078 228.35
13 350 35.68 9.7 3.385 4.6 346.08 120.77 164.12 0.00 0 −2.078 154.11
14 350 35.68 9.7 1.943 4.6 346.08 69.30 164.12 0.00 1.443 −2.078 228.35
15 185 18.86 11.3 6.27 3.6 213.10 118.24 67.89 0.00 −2.89 −1.078 178.89
16 185 18.86 11.3 0.5 3.6 213.10 9.43 67.89 0.00 2.885 −1.078 178.89
17 185 18.86 15.3 6.27 3.6 288.53 118.24 67.89 0.00 −2.89 −1.078 178.89
18 185 18.86 15.3 0.5 3.6 288.53 9.43 67.89 0.00 2.885 −1.078 178.89
A 6.5 6.77 3.6 3960.5 403.72 3.25 3.385 1.8 1312.08 1366.58 726.69 1977.97 0 0.722 210.24
B 4 6.77 3 2031 207.03 8.5 3.385 1.6 1759.79 700.81 331.25 946.02 0 0.922 175.86
C 5.6 6.77 3.6 3412.1 347.82 13.3 3.385 1.8 4625.96 1177.36 626.07 1704.10 0 0.722 181.13
D 0.8 5 1.6 160 16.31 7.3 3.385 3.8 119.06 55.21 61.98 37.46 0 −1.278 26.65
E 0.8 5 1.6 160 16.31 9.7 3.385 3.8 158.21 55.21 61.98 37.46 0 −1.278 26.65
14484 1476.40 11291.04 4997.63 3722.95 4703 4509.65

Calculation for damping

Based on Richart’s Formula we have
In the vertical direction
0.25 m(1 − υ)g 0.25 × 1476.4 × 0.75

Bz = = = 0.667
ρs r3z 2.03 × (5.89)3
0.425 0.425
→Dz = =√ = 0.52
Bz 0.667
√
Cz = 2Dz Kz m = 2 × 0.52 1000679.6 × 1476.4 = 39974 kN · sec/m
In the horizontal direction
(7 − 8ν) mg 5 1476.4
By = = × = 0.738
32 (1 − ν) ρs ry
3 32 × 0.75 2.04 × (5.89)3
0.288 0.288
Dy = =√ = 0.3352 and
By 0.738
√
Cy = 2Dy Ky m = 2 × 0.3352 900611.63 × 1476.4 = 24446 kN · sec/m
For the rocking mode
0.375(1 − ν)Jφy g 0.375 × 0.75 × 18588.4

Bφy = = = 1.013
ρs r5φy 2.03 × (4.798)5
0.15 0.15
Dφy = = √ = 0.074
(1 + Bφy ) Bφy 2.013 × 1.013
√
Cφy = 2Dφy Kφy m = 2 × 0.074 12511907.52 × 18588.4 = 71375 kN · sec/m
Calculation of natural frequencies

In the vertical direction

Kz 100679.6
ωz = = = 26 rad/sec (249 r.p.m)
m 1476.4
In the horizontal direction the equation of motion for free vibration is given by

m 0 ÿ Cx −Cy Zc
+
0 Jyφ φ̈ −Cy Zc Cφy + Cy Zc2 − WZc

ẏ Ky −Ky Zc y 0
× + =
φ̇ −Ky Zc Kφy + Ky Zc2 − WZc φ 0

For eigen value analysis we have

Ky − mλ −Ky Zc
=0
−Ky Zc Kφy + Ky Zc2 − WZc − Jφy λ
Here m = 1476.4; Jφ y = 18588.4; Zc = 2.52 m; W = 14484 kN; Ky =

900611.63 kN/m; Kφy = 12511907.52 kN/m

900611 − 1476.4λ −2269540
or, =0
−2269540 18194648 − 18588.4λ
The above matrix on expansion and simplification reduces to
λ2 − 1589λ + 409398 = 0;

1589 ± (1589)2 − 4 × 1 × 409398
λ= = 323.5, 1265
2
➔ ω2 = 17.98 rad/sec (172 r.p.m.); and ω3 = 35.56 rad/sec (340 r.p.m.)
Calculation of eigen vectors
For first mode, for ω = 17.98 rad/sec, we have

900611 − 1476.4 × 323.5 −2269540 φ11
=0
−2269540 18194648 − 18588.4 × 323.5 φ12
The above on expansion gives the following two equations
422995.6φ11 − 2269540φ12 = 0 and −2269540φ11 − 12181301φ12 = 0
Considering φ11 = 1.00 and solving the above homogenous equation we have,
φ12 = 0.186379442

φ11 1.00
Thus, =
φ12 0.186379442
For the Second mode we have

900611 − 1476.4 × 1265 −2269540 φ21
=0
−2269540 18194648 − 18588.4 × 1265 φ22
The above on expansion gives the following two equations:
−967035φ21 − 2269540φ22 = 0 and −2269540φ21 − 5319678φ22 = 0

Considering φ21 = 1.00 and solving the above homogenous equation we have
φ22 = −0.426092952
Thus the complete eigen vector matrix is given by

1.00 1.00
[ϕ] =
0.186379442 −0.426092952
Calculation of normalised eigen vectors

For the first mode

T 1476.4 0.0 1.00
{φ} [M] {φ} = 1.00 0.186379442
0.0 18588.4 0.186379442
√ √
The above on simplification gives, Mr = 2122.110 = 46.06
N
φ11
Thus dividing each term of the eigen vector by above we have, =
φ12

0.021707808
.
4.045889184 × 10−3

1476.4 0.0 1.00
{φ}T [M] {φ} = 1.00 −0.426092952
0.0 18588.4 −0.426092952
√ √
The above on simplification gives, Mr = 4851.22 = 69.65.
N
φ21
Thus dividing each term of the eigen vector by above we have, =
φ22

0.014357356
6.11756837 × 10−3
Thus the complete normalised eigen vector matrix is

21.707808 14.357356
[ϕ] = × 10−3
4.045889184 −6.11756837
Correction of damping matrix based on Rayleigh coeff icient

for modal analysis
We had already stated that damping matrix obtained from soil property is
non proportional and when considered in the analysis will not de-couple under
orthogonal transformation as such we correct the matrix enabling us to de-couple
the same20 .
20 Based on the theory of magnification factor damping may be ignored for this case for the ratio
of the fundamental frequencies of the foundation to the operating frequency of the machine is
more than 3.5. However for sake of clarification of the problem we continue to consider it in our
analysis.

Here,

Cx −Cy Zc 24446 −61604
[C] = =
−Cy Zc Cφy + Cy Zc2 − WZc −61604 190117.2
For the First mode we have

T −3 24446 61604
{φ} [C] {φ} = 21.707808 4.045889184 × 10
61604 190117.2

21.707808
× × 10−3
4.045889184
The above on simplification gives, {φ}T [C] {φ} = 3.8106 ➔ 2ζ1 ω1 = 3.8106 or
ζ1 = 0.105.
For the Second mode, we have

T −3 24446 61604
{φ} [C] {φ} = 14.357356 −6.11756837 ×10
61604 190117.2

14.357356
× × 10−3
−6.11756837
The above on simplification gives
{φ}T [C] {φ} = 22.97 ➔ 2D2 ω2 = 22.97 or D2 = 0.323.
Now considering the design damping as proportional Rayleigh damping,

we have
[C] = α[M] + β [K] or [φ]T [C] [φ] = α [φ]T [M] [φ] + β [φ]T [K] [φ]
and we have, 2D1 ω1 = α + βω12 and 2D2 ω2 = α + βω22 .

i.e. α + 323β = 3.8106 and α + 1265β = 22.97
Solving the above two simultaneous equations, we have: α = −2.7589 and
β = 0.0203
Substituting the above value of α and β we have,

1476.4 0 900611 −2269540
[C] = −2.7589 + 0.0203
0 18588.4 −2269540 18194648

14222 −46071
→ [C] = .
−46071 318233

Calculation of amplitude in the vertical mode
Vertical force = (7 + 6.76 + 51 + 23) = 87.76 sin 236t

We had calculated earlier that Dz = 0.52 and hence,
P0 sin ωm t/Kz 236

δz = , with, r = = 9.0.
(1 − r2 )2 + (2ζ r)2 26
87.76 sin 236t

➔ δz =
1000679.6 (1 − 81)2 + (2 × 0.52 × 9)2
= 1.08822 × 10−6 sin 236t m.
For, coupled sliding and rocking mode, we have

Lever arm = 3.6 + 2.0 − 2.52 = 3.08 m; Horizontal force = 87.76 sin 236t
Thus moment about the vertical centroid, Zc = 270 sin 236t

87.76
The force matrix can be represented as, {P} = sin 236t.
270
The equation of motion can be written as
& ' & '

[M] Ÿ + [C] Ẏ + [K] {Y} = {P}
With orthogonal transformation, we can write
& ' & '

[φ]T [M] [φ] ÿ + [φ]T [C] [φ] ẏ + [φ]T [K] [φ] {Y} = [φ]T {P}
which gives the following two equations
ÿ + 2D1 ω1 ẏ + ω12 y = p sin ωm t and θ̈ + 2D2 ω2 θ̇ + ω22 θ = m sin ωm t

T 21.707808 4.045889184 88 p
Here [φ] {P} = sin 236t →
14.357356 −6.11756837 270 m

3.0
= sin 236t
−0.388
i.e. ÿ + 2 × 0.105 × 17.98ẏ + 324y = 3.0 sin 236t → ÿ + 3.776ẏ + 324y =

3.0 sin 236t
and θ̈ + 2 × 0.323 × 35.56θ̇ + 1265θ = −0.388 sin 236t → θ̈ + 22.97θ̇ +
1265θ = −0.388 sin 236t
Thus for the horizontal translation, we have ÿ + 3.776ẏ + 324y = 3.0 sin 236t

236 3.776
where r = √ = 13.11 and ζ = √ = 0.104.
324 2 324
3 sin 236t
Hence, δy =
324 (1 − 171.8)2 + (2 × 0.104 × 13.11)2
= 5.4204 × 10−5 sin 236t
For the rocking mode
θ̈ + 22.97θ̇ + 1265θ = −0.388 sin 236t
236 22.97
where r = √ = 6.63 and ζ = √ = 0.322
1265 2 1265
−0.388 sin 236t

θ= = −7.1063 × 10−6 sin 236t
1265 (1 − 43.95)2 + (2 × 0.322 × 6.63)2
Thus in global co-ordinate, we have

Y 21.707808 14.357356 5.4204
= × 10−8
4.045889184 −6.11756837 −0.7106

107.462
= × 10−8 sin 236t
26.27
Net horizontal amplitude at top of foundation
Y= 107.463 × 10−8 + (3.6 − 2.52) × 26.27 × 10−8

= 1.358 × 10−6 m < 0.2 mm OK
Net horizontal amplitude at base of the foundation
Y = 107.463 × 10−8 − 2.52 × 26.27 × 10−8

= 41.263 × 10−8 m < 0.2 mm OK
We make here a very interesting comparison, shown in Figure 2.6.2, is the time
history response of the block foundation with non-proportional soil damping
and corrected proportional Rayleigh damping, we have obtained earlier.
It will be observed that values are quite closely matching and for practical
engineering work this is deemed sufficient.

Comparison of amplitude based on time history

0.000015
0.00001
0.000005
Amplitude
Displacement with non

0 proportional damping
1 23 45 67 89 111 133 155 177 199 221 243 265 Displacement with corrected
-0.000005 proportional damping
-0.00001
-0.000015
Time steps
Figure 2.6.2
Check of local vibration of the pedestals

Width of the pedestal = 800 mm; Depth of the pedestal = 5000 mm; Height
of pedestal = 1600 mm.
BD3 0.8 × 125

I= = = 8.33 m4
12 12
Considering the pedestal as cantilever beam,
3EI 3 × 300 × 106 × 8.33

Kh = = = 1.83 × 109 kN/m
L3 (1.6)3
Self weight of the pedestal = 0.8 × 5.0 × 1.6 × 25 = 160 kN; Weight from
machine = 3 × 200 + 3 × 350 = 1650 kN
→ Total weight = 1810 kN

2 Kh
Thus total mass(m) = W/g = 184.5 kN-sec /m and hence ω = =
m
1.83 × 109
= 3149 rad/sec
184.5
ωm 236
And, r= = = 0.07.
ωn 3149
As the frequency ratio is very low we neglect the damping we have,
P0 /k 51 sin 236t
y= = = 2.80 × 10−8 m < 0.2 mm.
(1 − r )
2 1.83 × 109 (1 − 0.072 )

Calculation based on embedment effect

We had already calculated before that based on property of soil individual soil
stiffness in various modes as
Kz = 1000679.6 kN/m,Cz = 39974kN.sec/m; Ky = 900611.63kN/m,

and Cy = 24446kN/m, Kφy = 12511907.52 kN/m, Cϕy = 71375 kN/m.
Embedment factor based on Richart’s table for soil stiffness

h
ηz = 1 + 0.6(1 − υ) , here h = 3.0 m as per the problem, and rz =5.89 m
rz
which gives,
h
ηz = 1.23 similarly, ηy = 1 + 0.55(2 − ν) and this gives ηy = 1.497 and
ry

h h 3
ηφy = 1 + 1.2(1 − ν) + 0.2(2 − ν) and this gives ηφy = 1.5625.
rφy rφy
Embedment factor based on Richart’s table for soil damping
1 + 1.9(1 − ν) rhz 1 + 1.9(2 − ν) rhy

αz = √ = 1.556; αy = √ = 2.20 and
ηz ηy
3
1 + 0.7(1 − ν) rhφy + 0.6(2 − ν) h
rφy
αφy = √ = 1.267.
ηφy
Thus considering the embedment factor the stiffness and damping value gets
modified to:
Kze = 1230836 kN/m, Cze = 62204 kN · sec/m

Kye = 1348214 kN/m, Cye = 53812 kN/m
e = 19549856 kN/m, C e = 90432 kN/m

and Kφy ϕy
For the vertical direction, we have

Kze 1230836
ωz = = = 29 rad/sec (277 r.p.m);
m 1476.4

We had calculated earlier that Dz = 0.73;
P0 sin ωm t/Kz 236

Thus δz = , where r = = 8.13
(1 − r2 )2 + (2ζ r)2 29
87.76 sin 236t

➔δz =
1230836 (1 − 66)2 + (2 × 0.52 × 8.13)2
= 1.07754 × 10−6 sin 236t m
In horizontal direction the equation of motion for free vibration is given by:

m 0 ÿ Cx −Cy Zc ẏ
+
0 Jyφ φ̈ −Cy Zc Cφy + Cy Zc2 − WZc φ̇

Ky −Ky Zc y 0
+ =
−Ky Zc Kφy + Ky Zc2 − WZc φ 0

Kye − mλ −Kye Zc
=0
−Kye Zc Kφy + Kye Zc2 − WZc − Jφy λ
Here m = 1476.4; Jφ y = 18588.4; Zc = 2.52 m; W = 14484 kN; Kye =

e
1348214 kN/m; Kφy = 19549856 kN/m.

1348214 − 1476.4λ −3397499
→ =0
−3397499 28075055 − 18588.4λ
λ2 − 2423λ + 958897 = 0 → λ = 498, 1925

➔ ω2 = 22.3 rad/sec (213 r.p.m.), and ω3 = 43.87 rad/sec (419 r.p.m.).
Now proceeding in the exact manner as explained in the previous case, we

arrive at the result:
Net horizontal amplitude at top of foundation
Y = 107.14 × 10−8 + (3.6 − 2.52) × 26.09 × 10−8

= 1.353 × 10−6 m < 0.2 mm OK
Net horizontal amplitude at the base of foundation
Y = 107.14 × 10−8 − 2.52 × 26.09 × 10−8 = 41.4 × 10−8 m < 0.2 mm OK

Calculation based on Wolf’s Model

Based on formulation proposed by Wolf
4Gr0 4 × 31855.25 × 5.89

Kz = = = 1000679.6
(1 − υ) (1 − 0.25)
rz 5.89 × 1000679.6 × 0.58
Cz = K z γ0 = = 27348 k · sec/m
Vs 125
r 2
z 5.89 2
mz = Kz μ 0 = × 1000679.6 × 0.095 = 211 k-sec2 /m
Vs 125
In the horizontal direction we have
8Gry 8 × 31855.25 × 5.89

Ky = = = 2001359 k/m
1−υ 0.75
ry 5.89 × 2001359 × 0.85
Cy = K y γ0 = = 80158 k · sec/m
Vs 125
r 2
y 5.89 2
my = Ky μ 0 = × 2001359 × 0.27 = 1200 k-sec2 /m
Vs 125
In rocking mode, we have
8Gr3θ 8 × 31855.25 × (4.8)3

Kφ y = = = 12511907.52 k/m
3 (1 − υ) 3(1 − 0.25)
r 2
y 4.8 2
Jφy = Kφy μ0 = × 12511907 × 0.24 = 4428 k-sec2 /m
Vs 125
0.3 0.3
γ0 = 3(1−υ)m
= 3 × 0.75 × 4428
= 0.242
1+ 8r5θ ρ
1+ 8 × (4.8)5 × 2.04
rφy 4.8 × 12511907 × 0.242

Thus, Cφy = Kφy γ0 = = 116271 k · sec/m
Vs 125
For vertical direction we have

Kz 1000679
ωz = = = 24.35 rad/sec (277 r.p.m)
m 1687

27348
Damping Ratio Dz = √ = 0.332
2 1000679 × 1687
P0 sin ωm t/Kz 236

Thus δz = where r = = 9.7
(1 − r2 )2 + (2Dr)2 24.34

87.76 sin 236t

➔ δz =
1000679 (1 − 94)2 + (2 × 0.332 × 9.7)2
= 0.94076 × 10−6 sin 236t m
For coupled sliding and rocking mode

m 0 ÿ Cx −Cy Zc ẏ
+
0 Jyφ φ̈ −Cy Zc Cφy + Cy Zc2 − WZc φ̇

Ky −Ky Zc y 0
+ =
−Ky Zc Kφy + Ky Zc − WZc
2 φ 0
For eigen value analysis we

Kye − mλ −Kye Zc
=0
−Kye Zc Kφy + Kye Zc2 − WZc − Jφy λ
Here m = 2676; Jφy = 23016; Zc = 2.52 m; W = 14484 kN; Kye = 2001359

kN/m; Kφy e = 12511907 kN/m.
It is to be noted that here mass and moment of inertia is the mass/inertia

of machine and foundation plus the mass/inertia of soil participating in the
vibration.
Thus

2001359 − 2676λ −5043425
=0
−5043425 25184838 − 23016λ
λ2 − 1842λ + 405381 = 0 ➔ λ = 255, 1587
Hence, ω2 = 15.96 rad/sec (152 r.p.m.) and ω3 = 39.83 rad/sec (380 r.p.m.)
Now proceeding in the manner as explained in the case of Richart’s model we
arrive at the result;
Net horizontal amplitude at top of the foundation
Y = 63 × 10−8 + (3.6 − 2.52) × 37.646 × 10−8

= 1.0365 × 10−6 m < 0.2 mm OK
Net horizontal amplitude at the base of foundation
Y = 63 × 10−8 − 2.52 × 37.646 × 10−8 = −31.86 × 10−8 m < 0.2 mm OK

The values obtained by the three methods are summarised hereafter:
Comparison of natural frequencies
Based on the method ωz (r.p.m) ωy (r.p.m) ωθy (r.p.m)
Richart’s formula 249 172 340

Richart with embeddment 277 213 419
Wolf’s method 277 152 380
Comparison of amplitude
Based on the method δz (mm) δy (mm) δθy
Richart’s formula 1.0882 × 10−3 1.358 × 10−3 41.263 × 10−8

Richart with embeddment 1.077 × 10−3 1.353 × 10−3 41.4 × 10−8
Wolf’s method 0.94076 × 10−3 1.0365 × 10−3 −31.86 × 10−8
• Based on Time History Analysis
We perform time history analysis for springs based on Richart’s method and
Wolf’s Method.
Here time history response has been done for 215 steps with complete soil
damping into consideration and shown in Figures 2.6.3 and 4.
Time History based on Newmark Beta Method with Richarts Spring

0.000008
AMPLITUDE
0.000006
0.000004
0.000002 Displacement in Y direction
0 Angular Rotation
-0.000002 1 17 33 49 65 81 97 113 129 145 161 177 193 209
-0.000004
-0.000006
Time Steps
Figure 2.6.3
Time History based on Newmark Beta Method with Wolf's Spring

Spring
0.000003
0.000002
AMPLITUDE
0.000001 Displacement in Y direction

Angular Rotation
0
-0.000001 1 20 39 58 77 96 115 134 153 172 191 210
-0.000002
-0.000003
Time Steps
Figure 2.6.4

2.7 ANALYSIS AND DESIGN OF MACHINE FOUNDATION

UNDER IMPACT LOADING
2.7.1 Introduction
In this section we will deal with foundations subjected to impact loading. These type
of foundations usually constitute of hammer foundations used for forging or hydraulic
stamps used to flatten steel billets to make plates out of them.
The arrangement of the hammer foundation is usually as shown Figure 2.7.1.
The hammer foundation, consists of a hammer or a tup which falls repeatedly on
an anvil. The anvil in turn is placed on an elastic pad resting on a massive RCC block.
The elastic pad is used to isolate the foundation from the surrounding and minimize
the harmful effect of the vibration induced by the hammer dropping on the anvil. The
elastic pad also acts as damper to reduce the net amplitude of vibration of the anvil
and the foundation.
Depending upon the functionality, the frame of the hammer may either rest on the
foundation block as shown above or may even rest on a separate foundation.
While planning the foundation it is usually ensured that the center line of the anvil
is concentric with the center of gravity of the base of the foundation. This ensures that
the amplitude of vibration is restricted to vertical translation only and does not give
rise to any coupled motion including rocking21 .
At times when the hammer is very heavy the foundation is further isolated by
providing elastic pad/springs along with dampers below the RCC block too.
Shown in Figure 2.7.2 is a hammer foundation where other than the anvil the RCC
block is also mounted on springs and dampers to isolate the transmittal of vibration
to the surrounding.
The springs or the elastic pad which are placed below the RCC block is usually
an expensive item and care should be taken to protect them from exposure to water,
chemicals, oils etc which could otherwise damage their properties. This is usually done
by providing a protective RCC trough all round the foundation and sealing the same
at the top of the foundation level.
The elastic material used under the anvil or the RCC block could be of cork, tim-
ber or even specialized mechanical springs and dampers supplied by vendors having
technological expertise in isolation techniques of these type of foundations.
2.7.1.1 How does the behavior of a mechanical system

under impact differ from externally applied
harmonic loads?
We do not tender any apologies for posing so fundamental a question, for in our
experience in teaching this subject, as well as interacting with professionals in the
industry for over two decades, we have been somewhat startled to find that though
people can arrive at the design values for various type of machine foundation quite
accurately by following the code stipulation blindly, but how does the characteristics
21 Hammer foundation having eccentric anvil is though uncommon but surely not rare. We will deal with
this particular case separately later.

Frame
Hammer/Tup
RCC Fdn.
Anvil
Elastic Pad
F.G.L.
Figure 2.7.1 General arrangement of a Typical Hammer Foundation.
Frame
Hammer/Tup
RCC Fdn.
RCC Trough
Anvil
Elastic Pads/Springs
Figure 2.7.2 General arrangement of a Typical Hammer Foundation mounted on spring with R.C.C.
Trough.

of the behavior differs in the above two cases. . . the picture has remained gray to
many. So before we take a plunge into the mathematical aspect of the design it would
be worthwhile to understand the conceptual aspect of the problem and reflect a bit on
how the two cases differ in transmitting the vibration to the system.
We hypothesize two pictures for this from our day to day life.
1 Imagine a boy continuously jumping on a plank supported at two ends for some
time.
2 A Karatika giving a vicious chop to the same plank at some point on it22 .
2.7.1.2 How do the planks behave under these two conditions?

For the first case if we have a stop watch we can measure the time taken by the boy
when he is at rest on the plank to the time he jumps (presuming with same monotony)
and again comes to rest on the plank. If we take this as his time period of vibration
T it is possible to find out the frequency of his motion from the relation, ω = 1/T.
If we now measure the weight of the boy we can say that the plank is subjected to
a continuous external force of W sin ωt, where W is the weight of the boy and ω
is the frequency with which he is jumping on the plank. Thus the plank is under a
forced harmonic load and will also produce amplitudes which will be a function of
the external force expressed as W sin ωt. This is called harmonic force23 .
While in the second case when the Karatika executes the chop he is transferring his
potential energy into a kinetic energy and is transferring this energy to the plank in a
very short period of time (may be some small fraction of a second) and then it ceases
to exist. This is quite unlike the earlier case when the external force continues to excite
the plank till the boy continues to jump on it.
These type of forces when induced on a body where it is subjected to force for a
very short time is known as an impact load or in technical term we call it a transient.
We consciously or otherwise often observe this phenomenon quit often in our day
to day life like
• A hammer used to put a nail in place

• A mallet used to hit a golf ball
• A ship hitting the jetty fenders when coming to rest on a port
These are all cases of impact forces acting on a system.

For instance, if we take the case of a mallet hitting the golf ball what we do is take
a swing up when we concentrate our potential energy and with the down swing of the
club we transfer the potential energy to kinetic energy which is then transferred to the
ball at the instant of impact.
Now suppose we connect the ball to a spring, we will observe that the ball starts
vibrating to and fro with respect to its mean position.
22 The two cases are mutually exclusive.

23 Block Foundation dealt earlier is a typical example of this when the unbalanced mass induces an external
force that is harmonic in nature.

To understand the effect of impact further we a take a step backward and formulate
a problem from our days of engineering mechanics/School Physics as hereafter.
Example 2.7.1
Shown in Figure 2.7.3 is a metal block of weight 100 kN suspended from a
point O by a mass less inextensible string having a length of 2.5m. It is released
from rest from a position 90 degree to vertical position of rest as shown below.
The block of 100 kN hits another metal block of weight 500 KN connected to a
spring of stiffness 2500 kN/m at point X. Considering the collision to be perfectly
elastic find out the amplitude of vibration of the body considering friction less
surface having
W1 = 100 kN
2500 mm
X W2 = 500 kN
K = 2500 kN/m
Figure 2.7.3 Conceptual diagram of the system.
• Un-damped motion.
• Damped Motion having a damper connected to W2 of magnitude 125 kN ·
sec/m
Solution:
When the body is released form its position of rest it takes a swing and hits the
500 kN body at point X.
The potential energy of the 100 kN body at its initial position = W1 h

1 W1
Kinetic energy of the 100 kN body at the point of impact = u21
2 g
Applying the law of conservation of energy i.e. KE = PE,


we have, u1 = 2gh m/sec: For h = 2.5 m → u1 = 7m/ sec
Now let,
u1 = Initial velocity of the body W1 @ 7.0 m/sec before collision;

u2 = Initial velocity of the body W2 @ 0.0 m/sec before collision;
v1 = Final velocity of the body W1 after collision, and
v2 = Final velocity of the body W2 after collision.
Then based on conservation of momentum and the collision being elastic we

have,
W1 W2 W1 W2
u1 + u2 = v1 + v2
g g g g
1 1 1 1
or, m1 u1 + m2 u2 = m1 v1 + m2 v2 and m1 u21 + m2 u22 = m1 v12 + m2 v22
2 2 2 2
Based on the above boundary conditions, we have
m1 u1 = m1 v1 + m2 v2 and m1 u21 = m1 v12 + m2 v22
Substituting the numerical data mentioned in the problem, one can have
100v1 + 500v2 = 700 and 100v12 + 500v22 = 4900
Solving the above two equations we have ; v1 = −3.90 m/ sec and v2 =

2.18 m / sec.
Here the negative value for v1 means that the 100 kN body will rebound back
with a velocity of 3.9 m/sec.
Now applying D’Alembert’s equation to the body connected to the spring we
have
mẍ + Kx = 0 where K = spring stiffness.
dv
→ m + Kx = 0, where v = velocity vector of the body
dt
dv dx dx
i.e. m + Kx = 0, as = v we have, mvdv + Kxdx = 0
dx dt dt
The above differential equation has boundary condition as at v = v0 , x = 0

and at v = 0, x = δ
0 δ
m
→ m vdv + Kxdx = 0, this on simplification gives, δ = v .
K
v0 0

v
Also, δ = ω where ω = the natural frequency of the structure.

K 2500 × 9.81
Thus, ω= = = 7 rad/ sec
m 500
v 2.18
The amplitude, δ = = = 0.3114 m.
ω 7
It is to be noted that from the above calculation we have managed to find out
only the magnitude of the maximum amplitude.
It does not tell us how the body will vibrate under this impact force.
To get this history let us consider the differential equation
mẍ + Kx = 0; and let, x = C1 sin ωt + C2 cos ωt
be the solution to the above.

Applying the boundary condition at t = 0 x = 0 and v = 2.18 m/sec, we have
At t = 0 x = 0 → C2 = 0 or x = C1 sin ωt and ẋ = C1 ω cos ωt at t =
0; → C1 = ωẋ = vω0 from which we deduce,
v0
x= sin ωt → x = 0.3114 sin 7t
ω
The above when plotted at time step of 0.05 seconds shows a curve as furnished
in Figure 2.7.4.
With damped vibration for single degree of freedom, we have seen earlier in
Chapter 3 (Vol. 1) that amplitude of vibration is given by
x = e−Dωn t [C1 cos ωd t + C2 sin ωd t]
√
where, ωd = ωn (1 − D2 ) and D = c/cc and cc = 2 km.
Displacement history under initial velocity v0

0.4
0.3
Amplitude (meter)
0.2
0.1
0 Amplitude (meter)
1
11
16
21
26
31
36
41
46
51
56
61
66
-0.1
-0.2
-0.3
-0.4
Time Steps
Figure 2.7.4 Displacement history undamped case.

Now, for t = 0 when x = 0 → C1 = 0 which reduces the above equation to
x = C2 e−Dωn t sin ωd t
Again for t = 0, ẋ = v0 we have, C2 = v0 /ωd which results in the equation,
v0
x= e−Dωn t sin ωd t
ωn (1 − D2 )
Here we have, v0 = 2.18 m/ sec, ωn = 7 rad/ sec, D = 0.175.

Substituting the above values, we have → x = 0.316e−1.225t sin 6.892t.
Plotting the above values at time step of 0.05 sec we see the time history curve
is as given in Figure 2.7.5.
Damped displacement under initial velocity v0

0.3
Amplitude (meter)
0.2
0.1
Amplitude (m)
0
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
-0.1
-0.2
Time Steps
Figure 2.7.5 Displacement history damped case.
The above response shows some very interesting results. While for un-damped
motion the curve follows a sinusoidal pattern, for damped case it initially starts with
peak amplitude and quickly dies down due to the inherent damping in the system
in contrary to the harmonic loading, where the body continues to vibrate under the
application of the externally applied force.
So far so good, we have managed to arrive at the behavior pattern of a system having
a single degree of freedom subjected to impact load albeit some idealization such as
• The string is inextensible and mass less

• The collision is perfectly elastic24
• The spring and damper is mass less having identified definite values.
24 There is no collision in nature that is perfectly elastic for some energy is always dissipated out in form
of heat or sound thus we usually use a term co-efficient of restitution. We will learn more about it
subsequently.

Now the question boils down to how does the above problem relates to a hammer
foundation which we are supposed to discuss herein?
To explain this, we need to clarify how does a hammer foundation work?
Based on the General arrangement of hammer foundation shown earlier, the ham-
mer or the tup either undergoes a free fall on the anvil or falls under a certain pressure
(for double acting hammers). It either flattens or forge the metal on the anvil to a desired
shape or may even crush it to lower particle size (in case of a crusher) depending upon
for what purpose the machine is being put to use.
Irrespective of its function, the basic point that remains unaltered is the following:
In contrary to the foundation supporting centrifugal or reciprocating type of
machines where the foundation is subjected to a constant external harmonic force
the hammer foundation induces a transient force at the point of collision and then
ceases to exist till the next blow is induced25 . Thus based on the above statement we
can postulate that for design of machine foundations of this type we need to analyze
the system subject to transient shocks. Hence as a first step let us see what type of
mathematical model is in vogue for analysis of these types of foundations.
2.7.2 Mathematical model of a hammer foundation

For foundations resting on ground supporting anvils mounted on elastic base we
usually consider a system having two degrees of freedom as shown hereafter.
Shown in Figure 2.7.6 is the mathematical model of a hammer foundation resting
on soil where,
m = mass of the hammer or the tup;

m1 = mass of the foundation block plus frame resting on it if any;
m2 = mass of the anvil resting on elastic pad;
k1 = soil spring to be calculated from Barkan or Richart’s formula26 ;
k2 = spring value for the elastic pad, this value is normally furnished by the vendor
supplying these pads;
c1 = damping of the soil to be obtained from Richart’s formula;
c2 = damping of the elastic pad again furnished by the vendor;
H = height of the free fall of the hammer;
x2 = amplitude vector of the anvil; and
x1 = amplitude vector of the Foundation.
We had already seen in earlier27 that for bodies having two degrees of freedom the
free equation of vibration is given by

m1 0 ẍ1 k + k2 −k2 x1
+ 1 = 0, (2.7.1)
0 m2 ẍ2 −k2 k2 x2
25 We hope by now the reader can smell the congruence with the worked out example 2.7.1.
26 Refer to section of block foundation for the formula of the springs and dampers.
27 Refer Chapter 5 (Vol. 1) on basic concepts in Structural Dynamics.

m2
x2
k2 c2
m1
x1
c1
k1
Figure 2.7.6 Mathematical model of hammer foundation resting on soil directly.
and since this is a statically coupled equation the damped free vibration of motion is
given by (Meirovitch 1975)

m1 0 ẍ1 c + c2 −c2 ẋ1 k + k2 −k2 x1
+ 1 + 1 =0 (2.7.2)
0 m2 ẍ2 −c2 c2 ẋ2 −k2 k2 x2
2.7.2.1 Mathematical model of foundation

resting inside a trough
In this case the mathematical model for analysis of the system is as shown in
Figure 2.7.7.
In the above mathematical model,
m = mass of the hammer or the tup;
m1 = mass of the trough resting on soil;
m2 = mass of the foundation plus hammer frame resting on it; if any,

m3
x3
k3 c3
m2
x2
k2 c2
x1
m1
k1 c1
Figure 2.7.7 Mathematical model of hammer foundation resting inside a trough.
m3 = mass of the anvil resting on elastic pad supplied by vendor;

k1 = soil spring to deduced from either Barkan, Richarts formula;
k2 = spring value for the elastic pad/spring on which the foundation is resting whose
value is normally furnished by the vendor supplying these pads;
k3 = spring value for the elastic pad supporting the anvil whose value is normally
furnished by the vendor supplying these pads;


c2 = damping of the elastic pad /spring supporting the foundation on the trough;
c3 = damping of the elastic pad supporting the anvil;
H = height of the free fall of the hammer;
x1 = amplitude of the trough resting on soil;
x2 = amplitude of the RCC block resting on springs; and
x3 = amplitude of the anvil resting on the elastic pad.
The free vibration of motion for the above is given by
⎡ ⎤⎧ ⎫ ⎡ ⎤⎧ ⎫
m1 0 0 ⎨ẍ1 ⎬ k 1 + k2 −k2 0 ⎨x1 ⎬
⎣0 m2 0 ⎦ ẍ2 + ⎣ −k2 k2 + k 3 −k3 ⎦ x2 = 0 and
⎩ ⎭ ⎩ ⎭
0 0 m3 ẍ3 0 −k3 k3 x3
the damped equation of motion is given by

⎡ ⎤⎧ ⎫ ⎡ ⎤⎧ ⎫
m1 0 0 ⎨ẍ1 ⎬ c 1 + c2 −c2 0 ⎨ẋ1 ⎬
⎣0 m2 0 ⎦ ẍ2 + ⎣ −c2 c2 + c 3 −c3 ⎦ ẋ2
⎩ ⎭ ⎩ ⎭
0 0 m3 ẍ3 0 −c3 c3 ẋ3
⎡ ⎤⎧ ⎫
k1 + k2 −k2 0 ⎨x1 ⎬
+ ⎣ −k2 k+ k 3 −k 3 ⎦ x2 = 0 (2.7.3)
2 ⎩ ⎭
0 −k3 k3 x3
Based on Equation (2.7.3), we have managed to formulate the equation of motion

for the hammer foundation, but how it will behave under transient load is yet to be
ascertained, though. Based on Example 2.7.1, we have shown the vibration charac-
teristics of a body having single degree of freedom under impact loading, we will now
extend this theory to multi-degrees of freedom.
2.7.2.2 Un-damped Response of a system under

impact loading having multi-degree of freedom
We write equation of motion in matrix notation as
& '
[M] Ẍ + [K] {X} = 0 (2.7.4)
Here [M] = A square mass matrix of the order n × n;

[K] = A square stiffness matrix of the order n × n;
{X} = A Column deflection matrix of order n × 1 (which means n rows and 1
column).
Based on the orthogonal property of the matrix28 we have
[ϕ]T [M][ϕ]{Ẍ} + [ϕ]T [K][ϕ]{X} = 0 which de-couples to (2.7.5)
28 Refer Chapter 5 (Vol. 1) for the orthogonal property of the Matrix.


N
N
{{ξ̈i } + ωi2 {ξi }} = 0 where {X} = [φ]{ξi } (2.7.6)
i=1 i=1
Now let {ξi } = {Ai sin ωi t + Bi cos ωi t}. (2.7.7)

N
N
Since, {X} = [φ] {ξi } , we have, {X} = [φ] {Ai sin ωi t + Bi sin ωi t}
i=1 i=1
Multiplying both sides of the above expression by the term [φ]T [M] we have

N
[φ]T [M] {X} = [φ]T [M] [φ] {Ai sin ωi t + Bi sin ωi t} which reduces to
i=1

N
[φ]T [M] {X} = {Ai sin ωi t + Bi sin ωi t}
i=1
Now imposing the boundary condition at t = 0 {X} = {0}, we have

N
{0} = {Ai sin ωi t + Bi cos ωi t}, which implies
i=1

N
{Bi } = {0}, thus we have, [φ]T [M]{X} = {Ai sin ωi t}
i=1
Again imposing the boundary condition at t = 0, {Ẋ} = {V0 } we have

N
[φ] [M] {V0 }
[φ]T [M] {V0 } = {Ai ωi } ⇒ {Ai } = , which gives
ωi
i=1

N T
[φ] [M] {V0 }
{X} = [φ] {sin ωi t} (2.7.8)
ωi
i=1
Based on the above we can clearly infer that for {sin ωi t} = {1}, we have the
maximum value of the amplitude vector.
We now explain further, the phenomenon based on a suitable numerical example.
For the numerical worked out problem below, we have deliberately used a
theoretical data with an objective that you can follow the process clearly.

Example 2.7.2
For a system having the following data find out the amplitude of vibration when
the mass m2 is subjected to an initial velocity of 0.5 m/sec.

30 −10 1 0 0.0
[K] = , [M] = , and {V} =
−10 30 0 1 0.5
Solution:
The free vibration of motion for the body is given by

m1 0 ẍ1 k11 −k12 x1
+ =0
0 m2 ẍ2 −k21 k22 x2

30 − λ −10
The eigen value solution of the problem is expressed as =0
−10 30 − λ
and this on simplification reduces to
(λ − 20) (λ − 40) = 0 which gives

λ = 20 or ⇒ ω1 = 4.35 rad/sec; and λ = 40 : ⇒ ω2 = 6.32 rad/sec.
Calculation of the eigen-vectors

For the first mode having λ = 20
10φ1 − 10φ2 = 0 and − 10φ1 + 10φ2 = 0
Thus for φ1 = 1, we have φ2 = 1
For the second mode having λ = 40

−10φ1 − 10φ2 = 0 and −10φ1 − 10φ2 = 0
Thus for φ1 = 1 we have φ2 = −1

1 1
This the complete eigen vector matrix as [ϕ] = .
1 −1
The normalized eigen vector, based on orthogonal transformation theory is

given by
√ √
1/√2 1/ √2
[ϕ]n =
1/ 2 −1/ 2

N . [φ]T [M]{V0 } /
We had deduced earlier that {X} = i=1 ωi [φ] {sin ωi t} ,
where {X} = [φ] {ξi } , thus based on mathematical symmetry we have {ξi } =
N . [φ]T [M]{V0 } /
i=1 ωi {sin ωi t} .
Hence for the first mode

1 1 1 0 0 1
{ξ1 } = √ √ × × sin 4.5t
2 2 0 1 0.5 4.5
the above on simplification reduces to [ξ1 ] = 0.07856 sin 4.5t


1 1 1 0 0 1
{ξ2 } = √ −√ × × sin 6.32t
2 2 0 1 0.5 6.32
the above on simplification reduces to {ξ2 } = −0.056 sin 6.32t

√
√
1/ √2 0.07856 sin 4.5t 1/√2
As {X} = [φ] {ξi } , we have, {X} =
−1/ 2 −0.056 sin 6.32t 1/ 2

0.0555 sin 4.5t − 0.0396 sin 6.32t
➔ {X} =
0.0555 sin 4.5t + 0.03965 sin 6.32t
The above when plotted at a time step of 0.05 sec gives plots as shown in
Figure 2.7.8.
Undamped motion of the system

0.12
0.1
0.08
0.06
0.04
Amplitude
x1
0.02
x2
0
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65
-0.02
-0.04
-0.06
-0.08
-0.1
Time Steps
Figure 2.7.8 Undamped response of a system under impact loading having multi-degree of
freedom.
The above plot shows how the two degree system body vibrates under a
transient initial velocity of 0.5 m/sec.

2.7.2.3 Damped response of a system under impact

loading having multi-degree of freedom
We had already discussed in our earlier chapters that undamped motion of vibra-
tion is an idealization which is practically rare in nature and as such it would now
be worthwhile to asses (Chowdhury et al. 2002) how damping affect the above
phenomenon.
We write equation of motion in matrix notation as
[M]{Ẍ} + [C]{Ẋ} + [K]{X} = 0 (2.7.9)
where
[M] = a square mass matrix of the order n × n;
[K] = a square stiffness matrix of the order n × n;
[C] = a square damping matrix of order n × n;
{X} = a column deflection matrix of order n × 1
(which means n rows and 1 column)
Based on the orthogonal property of the matrix we have
[ϕ]T [M][ϕ]{Ẍ} + [ϕ]T[C][ϕ]{Ẋ} + [ϕ]T [K][ϕ]{X} = 0 (2.7.10)
which de-couples to

N
N
{{ξ̈i } + 2Di ωi {ξ̇i } + ωi2 {ξi }} = 0 where {X} = [φ]{ξi } (2.7.11)
i=1 i=1
We have already proved earlier that for body having single degree of freedom the
free damped equation of motion is given by
x = e−Dωn t [C1 cos ωd t + C2 sin ωd t] (2.7.12)

where, ωd = ωn [1 − D2 ].
Thus in transformed co ordinate when the equations get de-coupled we can write
Let {ξi } = e−Di ωni t {Ai sin ωdi t + Bi cos ωdi t} , and (2.7.13)

N
since {X} = [φ] {ξi } , we have,
i=1

N
{X} = [φ] e−Di ωni t {Ai sin ωdi t + Bi cos ωdi t} (2.7.14)
i=1

Multiplying both sides of the above expression by the term [ϕ]T [M] we have

N
[φ]T [M]{X}= [φ]T [M][φ]e−Di ωni t {Ai sin ωdi t + Bi cos ωdi t} which reduces to
i=1

N
T
[φ] [M] {X} = e−Di ωni t {Ai sin ωdi t + Bi cos ωdi t}
i=1
Now imposing the boundary condition at t = 0 {X} = {0}, we have,

N
{0} = e−Di ωni t {Ai sin ωdi t + Bi cos ωdi t} which implies,
i=1

N
[Bi ] = [0] , thus we have [φ] [M] [X] = T
e−Di ωni t {Ai sin ωdi t}
i=1
Again imposing the boundary condition at t = 0 {Ẋ} = {V0 } we have,

N
[φ]T [M] {V0 }
[φ]T [M] {V0 } = {Ai ωdi } ⇒ {Ai } = , which gives
ωi
i=1

N
[ϕ]T [M]{V0 }
{X} = [ϕ]e−Di ωni t {sin ωi t} (2.7.15)
ωdi
i=1
It will be interesting to note that in this case the maximum amplitude does not occur
at ωi t = π/2 as in the case of undamped vibration.
To get the maximum amplitude we need to plot the complete time history which
was not required for the undamped case.
We now further explain the phenomenon based on a suitable numerical example.
Example 2.7.3
Repeat the problem worked out in Example 2.7.1 with following damping
ratio D1 = 0.15 and D2 = 0.20. All other parameters remain the same as
in Example 2.7.1.

Solution:
The free damped equation of motion for the problem is given by

m1 0 ẍ1 c −c12 ẋ1 k11 −k12 x1
+ 11 + =0
0 m2 ẍ2 −c21 c22 ẋ2 −k21 k22 x2
We had already proved earlier that based on eigen solution
⇒ ω1 = 4.35 rad/sec and ⇒ ω2 = 6.32 rad/sec.
The normalised eigen vector matrix for this case was calculated in the previous
example as
√ √
1/√2 1/ √2
[ϕ]n =
1/ 2 −1/ 2

N
[ϕ]T [M] {V0 }
Since {X} = [ϕ] e−Di ωni t {sin ωi t} and
ωdi
i=1

N
{X} = [ϕ] {ξi } we have
i=1

N
[ϕ]T [M] {V0 } −Di ωni t
{ξi } = e {sin ωi t}
ωdi
i=1
Thus for first mode we have

1 1 1 0 0 1
{ξ1 } = √ √ × √
2 2 0 1 0.5 4.5 (1−0.0225)
× e−0.675t × sin 4.449t
or {ξ1 } = 0.07946 · e−0.675t · sin 4.449t
For second mode we have

1 1 1 0 0
{ξ2 } = √ −√ × e−1.264t × sin 6.19t
2 2 0 1 0.5
or {ξ2 } = −0.0571168 e−1.264t sin 6.19t

Now since
√ √

N
1/√2 1/ √2
{X} = [ϕ] {ξi } we have, {X} =
1/ 2 −1/ 2
i=1

0.07946e−0.675t sin 4.49t
×
−0.0571168e−1264t sin 6.19t

0.0562e−0.675t sin 4.49t − 0.0404e−1.264t sin 6.19t
or {X} =
0.0562e−0.562t sin 4.49t + 0.0404e−1.264t sin 6.19t
The above equations when plotted at a time step of 0.05 sec, shows the history
as given in Figure 2.7.9.
Damped response of motion

0.08
0.06
0.04
Amplitude
x1
0.02
x2
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67
-0.02
-0.04
Time steps
Figure 2.7.9 Damped response for the two man system.
Observations:
• On studying the above mentioned plot you will see that the characteristic of the
curve is very similar to that plotted for the single degree of freedom.
• Also observe that the response is largely reduced compared to Example 2.7.2 due
to the consideration of damping in this case.
2.8 DESIGN OF HAMMER FOUNDATION
2.8.1 Design criteria for hammer foundation

Its time we emerge form the exotic world of Newtonian Mechanics and start looking
into the design parameters pertaining to hammer foundations.
In the design criteria we use the following nomenclatures
• Wh = weight of Hammer in kN;

• Wa = weight of the anvil in kN;
• Wfr = weight of hammer resting on foundation in kN;

• Wf = weight of the RCC foundation in kN;

• La × Ba = contact Area of the anvil in m2 ;
• L × B = contact are of the foundation in m2;
• Ea = Young’s Modulus of the elastic pad below the anvil in kN/m2 ;
• ta = thickness of the elastic pad in m;
• G = Dynamic shear modulus of the soil in kN/m2 ;
• g = acceleration due to gravity @ 9.81 m/sec2 ;
• Da = critical damping ratio of the elastic pad;
• Ds = critical damping ratio of the soil;
• p = pressure on piston of double acting hammer kN/m2 ;
• Ap = area of piston in m2 ;
• L = length of the stroke in m;
• α = correction factor @ 1.0 for well adjusted hammer and varies between 0.5 to
0.8 for double acting hammers. Usual design value taken is 0.65;
• k = coefficient of impact @ 0.5 for stamping hammers and 0.25 for forging
hammers;
• v = velocity of the hammer at the point of impact in m/sec;
• Va = velocity of anvil after the impact in m/sec;
• σs = allowable static bearing capacity of the soil in kN/m2 ; and
• σp = allowable stress of the elastic pad in kN/m2 .
2.8.1.1 Maximum permissible amplitudes for foundation

The maximum permissible amplitude of vibration for a hammer foundation shall not
exceed 1.2 mm.
2.8.1.2 Maximum permissible amplitudes for the anvil

The maximum permissible amplitude is usually dependent on the weight of the hammer
and is also at times prescribed by the vendor supplying the elastic pad who limits the
deflection based on the allowable stress of the elastic pad. Following Table 2.8.1
furnishes the allowable amplitude of the anvil based on hammer weight.
2.8.1.3 Minimum weight of foundation

The minimum weight of foundation is given by

Wa + Wfr
Wmin = Wh 8(1 + k)v −
Wh
Table 2.8.1
Weight of Max. permissible

Sl. No. hammer (KN) amplitude (mm)
1 30 3–4
2 20 2
3 Upto 10 1

2.8.1.4 Minimum base area of the foundation

The minimum base area of the foundation may be obtained from the following
expression
20(1 + k)
Amin = v · Wh
σs
2.8.1.5 Minimum thickness of foundation (Major 1980)

This is given in Table 2.8.2.
2.8.1.6 Velocity, V, of hammer at point of impact

• For free falling hammer

V = α 2gH
If energy of impact Ei is given by the manufacturer then
Ei
H= and V = 2gH
Wh
• For double acting hammer

2g(Wh + pAp )l
V = 0.65
Wh
2.8.1.7 Velocity of anvil after of impact
(1 + k)
V= v
Wa
1+ Wh
Table 2.8.2
Weight of Thickness of
Sl. No. hammer (kN) foundation (mm)
1 >60 >2250
2 60 2250
3 40 1750
4 20 1250
5 10 1000

2.8.1.8 Calculation of natural frequency based on IS-2974

ka
• Calculate ωa =
ma

kz
• Calculate ωz =
ma + mf + mfr
• Calculate ωn1 and ωn2 from the equation
ma
ωn4 − (ωa2 + ωz2 ) (1 + α) ωn2 + (1 + α)ωa2 ωz2 = 0; α=
mf + mfr

where, ka = EataAa and kz = (1−υ)
4Gr0
in which, r0 = L×B
π and, G = dynamic shear
modulus of the soil.
When supported on short bearing piles
kp · kz
kz =
kp + kz
where, kp = vertical stiffness of pile which may be obtained from formulas derived
W W
earlier and ma = Wg a , mf = g f and mfr = gfr .
2.8.1.9 Amplitude of vibration based on IS-2974

• The amplitude of the RCC foundation (x1 ) is given by

−(ωa2 − ωn2
2 )(ω2 − ω2 )
a n1 sin ωn1 t sin ωn2 t
x1 = V −
ωa2 (ωn1
2 − ω2 )f
n2 n2
ωn1 ωn2
• The amplitude of vibration of the anvil (x2 ) is given by

V (ωa2 − ωn2
2 ) sin ω t
n1 (ω2 − ωn1
2 ) sin ω t
n2
x2 = − a
(ωn1
2 − ω2 )
n2
ωn1 ωn2
2.8.1.10 Stability of pad between anvil and block

Total deflection of the pad under impact is given by
δtot = δst + δdyn
Wa + Wfr V
where δst = and δdyn =
Ka ωna

2.8.1.11 The loading intensity on the pad

The loading intensity on the pad is given by
Ka δtot
σp =
Aa
2.8.1.12 Stress in soil below foundation

The stress in soil below foundation is given by
Wa + Wf + Wfr + Kz ψ
σs =
Af
V̂ 1+k
in which, ψ = , where V̂ = V.
2π × fz 1+
Wf
Wa
2.8.2 Discussion on the IS-code method of analysis

The discussion should not be deemed as a criticism of the method, for normal
engineering design of standard hammer foundations the method advocated by the
code is adequate. But when a foundation subjected to loading of large magnitude
(heavy hammer >40 kN), the method proposed by the code may lead to conservative
and expensive design. Moreover if the environmental criteria calls for more stringent
restriction of amplitude or transmittal of waves, a designer may find it difficult to meet
the requirements based on IS-code.
The reasons attributable for the same may be summarized as follows:
• IS-code method does not take damping of the pad or that of the soil into considera-
tion. It has been observed that damping plays a very significant role in minimizing
the amplitude of vibration for such hammer foundation (Novak and El Hifnawy
1983).
• It also does not take into consideration the embedment effect which could play a
very significant role for heavy hammer foundation when the depth of the block
could be quite large.
• The IS-code formula of kz = 7.6Gr0 apparently looks overestimated29 .
• The dynamic displacement (δdyn ) is based on uncoupled form when the actual
response should based be coupled response. This could either under-estimate or
could also over estimate the stress induced in the foundation.
Based on the above the design procedure suggested herein may be structured as
follows.
29 For if we equate 4Gr0 /(1 − v) = 7.6Gr0 . We get v = 0.473 => 0.5. Poisson’s ratio @ 0.5 depicts
perfectly plastic clay which is rarely obtained. Value of v is usually taken as 0.4.

2.8.3 Check list for analysis of hammer foundation

Check if the following data regarding the machine and the foundation are available
with you
• Weight of the hammer.

• Type of hammer (free falling or double acting etc.).
• Sufficient data to calculate the velocity of hammer at point of collision.
• (like energy of impact, height of free fall etc.).
• Geometric dimension and weight of anvil.
• Geometric dimension, stiffness and damping property of the elastic pad supporting
the anvil.
• Anchoring detail of the elastic pad.
• Stiffness and damping data if mechanical springs and dampers are used in lieu of
elastic pads.
• Mechanical detail of the springs and dampers including their fixing detail as
suggested by the supplier.
• Anchoring detail of the frame supporting the hammer on the foundation.
• Top elevation of the anvil based on the mechanical process.
• Allowable bearing capacity of the soil.
• Dynamic shear modulus of the soil.
• Grade of concrete.
2.8.4 Other techniques of analysis of Hammer foundation

For most of the cases analysis as mentioned above suffice. However there are cases
where due to the massiveness of the foundation and hammer more detailed analysis is
envisaged where the effect of generated shock to its surrounding could be significant30 .
In such cases the best way to analyze such problem would be to resort to FEM
analysis of such foundations.
Shown in Figure 2.8.1 is a conceptual Finite element mathematical model of hammer
foundation housed inside a structural building. If the hammer foundation is very heavy,
though the hammer foundation itself may be within the acceptable limit of codal
stipulation can yet have adverse effect on the building in which it is located.
In such cases studying the problem based on FEM could be quit advantageous.
2.8.4.1 Selection of elements

• Here the anvil is modeled as a beam element having thee degrees of freedom
(horizontal, vertical and rotational degrees).
• The elastic pad below the anvil which is usually modeled as a spring can be
converted to equivalent truss element having stiffness @ AE/L for computer
implementation.
30 It has not been uncommon that the shocks generated by hammer foundation has done secondary damages
to the building in which it is placed or have rendered crane girders unserviceable at the Gantry level due
to distortion.

Anvil modeled as beam element Pad modeled

Lumped mass(Typ) as truss
element
Figure 2.8.1 Finite element model of hammer foundation.
• The block foundation itself can be modeled as a 2D plane stress element having
incompatible modes.
• The soil medium can be modeled as 2D plane strain element again having
incompatible modes.
• Finally the building and the foundation may be modeled as a plane frame
constituting of beam elements connected to the soil elements.

2.8.4.2 Boundary conditions

This is surely a major problem for failing to take a correct judgment can render the
analysis useless for waves generated due to the transient shock from the hammer may
get reflected from the boundary and generate spurious response of the foundation as
well as the building rendering the analysis useless.
We had already discussed earlier that for this type of infinite domain problem the
boundary should be extended far enough ensuring that no reflection of the waves take
place.
This on the other hand makes the analysis voluminous and also expensive in terms
of man-hours.
Moreover from practical point of view geo-technical data may not be available to
the depth required to ensure that no reflection takes place.
This can however be done by following techniques:
To provide spring and dash-pots having high damping value at the boundaries which
ensures that it absorbs all energy transmitted to it 31 .
Provide paraxial or quite boundaries to suppress the spurious modes32 .
Providing boundaries at a distance at least 2.5 times the length of the Rayleigh waves
to ensure that radiation damping is good enough to dissipate away the energy.
2.8.4.3 Material input

Following, material input shall be provided for the analysis:
• Dynamic Modulus of concrete for the RCC block.

• Dynamic modulus of the elastic pad.
• Dynamic shear modulus of the soil33 .
• Poisson’s ratio of the soil.
• Damping property of elastic pad below anvil (in this case the truss elements).
• Dynamic elastic modulus of the structure enabling the computer to generate the
stiffness matrix.
2.8.4.4 Input loading

The velocity of the anvil after impact shall be directly provided as input velocity for
the nodes of the beam element where masses are lumped.
2.8.4.5 Method of analysis

Due to the heterogeneous properties of the material we advocate here a time history
analysis having a time step of 0.1Tn where Tn is the least period of the system.
31 Unfortunately most of the commercially available FEM software does not have this feature of directly
inputting dash pots except ANSYS.
32 Refer to the Chapter 5 (Vol. 1) for further explanation.
33 If the soil is layered then for each layer a separate value of shear modulus has to be provided enabling
the computer to develop the material stiffness matrix for the plane strain element.

If the building system is found to have unacceptable amplitude ways and means
have to be sought to reduce the vibration transmitted to it by either providing air gaps
around the foundation or by providing suitable dampers around the foundation to
absorb this energy.
Example 2.8.1
A hammer foundation (Figure 2.8.2) having the following data has to be designed
for a particular site.
Frame Column(Typ.) 1400
2000
1400
1150 1250 1250 1150
1290
2150
Figure 2.8.2 General arrangement of the hammer foundation.

• Calculate the natural frequency of vibration

• The damped amplitude of vibration
• Compare the same with time history response
• Re-analyse the foundation based on IS-Code or Barkan’s mathod and
compare the results.
Design input data

1 Weight of hammer = 20 kN
2 Weight of frame = 85 kN
3 Plan area of anvil = 2.0 m × 2.5 m
4 Thickness of oak pad = 0.4 m
5 Height of free fall = 1.8 m
6 Unit weight of soil = 10 kN/m3
7 Dynamic shear modulus of soil = 120,0000 kN/m2
8 Poisson’s ratio = 0.40
9 Dynamic modulus of oak wood = 500,0000 kN/m2
10 Damping ratio for oak = 0.10
11 Plan area of foundation = 5 m × 5 m
12 Co-efficient of restitution = 0.65
13 Allowable bearing capacity of soil = 180 kN/m2
14 Air gap between anvil and foundation = 100 mm
Let height of the R.C.C. anvil be = 1250 mm
Weight of anvil (Wa ) = 2.5 × 2.0 × 1.25 × 25 = 156.25 kN
Weight of hammer = 20 kN √
Velocity of hammer at point of collision = 2gH = v = 2 × 9.81 × 1.80 =
5.94 m/sec
Velocity of Anvil after the impact
(1 + k) 1.65
V= v = 156.25
5.94 = 1.112 m/sec
1+ Wa 1+ 20
Wh
Minimum weight of foundation

Wa + Wfr
Wmin = Wh 8(1 + k)v −
Wh

156.25 + 85
= 20 8 × 1.65 × 5.94 − = 1327.16 kN
20
1327.16
Thus minimum depth required = = 2.12 m ∼
= 2.15 m(say)
5 × 5 × 25
Thus based on the sketch furnished earlier,

Weight of foundation = (5 × 5 × 3.44 − 2.7 × 2.2 × 1.29) × 25 =1958 kN.

Weight of the oak pad = 2.5 × 2.0 × 0.4 × 20 = 40 kN
Minimum base area required = Amin

20(1 + k) 20 × 1.65
= v · Wh = × 5.94 × 20 = 21.78 > 25 m2 OK
σs 180
Stiffness properties
EAp 500 × 104 × 5

For Pad stiffness (k2 ) = = = 625 × 104 kN/m
t 0.4
For Soil

25
Equivalent Radius r0 = = 2.82 m
π
4Gr0 4 × 120 × 104 × 2.82

kz = = = 22560000 kN/m
(1 − υ) 0.6
Damping properties
40
Mass of oak pad = = 4.0 kN sec2 /m
9.81
√
Cc = 2 km = 2 625 × 105 × 4 = 31623 kN · sec/m
√
C = D × 2 km = 0.1 × 31623 = 3162 kN · sec/m
1958 + 156 + 85
Mass of foundation and machine = = 224 kN · sec2 /m
9.81
0.25 (1 − υ) mg 0.25 × 0.6 × 224 × 9.81

Bz = = = 0.773
ρs r30 19 × (2.82)3
0.425
D= = 0.4832
Bz
Thus damping of the soil is given by

√ √
Cs = D × 2 km = 2 × 0.4832 × 22560000 × 224 = 68699 kN · sec/m
The mathematical model

The mathematical model perceived is given in Figure 2.8.3.

m2
x2
k2
c2
m1
x1
k1 c1
Figure 2.8.3 Mathematical model of the foundation.
The equation of motion for the above is given by

m1 0 ẍ1 C + C2 −C2 ẋ1
+ 1
0 m2 ẍ2 −C2 C2 ẋ2

k1 + k2 −k2 x1
+ =0
−k2 k2 x2
For free vibration we have

m1 0 ẍ1 k + k2 −k2 x1
+ 1 =0
0 m2 ẍ2 −k2 k2 x2

208 0 ẍ1 850.6 −625 x1
➔ + × 105 = 0
0 16 ẍ2 −625 625 x2
For eigen solution we have

85060000 − 208λ −62500000 x1
=0
−62500000 62500000 − 16λ x2

The above on expansion and simplification reduces to
λ2 − 4315192λ + 4.2367788 × 1011 = 0; → λ1 = 100525 and

λ2 = 4214667 ⇒ ω1 = 317 rad/sec and ω2 = 2052 rad/sec
Calculation of eigen vectors
For the first mode

85060000 − 20909200 −62500000 φ11
= 0;
−62500000 62496672 φ12
for φ11 = 1.0 we have φ12 = 1.0264
For the second mode

85060000 − 8.7665074 × 108 −62500000 φ21
= 0; for φ21 = 1.0
−62500000 −4934672 φ22
we have φ22 = −12.665
Thus the complete eigen vector is

1.0 1.0
[ϕ] =
1.0264 −12.665
Normalization of the eigen vectors
For the first mode

0 208
1.00
{φ}T [M] {φ} = 1.00 1.0264 = 224.856 and
16 1.0264 0
√
0.066688
Mr = 224.856 = 14.495 → {ϕ1 }N =
0.06844
For the second mode

T 208 0 1.00
{φ} [M] {φ} = 1.00 −12.665 = 2774 and
0 16 −12.665
√
0.01898
Mr = 2774 = 52.67 → {ϕ1 }N =
−0.240

0.0667 0.01898
Thus the complete normalized eigen vector is [ϕ]N = .
0.0684 −0.240

Damping matrix and its correction for de-coupling

c1 + c2 −c2 71861 −3162
The damping matrix is given by [C] = =
−c2 c2 −3162 3162
The above matrix on orthogonal transformation does not de-couple as such

we need to correct it.
For the first mode

71861 −3162 0.0667
{φ}T [C] {φ} = 0.0667 0.0684 = 305
−3162 3162 0.0684
→ 2D1 ω1 = 305 ➔ D1 = 0.480.
For the second mode

T 71861 −3162 0.01898
{φ} [C] {φ} =0.01898 0.0240 = 208
−3162 3162 0.240
→ 2D2 ω2 = 208 ➔ D2 = 0.050.
Based on Rayleigh damping we know that on orthogonal transformation the

expression should reduce to
2D1 ω1 = α + βω12 and 2D2 ω2 = α + βω22
or α + 100525β = 305 and α + 4214667β = 208
the equation on solving gives → α = 302.63 and β = 2.3577 × 10−5

Thus, based on Rayleigh damping

208 0
[C] = α [M] + β [K] → [C] = 302.63 + 2.3577 × 10−5
0 16

850.6 −625 64952 1474
× × 105 =
−625 625 1474 6316
Calculation of amplitude
The equation for damped amplitude is given by

N
[ϕ]T [M]{V0 }
{X} = [ϕ]e−Di ωnit {sin ωi t}
ωdi
i=1

Thus for the first mode we have

N
[ϕ]T [M]{V0 }
{ξi } = [ϕ]e−Di ωtni {sin ωi t}
ωdi
i=1

208 0 0 1
or ξ1 = 0.0667 0.0684 × √
0 16 1.112 317 (1 − 0.2304)
−0.48×317t
√
×e × sin(317 1 − 0.2304)t
→ ξ1 = 4.3776e−152.16t × sin 278t × 10−3

208 0 0
ξ2 = 0.01898 0.240
0 16 1.112
1 √
× √ × e−102.6t × sin(2052 1 − 0.0025)t
2052 (1 − 0.0025)
→ ξ2 = −2.08398e−102.6t × sin 2049t × 10−3
Since {X} = [ϕ] {ξ }, hence

0.0667 0.01898 4.3776e−152.16t sin 278t
{X} = × 10−3
0.0684 −0.240 −2.08398e−102.6t sin 2049t

0.29198e−152.16t sin 278t − 0.03955e−102.6t sin 2049t
or {X} = ×10−3 m.
0.299428e−152.16t sin 2049t + 0.50015e−102.6t sin 2049t
The above when plotted at a time step of 0.0005 seconds shows displacement
plots as depicted in Figure 2.8.4.
Displacement History of Hammer Foundation

0.6 with time step of 0.0005 seconds
0.5
Amplitude (mm)
0.4
0.3 Displacement amplitude
0.2 of foundation (mm)
0 of anvil (mm)
-0.1 1 14 27 40 53 66 79 92 105 118 131 144
-0.2
-0.3 Time steps
Figure 2.8.4 Displacement history of foundation and the anvil.

Time History analysis based on Newmark-beta method

Here we do time history analysis of the equation

208 0 ẍ1 71861 −3162 ẋ1
+
0 16 ẍ2 −3162 3162 ẋ2

85060000 −62500000 x1
+ =0
−62500000 62600000 x2
having boundary condition at t = 0 v = 1.112 m/sec. Solution is given in

Figure 2.8.5.
Time history plot of Hammer Foundation

(Newmark-beta Method)
0.0006
0.0004
Amplitude(meter)
0.0002
Amplitude of fdn.
0 Amplitude of anvil
1 14 27 40 53 66 79 92 105 118 131 144
-0.0002
-0.0004
Time steps
Figure 2.8.5 Time history response of amplitude for foundation and the anvil.
It is to be noted that we use here the original non proportional damping matrix
and not the corrected one used above. we give the following results for 98 steps
(explanation in next page).
Next we compare the response of the foundation and the anvil separately to see
what is the variation in the results. The results are as plotted in Figures 2.8.6 to 7.
Comparison of amplitude of foundation

based on Time History and Closed form
0.2
0.15
Amplitude(mm)
Amplitude of foundation
0.1 based on Newmark Method
of foundation(mm)
0
1 15 29 43 57 71 85 99 113 127 141
-0.05
Time steps
Figure 2.8.6 Comparison of response, time history versus approximate damping.

Sl. Time
No. step x1 (disp) x1 (vel) x1 (acc) x2 (disp) x2 (vel) x2 (acc)
0 0 0 0 0 0.00 × 10+00 1.112 0

1 0.0005 8.03×10−06 3.21 × 10−02 1.28 × 10+02 4.42 × 10−04 0.65707 −1819.7
2 0.001 3.90 × 10−05 9.18 × 10−02 1.10 × 10+02 5.39 × 10−04 −0.2687 −1883.2
3 0.0015 9.10 × 10−05 1.16 × 10−01 −1.27 × 10+01 2.58 × 10−04 −0.855 −462.04
4 0.002 1.40 × 10−04 7.99 × 10−02 −1.32 × 10+02 −1.24 × 10−04 −0.6752 1181.16
5 0.0025 1.62 × 10−04 8.62 × 10−03 −1.53 × 10+02 −2.80 × 10−04 0.05021 1720.53
6 0.003 1.53 × 10−04 −4.58 × 10−02 −6.51 × 10+01 −9.60 × 10−05 0.68719 827.405
7 0.0035 1.29 × 10−04 −4.89 × 10−02 5.27 × 10+01 2.58 × 10−04 0.72961 −657.74
8 0.004 1.15 × 10−04 −9.27 × 10−03 1.06 × 10+02 4.88 × 10−04 0.19049 −1498.7
9 0.0045 1.20 × 10−04 3.21 × 10−02 5.96 × 10+01 4.22 × 10−04 −0.4548 −1082.6
10 0.005 1.38 × 10−04 3.73 × 10−02 −3.87 × 10+01 1.36 × 10−04 −0.6883 148.758
11 0.0055 1.47 × 10−04 1.53 × 10−03 −1.05 × 10+02 −1.27 × 10−04 −0.365 1144.36
12 0.006 1.36 × 10−04 −4.63 × 10−02 −8.68×10+01 −1.67×10−04 0.20483 1135
13 0.0065 1.07 × 10−04 −6.94 × 10−02 −5.62 × 10+00 2.01 × 10−05 0.54351 219.711
14 0.007 7.67 × 10−05 −5.33 × 10−02 7.01 × 10+01 2.56 × 10−04 0.40073 −790.8
15 0.0075 5.95 × 10−05 −1.53 × 10−02 8.19 × 10+01 3.39 × 10−04 −0.0677 −1082.9
16 0.008 5.87 × 10−05 1.20 × 10−02 2.72 × 10+01 2.07 × 10−04 −0.4603 −487.52
17 0.0085 6.37 × 10−05 7.88 × 10−03 −4.36 × 10+01 −2.55 × 10−05 −0.4714 443.13
18 0.009 6.04 × 10−05 −2.10 × 10−02 −7.19 × 10+01 −1.75 × 10−04 −0.1257 939.469
19 0.0095 4.30 × 10−05 −4.85 × 10−02 −3.83 × 10+01 −1.39 × 10−04 0.27068 646.244
20 0.01 1.79 × 10−05 −5.18 × 10−02 2.53 × 10+01 2.89 × 10−05 0.39924 −132.01
21 0.0105 −2.27 × 10−06 −2.89 × 10−02 6.61 × 10+01 1.74 × 10−04 0.18314 −732.36
22 0.011 −9.23 × 10−06 1.09 × 10−03 5.39 × 10+01 1.77 × 10−04 −0.1732 −692.94
23 0.0115 −5.11 × 10−06 1.54 × 10−02 3.34 × 10+00 4.07 × 10−05 −0.372 −102.31
24 0.012 1.84 × 10−07 5.76 × 10−03 −4.19 × 10+01 −1.19 × 10−04 −0.2675 520.295
25 0.0125 −2.46 × 10−06 −1.63 × 10−02 −4.64 × 10+01 −1.78 × 10−04 0.03175 676.689
26 0.013 −1.42 × 10−05 −3.05 × 10−02 −1.02 × 10+01 −1.02 × 10−04 0.27195 284.117
27 0.0135 −2.79 × 10−05 −2.46 × 10−02 3.39 × 10+01 3.30 × 10−05 0.26895 −296.12
28 0.014 −3.50 × 10−05 −3.58 × 10−03 5.00 × 10+01 1.12 × 10−04 0.04846 −585.85
29 0.0145 −3.19 × 10−05 1.58 × 10−02 2.75 × 10+01 7.62 × 10−05 −0.1932 −380.92
30 0.015 −2.31 × 10−05 1.96 × 10−02 −1.25 × 10+01 −3.73 × 10−05 −0.2607 111.02
31 0.0155 −1.64 × 10−05 7.21 × 10−03 −3.70 × 10+01 −1.31 × 10−04 −0.1148 472.61
32 0.016 −1.68 × 10−05 −9.05 × 10−03 −2.81 × 10+01 −1.32 × 10−04 0.11024 427.577
33 0.0165 −2.29 × 10−05 −1.51 × 10−02 3.79 × 10+00 −4.75 × 10−05 0.22911 47.8826
34 0.017 −2.83 × 10−05 −6.44 × 10−03 3.09 × 10+01 4.91 × 10−05 0.1574 −334.73
35 0.0175 −2.76 × 10−05 9.28 × 10−03 3.19 × 10+01 8.09 × 10−05 −0.0303 −415.92
36 0.018 −2.04 × 10−05 1.93 × 10−02 8.03 × 10+00 2.99 × 10−05 −0.1738 −158.38
37 0.0185 −1.15 × 10−05 1.63 × 10−02 −1.97 × 10+01 −5.43 × 10−05 −0.1628 202.42
38 0.019 −6.40 × 10−06 4.13 × 10−03 −2.91 × 10+01 −9.99 × 10−05 −0.0197 369.994
39 0.0195 −7.06 × 10−06 −6.79 × 10−03 −1.45 × 10+01 −7.24 × 10−05 0.12982 228.176
40 0.02 −1.07 × 10−05 −7.89 × 10−03 1.01 × 10+01 1.62 × 10−06 0.1662 −82.648
41 0.0205 −1.25 × 10−05 6.95 × 10−04 2.42 × 10+01 6.08 × 10−05 0.07049 −300.2
42 0.021 −9.58 × 10−06 1.11 × 10−02 1.74 × 10+01 6.10 × 10−05 −0.0695 −259.89
43 0.0215 −3.15 × 10−06 1.46 × 10−02 −3.19 × 10+00 8.95 × 10−06 −0.1388 −16.976
44 0.022 2.72 × 10−06 8.86 × 10−03 −1.99 × 10+01 −4.79 × 10−05 −0.0887 217.091
45 0.0225 4.66 × 10−06 −1.10 × 10−03 −1.99 × 10+01 −6.27 × 10−05 0.02976 256.868
46 0.023 2.58 × 10−06 −7.24 × 10−03 −4.67 × 10+00 −2.62 × 10−05 0.11601 88.1138
47 0.0235 −5.62 × 10−07 −5.33 × 10−03 1.23 × 10+01 2.88 × 10−05 0.10398 −136.24
(Continued)

Sl. Time
No. step x1 (disp) x1 (vel) x1 (acc) x2 (disp) x2 (vel) x2 (acc)
48 0.024 −1.37 × 10−06 2.09 × 10−03 1.74 × 10+01 5.77 × 10−05 0.01174 −232.72
49 0.0245 1.23 × 10−06 8.32 × 10−03 7.58 × 10+00 4.05 × 10−05 −0.0804 −136.01
50 0.025 5.37 × 10−06 8.24 × 10−03 −7.90 × 10+00 −4.45 × 10−06 −0.0995 59.672
51 0.0255 7.99 × 10−06 2.22 × 10−03 −1.62 × 10+01 −3.86 × 10−05 −0.0371 189.843
52 0.026 7.37 × 10−06 −4.66 × 10−03 −1.13 × 10+01 −3.55 × 10−05 0.04953 156.874
53 0.0265 4.44 × 10−06 −7.06 × 10−03 1.72 × 10+00 −8.53 × 10−07 0.08917 1.66764
54 0.027 1.76 × 10−06 −3.66 × 10−03 1.19 × 10+01 3.50 × 10−05 0.05426 −141.3
55 0.0275 1.38 × 10−06 2.16 × 10−03 1.14 × 10+01 4.33 × 10−05 −0.0209 −159.34
56 0.028 3.28 × 10−06 5.42 × 10−03 1.64 × 10+00 1.99 × 10−05 −0.073 −49.244
57 0.0285 5.54 × 10−06 3.65 × 10−03 −8.73 × 10+00 −1.41 × 10−05 −0.0629 89.9718
58 0.029 6.11 × 10−06 −1.38 × 10−03 −1.14 × 10+01 −3.09 × 10−05 −0.0041 145.004
59 0.0295 4.41 × 10−06 −5.42 × 10−03 −4.82 × 10+00 −1.89 × 10−05 0.05205 79.6655
60 0.03 1.71 × 10−06 −5.39 × 10−03 4.92 × 10+00 9.41 × 10−06 0.06115 −43.237
61 0.0305 −6.34 × 10−08 −1.70 × 10−03 9.86 × 10+00 2.97 × 10−05 0.02016 −120.74
62 0.031 1.10 × 10−07 2.39 × 10−03 6.52 × 10+00 2.63 × 10−05 −0.0338 −95.228
63 0.0315 1.61 × 10−06 3.61 × 10−03 −1.66 × 10+00 3.69 × 10−06 −0.0567 3.79594
64 0.032 2.83 × 10−06 1.26 × 10−03 −7.72 × 10+00 −1.87 × 10−05 −0.033 90.9929
65 0.0325 2.54 × 10−06 −2.42 × 10−03 −7.02 × 10+00 −2.34 × 10−05 0.01427 98.0753
66 0.033 8.48 × 10−07 −4.35 × 10−03 −7.01 × 10−01 −8.48 × 10−06 0.04544 26.603
67 0.0335 −1.01 × 10−06 −3.10 × 10−03 5.71 × 10+00 1.22 × 10−05 0.03721 −59.519
68 0.034 −1.76 × 10−06 1.08 × 10−04 7.11 × 10+00 2.14 × 10−05 −0.0003 −90.463
69 0.0345 −1.09 × 10−06 2.60 × 10−03 2.84 × 10+00 1.27 × 10−05 −0.0345 −46.558
70 0.035 1.90 × 10−07 2.51 × 10−03 −3.20 × 10+00 −5.56 × 10−06 −0.0385 30.5594
71 0.0355 8.65 × 10−07 1.95 × 10−04 −6.05 × 10+00 −1.81 × 10−05 −0.0118 76.5631
72 0.036 3.52 × 10−07 −2.25 × 10−03 −3.73 × 10+00 −1.56 × 10−05 0.02179 57.6405
73 0.0365 −9.15 × 10−07 −2.82 × 10−03 1.44 × 10+00 −1.45 × 10−06 0.03487 −5.3432
74 0.037 −1.92 × 10−06 −1.19 × 10−03 5.09 × 10+00 1.20 × 10−05 0.01895 −58.344
75 0.0375 −1.92 × 10−06 1.19 × 10−03 4.44 × 10+00 1.41 × 10−05 −0.0107 −60.105
76 0.038 −1.02 × 10−06 2.40 × 10−03 3.82 × 10−01 4.12 × 10−06 −0.0291 −13.827
77 0.0385 −1.79 × 10−08 1.60 × 10−03 −3.56 × 10+00 −8.86 × 10−06 −0.0228 39.3595
78 0.039 2.96 × 10−07 −3.50 × 10−04 −4.26 × 10+00 −1.43 × 10−05 0.0012 56.5166
79 0.0395 −2.38 × 10−07 −1.79 × 10−03 −1.48 × 10+00 −8.42 × 10−06 0.02214 27.2158
80 0.04 −1.08 × 10−06 −1.59 × 10−03 2.26 × 10+00 3.04 × 10−06 0.02367 −21.079
81 0.0405 −1.49 × 10−06 −5.06 × 10−05 3.89 × 10+00 1.05 × 10−05 0.00634 −48.257
82 0.041 −1.13 × 10−06 1.49 × 10−03 2.29 × 10+00 8.53 × 10−06 −0.0144 −34.597
83 0.0415 −3.02 × 10−07 1.82 × 10−03 −9.82 × 10−01 −4.86 × 10−07 −0.0217 5.36525
84 0.042 3.49 × 10−07 7.82 × 10−04 −3.18 × 10+00 −8.65 × 10−06 −0.011 37.4775
85 0.0425 3.75 × 10−07 −6.77 × 10−04 −2.66 × 10+00 −9.49 × 10−06 0.00762 36.8928
86 0.043 −1.34 × 10−07 −1.36 × 10−03 −7.52 × 10−02 −2.94 × 10−06 0.01859 7.00279
87 0.0435 −6.73 × 10−07 −7.98 × 10−04 2.32 × 10+00 5.19 × 10−06 0.0139 −25.792
88 0.044 −7.62 × 10−07 4.42 × 10−04 2.64 × 10+00 8.33 × 10−06 −0.0013 −35.146
89 0.0445 −3.25 × 10−07 1.31 × 10−03 8.13 × 10−01 4.48 × 10−06 −0.0141 −15.722
90 0.045 2.84 × 10−07 1.13 × 10−03 −1.52 × 10+00 −2.63 × 10−06 −0.0144 14.4488
91 0.0455 6.01 × 10−07 1.39 × 10−04 −2.45 × 10+00 −7.01 × 10−06 −0.0032 30.4011
92 0.046 4.31 × 10−07 −8.16 × 10−04 −1.37 × 10+00 −5.40 × 10−06 0.00962 20.7196
93 0.0465 −1.97 × 10−08 −9.89 × 10−04 6.82 × 10−01 4.16 × 10−07 0.01365 −4.5961
94 0.047 −3.47 × 10−07 −3.20 × 10−04 1.99 × 10+00 5.45 × 10−06 0.0065 −24.005
95 0.0475 −2.84 × 10−07 5.72 × 10−04 1.58 × 10+00 5.79 × 10−06 −0.0052 −22.597
96 0.048 9.65 × 10−08 9.51 × 10−04 −6.63 × 10−02 1.59 × 10−06 −0.0116 −3.3553
97 0.0485 4.72 × 10−07 5.52 × 10−04 −1.53 × 10+00 −3.39 × 10−06 −0.0083 16.8157
98 0.049 5.49 × 10−07 −2.43 × 10−04 −1.66 × 10+00 −5.11 × 10−06 0.00138 21.7888

Comparison of amplitude of the anvil

based on Time History and Closed form
0.6
0.5
0.4
Amplitude(mm)
0.3
0.2 Amplitude of anvil based
0.1 on Newmark Method
0 Displacement amplitude
-0.1 1 16 31 46 61 76 91 106 121 136 of anvil(mm)
-0.2
-0.3
-0.4
Time steps
Figure 2.8.7 Comparison of response time history versus approximate damping.
The above displacements plot show some very interesting results
• The closed form solution and the time history results are very closely
matching
• Since the time history is done with the original soil damping matrix and com-
pared with modal response having damping matrix corrected to Rayleigh
format (and yet it gives reasonably good results), it may be concluded that
the technique of separating the damping for each mode and correcting the
damping matrix based on Rayleigh damping may well be adapted without
any significant error in cases where the damping matrix is non-proportional.
Analysis based on Code

ka 62500000
ωa = = = 1979 rad/sec
ma 16

kz 2560000 × 9.81
ωz = = = 317 rad/sec
ma + mf + mfr 156 + 1958 + 85
ma 156
α= = = 0.076
mf + mfr 1958 + 85
Equation for natural frequency is given by
ωn4 − (ωa2 + ωz2 )(1 + α)ωn2 + (1 + α)ωa2 ωz2 = 0
Substituting the numerical values calculated above, we have
ωn4 − 4322216.7ωn2 + 4.2346974 × 1011 = 0.

The above on solving gives, ωn12 = 100302.9 → ω

n1 = 316.7 rad/sec and
ωn2
2 = 4221913.9 → ωn2 = 2054 rad/sec.
The amplitude of vibration is given by the formula

−(ωa2 − ωn2
2 )(ω2 − ω2 )
a n1 sin ωn1 t sin ωn2 t
x1 = V − for the foundation and,
ωa2 (ωn1
2 − ω2 )f
n2 n2
ωn1 ωn2

V (ωa2 − ωn2
2 ) sin ω t
n1 (ωa2 − ωn1
2 ) sin ω t
n2
x2 = 2 − for the anvil.
(ωn1 − ωn22 ) ωn1 ωn2
Substituting the numerical values as calculated above and plotting @ 0.0005

sec time steps we time history plot for this case as shown in Figures 2.8.8 and 9.
Undamped response of hammer foundation

Amplitude of foundation block
0.4
0.3
0.2
0.1
0 Amplitude of foundation
1
13
25
37
49
61
73
85
97
109
121
133
145
-0.1
-0.2
-0.3
-0.4 Time Steps
Figure 2.8.8 Amplitude of foundation as per IS-code.
1 Undamped response of anvil

0.8
Amplitude of anvil(mm)
0.6
0.4
0.2
0 Amplitude of anvil
1
11
21
31
41
51
61
71
81
91
101
111
121
131
141
-0.2
-0.4
-0.6
-0.8
-1 Time steps
Figure 2.8.9 Amplitude of anvil as per IS-code.
The maximum amplitude of vibration is 0.284 mm in lieu of 0.156 mm for

damped case. This proves that without considering the effect of damping the
designer may have to make the foundation more heavy (thus expensive) to reduce
the amplitude if the value exceeds the acceptable limit as shown in Figure 2.8.9.
In this case the undamped amplitude of the anvil is 0.754 mm in lieu of
0.471 mm when damping is considered.

We give below a compare the various data as calculated above.

Comparison of results of the hammer foundation by various techniques:
Close Close Time history

formed formed response of
Engineering (undamped (damped the damped
parameter solution) Solution) equation Remarks
1st Natural 316.7 317 317 Practically no difference

frequency
2nd Natural 2054 2052 2052 Practically no difference
frequency
Amplitude of 0.754 0.471 0.539 Significant difference between un-
anvil (mm) damped and damped amplitude but
marginal difference with time history
Amplitude of 0.286 0.156 0.1622 Comments same as above
foundation (mm)
2.9 DESIGN OF ECCENTRICALLY LOADED HAMMER FOUNDATION
2.9.1 Mathematical formulation of anvil placed

eccentrically on a foundation
Sometimes due to the functional requirements anvils may be placed eccentric to the
RCC block as shown in Figure 2.9.1.
In this case the hammer hits the anvil concentrically, but as the anvil is placed at
an eccentricity e mm, say, the foundation block other than vertical mode also gets
subjected to a coupled horizontal and rocking mode.
The mathematical model used for such case is as shown in Figure 2.9.2.
Here additional sliding (x) and rocking mode (θ ) is also simulated due to the eccentric
impact.
Hammer(Wh)
e Anvil(Wa)
Foundation(Wf)
Figure 2.9.1 Eccentrically loaded anvil.

m2
z2
k2 c2
Cx
Cθ kθ m1 z1
kx
c1
k1
Figure 2.9.2 Mathematical model of eccentrically loaded anvil.
The nomenclatures used here are:
m1 = mass of the foundation block plus frame resting on it if any;

m2 = mass of the anvil resting on elastic pad;
k1 = soil spring to be calculated from Barkan or Richart’s formula34 ;
k2 = spring value for the elastic pad, this value is normally furnished by the vendor
supplying these pads;
kx = soil spring to be calculated from Barkan or Richart’s formula in x direction;
kθ = soil spring to be calculated from Barkan or Richart’s formula in θ direction;
c2 = damping of the elastic pad again furnished by the vendor;
cx = damping of the soil to be obtained from Richart’s formula in x direction;
cθ = damping of the soil to be obtained from Richart’s formula in θ direction;
z2 = amplitude vector of the anvil in vertical direction;
z1 = amplitude vector of the Foundation in vertical direction;
x = amplitude vector of the Foundation in horizontal direction;
θ = rotational amplitude of the foundation, and
e = eccentricity of the anvil with respect to the c.g. of the foundation.
34 Refer to section on block foundation for the formula of the springs and dampers.

Based on d’Alembert’s equation and the free body diagram the equation of motion
for vibration are as follows:
m2 z̈2 + k2 (z2 − z1 ) = 0
m1 z̈2 + k1 (z1 − θe) + k2 (z1 − z2 ) = 0
m1 ẍ + kx (x − Zc θ ) = 0
J θ̈ − kx Zc x + [kθ − Wf Zc + k1 e2 + kx Zc2 ]θ − k1 ez1 = 0 (2.9.1)
The above when written in matrix form gives the equation

⎡ ⎤⎧ ⎫
m1 0 0 0 ⎪ z̈1 ⎪
⎪
⎨ ẍ ⎪
⎬
⎢0 m1 0 0⎥
⎢ ⎥
⎣0 0 m2 0⎦ ⎪ z̈2 ⎪
⎪
⎩ θ̈ ⎪
⎭
0 0 0 J
⎡ ⎤⎧ ⎫
k1 + k2 0 −k2 −k1 e ⎪
⎪z1 ⎪
⎪
⎢ 0 kx 0 −kx Zc ⎥⎨x⎬
+⎢
⎣ −k2
⎥
⎦ ⎪z2 ⎪ = {0}
0 k2 0 ⎪
⎩ ⎪ ⎭
−k1 e −kx Zc 0 (kθ − Wf Zc + k1 e + kx Zc )
2 2 θ
(2.9.2)
2.9.2 Damped equation of motion with eccentric anvil

The damped equation of motion is given by
[M]{Ẍ} + [C]{Ẋ} + [K]{X} = 0 (2.9.3)
Here,
[M] = a square mass matrix of the order 4 × 4 as shown above

[K] = a square stiffness matrix of the order 4 × 4 as shown above
{X} = a Column deflection matrix of order 4 × 1 (which means 4 rows and 1 column)
[C] = a square damping matrix of order 4 × 4.
The equation being statically coupled, the damping matrix is given by
⎡ ⎤
c1 + c2 0 −c2 −c1 e
⎢ 0 cx 0 −cx Zc ⎥
[C] = ⎢
⎣ −c2
⎥
⎦ (2.9.4)
0 c2 0
−c1 e −cx Zc 0 (cθ − Wf Zc + c1 e2 + cx Zc2 )
Based on the above matrices one can now do the analysis in identical fashion as
shown earlier35 once the initial velocity of the anvil after the impact is known.
35 Here the order of matrices being 4 × 4 eigen solution may be done by Bairstow’s method or one can
directly solve for them in solution tools in computer like MATHCAD or MATLAB etc.

Figure 2.10.1 Typical reinforcement detail for hammer foundation.
2.10 DETAILS OF DESIGN
2.10.1 Reinforcement detailing

Usually for this type of foundation moments induced and reinforcement required for
the same are nominal and nominal reinforcement is only provided.
Usually 16 mm diameter bar is placed @ 150/200 mm c/c subject to the minimum
condition that the volume of reinforcement shall not be less than 25 Kg/m3 .
In this type of foundations, the reinforcement shall be placed along the three axis
and also diagonally at the corner to prevent shear failure.
Additional reinforcement shall be provided at the top side of the foundation block
than at other sides.
Topmost layer of reinforcement shall be provided with a cover not less than 50 mm.
The sketch in Figure 2.10.1 shows a typical detail of re-bars for hammer foundations.
2.10.2 Construction procedure

The foundation block should preferably be cast in one go. If this is not possible and
a construction joint is unavoidable then such joints shall be horizontal in orientation
and measures shall be taken to provide a proper joint by providing dowels of 12 mm
or 16 mm dia bars embedded at 60 mm center to center to depth of at least 300 mm
at both sides of the joint.
Before placing the next layer of concrete the previously laid layer of concrete should
be roughened, cleaned thoroughly and washed by water jet and then covered by a layer
of rich 1:2 cement grout (1 cement, 2 sand) at least 20 mm thick. Concrete should be
placed not later than 2 hours after the grout is laid.

2.11 VIBRATION MEASURING INSTRUMENTS
2.11.1 Some background on vibration measuring

instruments and their application
Design of machine foundation is not only a case of design of foundation and restricting
the amplitude of vibration within the acceptable limit. In many cases if there are
sensitive instruments around the foundation a foundation specialist is required to
measure the vibration amplitude of the surrounding instruments (even though the
foundation itself is safe) to ensure their safe operation. There could be properties
or heritage buildings around the foundations that would require protection from the
foundation vibration. In such cases an important function is to measure the vibration
induced by the machine and its effect on the surrounding. This is usually done by
instruments which are generally termed as vibration pick up instruments.
In this section we will see what the theoretical background for development of these
machines is and how they are used to measure such vibrations.
2.11.2 Response due to motion of the support

In many situations vibration of a system is not due to forces acting directly on the
mass but resulting from the motion of the base. Consider the situations shown in
Figure 2.11.1.
Figure 2.11.1(a) shows the basics of a ground-measuring device. A vibration meter
shown therein measures only a relative displacement of the ground. The relative motion
is usually converted to an electric voltage by making the seismic mass a magnet mov-
ing relative to coils fixed in the case shown in Figures 2.11.1(b) and (c). The electric
voltage produced is a measure of the ground displacement. These types of instruments
are called velocity meters. The voltage generated is proportional to the rate of cutting
of the magnetic field and the output of the system will be proportional to the velocity
of the vibrating body. If a rotating drum is fixed and a needle is moving on a drum,
the relative motion of the instrument-soil system will be recorded on the drum [rela-
tive displacement = [z2 (t) −z1 (t)], z2 (t) being the displacement of the mass]. A basic
description of a seismic pick up is shown in Figure 2.11.1(c). The relative displacement
is the e. m. f. produced in the coil is the electrical signal from it mechanical counter-
part. A hypothetical pick up is shown in Figure 2.11.1(d) wherein a magnetic material
moves up and down in a electro-magnetic field and the electrical signal produced is a
measure of the actual ground displacement.
2.11.3 Vibration pick-ups

The vibration-displacement amplitudes are most often measured in soils and founda-
tions ranging from millionths to thousandths of a centimetre and occur at frequencies
ranging from less than 10 Hz to more than 100 Hz. The instruments required to mea-
sure motions of this magnitude are designed on the basis of a single-degree-of-freedom
system. Instruments based on this design have two distinct advantages. First, in the
S.D.O.F. system, the suspended mass is used as a reference from which vibrations
are measured because in cases such as ground-motion measurements no reference is
available. The second, some electrical phenomena are readily adapted to measuring

Rotating drum
Vibration meter z2(t)
m z2(t) m
z2-z1
z2-z1
z1(t) z1(t)
(a). General set up (b). General set up of vibration pick up
Electromagnetic field
Rotating drum
z2(t)
m z2-z1
Displacement here is same as that of the ground;

i.e. the actual displacement is measured
z1(t)
(c). Seismic pick up (d). Hypothetical ground displacement measuring device
Figure 2.11.1 Seismic pick up.
z2(t)
m
k c
z1(t)
Figure 2.11.2 Displacement pick up.
the response of the system by producing an electrical signal that can be observed with
an oscilloscope or recorded for subsequent analysis.
An instrument that converts mechanical motion into an electrical signal is called a
transducer. For vibration measurements there are three general types of transducers,
namely, velocity, acceleration and displacement transducers.
2.11.3.1 Displacement transducer

In Figure 2.11.2 a schematic sketch of a displacement transducer is shown. Here we
record Zr (t) = z2 (t) − z1 (t).
Now if it so happens that
Zr (t) z2 (t)
⇒1 and 1
z1 (t) z1 (t)

then whatever Zr we measure will be the z1 (t).

If ż2 (t) and z̈2 (t) are velocities and accelerations of the mass, one can write the
equation of motion of the system as
m z̈2 + c(ż2 − ż1 ) + k(z2 − z1 ) = 0 (2.11.1)
Again, let
Z(t) = z2 (t) − z1 (t), so Ż(t) = ż2 (t) − ż1 (t),

and Z̈(t) = z̈2 (t) − z̈1 (t), also, z̈2 (t) = Z̈(t) + z̈1 (t).
Substituting above, one can write:
mZ̈(t) + cŻ(t) + kZ(t) = −mz̈1 (t) (2.11.2)
If we assume a ground motion of the type: z1 (t) = A sin ωt

⇒ mz̈1 (t) = −mA ω2 sin ωt: Substituting the above, Equation (2.11.2) reduces to:
mZ̈(t) + cŻ(t) + kZ(t) = mA ω2 sin ωt (2.11.3)
Solution of Equation (2.11.3) is
mAω2 /k
Z= sin (ωt − φ) = r2 κA sin(ωt − φ) = X sin(ωt − φ)
(1 − r2 )2 + (2Dr)2
2Dr 1
and tan φ = where κ = (2.11.4)
1 − r2 (1 − r2 )2 + (2Dr)2
Solution of Equation (2.10.4) is shown in Figure 2.11.3. It shows that when r is

very high, Zmax /A approaches one. That is if we choose pick-ups having a very low
ωn (natural frequency of m-k system), r will be very high. This will result in Z ≡ A.
Also the phase angle φ, between the exciting force (ground motion) and the instrument
should be nearly zero or 180◦ . Figure 2.11.3 shows that φ is nearly 180◦ for large values
of r. Regarding D of the transducer, we see that the curve with D = 0.6 is better suited
as the amplitude is not amplified near the natural frequency and secondly Zmax /A
reaches unity faster.
2.11.3.2 Instrument with low natural frequency

Instruments with low natural frequency, the r-values approach a large value and the
relative displacement Z approaches A, regardless of the damping, D [Figure 2.11.3].
The mass here remains stationary while the supporting case moves with the vibrating
body. The instrument just described is the basis of what is known as seismometer.
These instruments are of large size as the relative motion of the seismic mass must
be of the same order of magnitude as that of the vibration to be measured. Since the

D = 0.05
D=0
180
8 D = 0.0
140
Phase angle,
90
6 0.707
40
Zmax/A
D = 0.4
accelerometer
0
0 0.5 1 1.5 2 2.5 3
4
Frequency ratio, r
D = 0.6 D = 0.707
1
seismometer
0 1 2 4 6 8
0.2
Frequency ratio, r
Figure 2.11.3 Response curves.
seismic mass (m as shown in Figure 2.11.2) is a magnet moving relative to coils fixed
in the case, the voltage generated is proportional to the velocity of the vibrating body.
The displacement and acceleration can be obtained from this velocity type transducer
through integrator and differentiator provided in most signal-recording units.
2.11.3.3 Instrument with high natural frequency

Instruments with high natural frequency, the r-values approach a very low value
and the relative displacement Z approaches A, regardless of the damping, D [Figure
2.11.3]. Again, the denominator of Equation (2.11.4),

i.e. (1 − r2 )2 + (2Dr)2 approaches unity as r → 0.
Under this condition, Equation (2.10.4)
ω2 A Acceleration
➔ Zmax = Ar2 = = ,
ωn
2 ωn2
implying that Zmax is now proportional to the acceleration of the motion, 1/ωn2 being
the constant of proportionality. Range of accelerometer can be seen from a magni-
fied plot of √ 1
for various values of D. For D = 0.7, the useful range is
2 2 2 (1−r ) +(2Dr)

1.04
D=0
1.02 D = 0.6
1 1
(1- r2)2 + (2Dr)2
0.98 D = 0.65
D = 0.75
0.96
D = 0.7
0 0.2 0.4 0.6 0.8 1

Frequency ratio, r
Figure 2.11.4 Acceleration error with varying frequency for various D.
0 ≤ D ≤ 0.2 with a maximum error of 0.01 percent (Figure 2.11.4). Electromagnetic

type of accelerometers generally utilizes a damping around D = 0.7, which not only
extends the useful frequency range but also prevents phase distortion for complex
waves.
2.11.3.4 Velocity transducers

The displacement transducer described in the preceding may also be used as a
velocity transducer. The relation between the relative velocity
amplitude and the
ground-velocity amplitude is identical to Zmax /A = r2 / (1 − r2 )2 + (2Dr)2 , since
multiplication of Zmax and A by ω leaves the equation unchanged.
Output from a velocity transducer is generated by coil moving through a magnetic
field as mentioned earlier (Figure 2.11.5). Since voltage induced in the coil is directly
proportional to the relative velocity between the coil and the magnetic field, either the
coil or the magnet is made part of the mass and the other component is attached to
the frame.
2Di ω 2Dω
ωni ωn
Phase angles : tan φ1 = 2 : tan φ = 2 (2.11.5)
ω
1− ωni 1 − ωωn
in which ω = operating frequency, frequency of the forcing function (say 50 Hz);

ωni = natural frequency of the instrument (say 4.75 Hz); ωn = natural frequency of
(soil mass + footing); Di = Damping coefficient of the instrument; D = damping
coefficient of (soil mass + footing).
For a soil-footing vibration system:
Phase angles for the above system (Figure 2.11.6) may be written as
2Di ω/ωni 2Dω /ωn

tan φ1 = 2 ; tan φ = 2 (2.11.6)
1 − ωωn 1 − ωωn
i

m z 2 (t)
F0 sin t
Z(t) = A sin ( t- 1-
k c = relative displacement
z1(t) = A sin( t- (as Z ≈ A)
Figure 2.11.5 Ground displacement measuring instrument.
Pick up F(t), A(t)

F = F0 sin t F0 sin t
footing A0 sin t
t
soil
Figure 2.11.6 Footing-soil system.
where ωn is the natural frequency of soil + footing system; other terms are as defined
in the preceding.
To obtain D one may use (φ1 + φ) and (φ − φ1 ), if possible.
2.11.3.5 Acceleration transducers

There is no magnetic field here. The Figure 2.11.7 shows a schematic sketch of an
accelerometer. The principle is “when there is a pressure difference between the faces
it produces a voltage difference”, which is a measure of the force acting and hence the
acceleration of the mass of the crystal.
For the mass m of the crystal:
mz̈2 + c(ż2 − ż1 ) + k(z2 − z1 ) = 0; or mz̈2 + cż2 + kz2 = cż1 + kz1 (2.11.7)
Assuming,
z1 = A1 sin ωt, ż1 = A1 ω cos ωt, hence cż1 + kz1 = cA1 ω cos ωt + kA1 sin ωt
Thus√the right hand side of Equation (2.11.7) reduces to F sin (ω + φ1 ): in which
F = A1 c2 ω2 + k2 and tan φ1 = cω k
= 2D ωωn . The angle φ1 is the angle between force
(F) and the displacement of the ground z1 .

Piezo-electric crystal
m Accln. measured
z2(t)
k c O
z1(t)
Figure 2.11.7 Accelerometers.
Solution of Equation (2.11.7) is given by

√
(A1 c2 ω2 + k2 )/k
z2 = sin (ωt + φ1 − φ2 ) = A2 sin (ωt + φ1 − φ2 ) (2.11.8)
(1 − r2 )2 + (2Dr)2

A2 z2 max 1 + (2Dr)2
= = (2.11.9)
A1 A1 (1 − r2 )2 + (2Dr)2
To have A2 /A1 ⇒ 1, r-value should very low. For a typical instrument where
fn = 3200 Hz (say) which much higher than the operating frequency normally encoun-
tered in practice. Within a range of r ≤ 0.2, such a situation is encountered. These are
shown in Figures 2.11.8 to 10.
2.11.3.5.1 Phase angles

2Dω
Between z1 and generating force: φ1 → tan φ1 = ;
ωn
2Dω/ωn
Between generating force and z2 : φ2 → tan φ2 = 2 (2.11.10)
1 − ωωn
Between z2 and z1

−1 2Dr −1 −1 2Dr3
➔ φ = (φ2 − φ1 ) = tan − tan (2Dr) = tan
1 − r2 1 − r2 (1 − 4D2 )
(2.11.11)
An ideal accelerometer is the one in which instrument mass is servo-controlled to

have zero relative displacement; the force necessary to accomplish this becomes a
measure of the acceleration.

D=0
180
140
Phase angle,
90
A2/A1
Range for making A2/A1 = 1 40 0.707
0
1 0 0.5 1 1.5 2 2.5 3
Frequency ratio, r
0 6
1 2 4
r
√2
Figure 2.11.8 Accelerometer response.
D=.05 0.01 0.15

in radian
0.5 1
0
0 2 4 6
r
Figure 2.11.9 Phase angles.
φ1 z1
( φ2-φ1)
cωz1 F
kz1 z2
cωz1
φ2
kz2
Reference
mω2z2 ωt
Figure 2.11.10 Vector diagram for phase angles.

The piezoelectric crystals are mounted in such a manner that under acceleration
they are either compressed or bent to produce an e.m.f. which is ultimately converted
to electrical signals. The natural frequencies of such accelerometers can be made very
large, say in the range of 50 000 Hz. The piezoelectric crystal mounted accelerometers
can be made very small in size, may be of the range of 10 mm in diameter and height
and are very rugged to withstand a shock as high as 10 000 g acceleration. A typical
instrument may have fni → 1 Hz to 5 Hz useful frequency, foperating → 10 Hz to
2000 Hz which means r is more than 10. Sensitivity of such instruments may be in the
range 20 mV/(cm/sec) to 350 mV/(cm/sec) with maximum displacement = 0.5 cm
(double amplitude) [Note Arms = 0.707 A].
Sensitivity of crystal-type accelerometer is denoted either in terms of charge, i.e. pico-
coulombs = pC = 10−12 coulombs per g or in terms of voltage, i.e. millivolts = mV =
10−3 V per g. Sensitivity of a crystal-type accelerometer can be established from: say a
typical crystal accelerometer is 25 pC/g with crystal capacitance equal to 500 pF (pico-
farads). Voltage from the classical equation E = Q/C, gives the sensitivity = 25/500 =
0.05 V/g or 50 mV/g as sensitivity in terms of voltage. Again, if the accelerometer is
connected to a vacuum tube voltmeter through a 3 m long cable of capacitance 300
pF, the open circuit output voltage of the accelerometer is reduced to (50)(500)/(500
+ 300) = 31.3 mV/g. This loss can be avoided by using a charge amplifier, in which
case the capacitance of the cable has no effect.
2.11.3.6 Amplitude distortion

Normally the measured vibrations consist of a number of harmonic motions of various
frequencies. Amplitude distortion occurs in an accelerometer if the acceleration of one
harmonic is amplified more than another. From a harmonic solution, the amplitude
of acceleration can be written as ω2 A. For an equal amplification to acceleration, it
is desirable to have κ/ωn2 nearly same for all frequencies. For r = 0, κ = 1. Thus, the
amplitude distortion can be defined as the change in κ/ωn2 with respect to r = 0. The
percent amplitude distortion is
κ 1
ωn2
− ωn2
100 × 1
= 100 × (κ − 1)
ωn2
It can be observed from Figure 2.11.11 that accelerometer should be built with D
lying between 0.6 and 0.7 to minimize the amplitude distortion.
2.11.3.7 Phase distortion

This distortion occurs if the relative phase of the harmonics recorded is different from
that of the vibration to be measured. For zero distortion, the shift φ should increase
linearly with frequency of the harmonic motion. The phase shift at r = 1 is always π /2.
For zero distortion, the phase shift for 0 < r < 1 should be 90r degree. Hence phase
distortion in an accelerometer can be defined as: Phase distortion = (φ − 90r) degree.
It can be noticed in Figure 2.11.11 that appropriate damping in an accelerometer is
necessary for minimizing the phase distortion.

To record any complex wave without changing its shape, the phase of all harmonics
must remain unchanged with respect to the fundamental. This requires that the phase
angle be zero or that all the harmonic components must be shifted equally. The first
case of zero phase shift corresponds to D = 0 for r < 1. The second case of equal time-
wise shift of all harmonics is nearly satisfied for d = 0.7 for r < 1 (Figure 2.11.12)
and when D = 0.70, the phase for r < 1 can be expressed by φ ≈ π r/2. Thus for d = 0
or 0.70, the phase distortion is completely eliminated.
10
D=0
Amplitude distortion in %
D = 0.6
+
0
D = 0.65
D = 0.75
D = 0.7
10
0 0.2 0.4 0.6 0.8 1
Frequency ratio, r
Figure 2.11.11 Amplitude distortion in accelerometer.
Frequency ratio, r
0 0.2 0.4 0.6 0.8 1 1.2
2
0 D = 0.75
-2 0.70
Phase distortion
-4
0.6
-6
-8 0.5
-10
-12
Figure 2.11.12 Phase distortion in accelerometer.
Example 2.11.1
1 A manufacturer of vibration measuring instruments gives the following

specif ication for one of its vibration pick ups;
Frequency range: velocity response flat from 10 Hz to 1000 Hz.
Sensitivity: 0.096 V/cm/sec, both volts and velocity in rms values.

Amplitude range: Almost no lower limit to maximum stroke between stops

of 0.60 in.
a This instrument was used to measure the vibration of a machine with

a known frequency of 30 Hz. If a reading of 0.024 V is indicated,
determine the rms amplitude.
b Could this instrument be used to measure the vibration of a machine
with known frequency of 12 Hz and double amplitude of 0.80 cm?
Give reasons.
Solution:
a Voltage = 0.024 V : Sensitivity = 0.096 V/(cm/sec) : Velocity = 0.024/0.096

cm/sec = 0.25 cm/sec.
f = 30 Hz : ω = 2π × 30 rad/sec : Amplitude = velocity/ω = 0.132 × 10−2
cm = 0.0133 mm.
f = 12 Hz : ω = 2π × 12 rad/sec : Amplitude = 0.40 cm.
Velocity = 2π × 12 × 0.40 = 30.159 cm/sec
b Now for f = 10 Hz : ω = 2π× 10, amplitude = 0.3 × 2.54 = 0.762 cm:
velocity = 0.762 × 20π = 47.88 cm/sec.
For f = 1000 Hz, velocity = 0.762 × 2000π = 478877.8 cm/sec
Velocity required is 30.159 cm/sec.
→ So the instrument cannot be used.
2 The sensitivity of a certain crystal accelerometer is given as 18 pC/g, with

its capacitance equal to 450 pF. It is used with a vacuum tube voltmeter and
its cable is 5 m long with a capacitance of 50 pF/m. Determine its voltage
output per g. Ans. E = 25.7 mV/g.
Solution:
Sensitivity = 18 pC/g, crystal capacitance = 450 pF.

Sensitivity in terms of voltage = 18/450 0.04 V/g [E = Q/C].
Total cable capacitance = 50 × 5 = 250 pF
Output voltage = 40 × 450/(450 + 250) = 25.7 mV/g
3 A vibration pickup has a sensitivity of 40 mV/(cm/sec) between f = 10 Hz

to 2000 Hz. If 1 g acceleration is maintained over this frequency range,
what will be the output voltage at (a) 10 Hz and (b) at 2000 Hz. Ans. (a)
624.5 mV, (b) 3.123 mV.
Solution:
Sensitivity = 40 mV/(cm/sec) = 0.04 V/(cm/sec)
a F = 10 Hz → ω = 2π × 10 = 62.83 rad/sec

Acceleration: 1 g = 981 cm/sec2 → velocity = 981/62.83 = 15.61356 cm/sec.

Voltage = (40 × 981)/62.83 mV = 624.5 mV
b f = 2000 Hz → ω = 2π× 2000 rad/sec
Velocity = 981/(2π× 2000)
Voltage = (40 × 981)/(2π× 2000) = 3.123 mV.
2.12 EVALUATION OF FRICTION DAMPING FROM ENERGY

CONSIDERATION
E = F0 Aπ sin φ, phase angle φ at resonance is 90◦ , and the energy input is F0 Aπ .

Energy dissipation, using friction concept is 4fA. Now, if 4f /π F0 < 1, energy input
exceeds the energy dissipation, and the excess energy accumulated over the cycles
builds up the amplitude of oscillation.
Consider an embedded footing, Figure 2.12.1.
Governing equation is
mz̈ + cż + Ceq ż + kz = F0 sin ωt (2.12.1)
Steady state solution is
F0 /k
A= 2 sin (ωt − φ) = A0 sin (ωt − φ) (2.12.2)
(1 − r2 )2 + Ceq ωk + (2Dr)2
A0 can be obtained from

4f 2
F0 1− πF0
A0 = . (2.12.3)
k
(1 − r2 )2 + (2Dr)2
If we have two observations with forcing functions: F1 (t) = F01 sin ωt and F2 (t) =
F02 sin ωt, there will be two responses namely, A01 sin (ωt − φ) and A02 sin (ωt − φ).
F = F0 sin t
f/2 f/2
m
⇒
k c
Figure 2.12.1 Vibration of a footing with side friction.

A0
r
r=1
Figure 2.12.2
Now using Equation (2.12.3) we can write

4f 2
A01 F01 1− πF01
= (2.12.4)
A02 F02 4f 2
1− πF02
A01 , A02 , F01 and F02 are known and hence we can obtain a representative value of
f , the friction force.
Now, for D = 0 Equation (2.12.2) gives
2
1 − 4f
F0 πF
A0 =
k 1 − r2
and the response is given in Figure 2.12.2.

But 4f < πF0 has to be satisfied for a real system, i.e. 4fA0 sin φ < π F0 A0 sin φ.
Hence, work done by the friction force is less than the work done by exciting force.
This implies building up of energy and hence a resonant situation will arise.
If f is large we have to use more exact analysis for solution as the motion cannot be
assumed to be harmonic.
2.13 VIBRATION ISOLATION
Vibratory forces generated by machines and engines are often unavoidable; however,
their effect on a dynamical system can be reduced substantially by properly designed
springs, which are referred to as isolators. Protection of the base against the action of
driving forces is called active isolation and protection against kinematic disturbances
is called passive isolation. Thus, when the noise-making source itself is isolated from
other structures, the isolation is an active isolation whereas when other structures
are isolated from the noise making sources, the isolation is a passive one. In active
isolation the basic problem is that of determining the force transmitted to the base;
in the theory of passive isolation, it is the problem of finding the amplitude of the
vibration the object is to be protected is forced into.

F = F0 sin ωt
k c
Figure 2.13.1 Vibration isolation (active).
mω2x
FTr
F
cωx
φ
x kx
Figure 2.13.2 Vector diagram for the system shown in Figure 2.13.1.
2.13.1 Active isolation

Let us consider a system shown in Figure 2.13.1. Let F = F0 sin ωt acting on a SDOF
system.
The force transmitted to the ground through springs and damper:
cω 2
2
FTr = (kx) + (cωx) = kx 1 +
2 (2.13.1)
k
The vector diagram can be shown as in Figure 2.13.2.

Solution of the problem can be written as
F0
k F0
x= sin (ωt − φ) = κ sin (ωt − φ) (2.13.2)
(1 − r2 )2 + (2Dr)2 k
Thus from Eqns. (7.13.1) and (7.13.2) one can write

FTr 1 + (2Dr)2
= (2.13.3)
F0 (1 − r2 )2 + (2Dr)2
This is identical to the one developed for accelerometers.

Thus the problem of isolating a mass from motion of the support point is identical
to that of isolating the disturbing forces. The ratio, FTr /F0 is called the transmissibility.
Without the loss of generality, setting D = 0, Equation (2.13.3) can be written as
FTr = κF0 (2.13.4)
Hence, the efficiency of active isolation depends on the magnitude of the magnifica-
tion factor κ to be used. This requires accordingly a low value of the natural frequency
ωn , which may be achieved by reducing the stiffness of the mounting of the machine
or by increasing the vibrating mass. √
Transmissibility
√ is less than unity only for r > 2. → Isolation is possible only for
ω/ωn > 2 (refer to Figure 2.11.3). An un-damped spring is superior to a damped
system in reducing transmissibility. However, to reduce amplitude near resonance
some damping is desirable.
It should be noted that vibration isolation of slow-speed machines (when the fre-
quency of the disturbance is not high) may require a very low natural frequency and
accordingly impractically great flexibility of vibration absorbers. To overcome this dif-
ficulty the vibrating mass is artificially increased in such cases. This serves a twofold
objective; first, the natural frequency is reduced and, second, sufficient stiffness of the
system is preserved.
It is possible to reduce the amplitude of vibration by supporting the machine on a
large mass or by other means is shown in Figure 2.13.3.
Again a set of elastic constraints (vibration isolators), in the form of steel springs or
rubber elements are introduced under the frame of the machine to be isolated.
m Machine
(a) (b)
Machine
(c) (d)
Figure 2.13.3 Active isolation.

Two equally efficient types of mounting are in use. These are supporting type when
vibration isolators are placed under the base of the machine (Figure 2.13.3a) and
suspension type when vibration isolators are placed above the bottom of the base in
the latter case the vibration isolators may be either in compression (Figure 2.13.3b)
or in tension (Figure 2.13.3c). If horizontal vibration prevails in the machine to be
isolated, a pendulum type suspension may be used to advantage (Figure 2.13.3d).
To keep transmissibility same, k must be increased in the same ratio so that
(m + M)/k remains the same.
Say for some transmissibility, if we increase the mass m to m + M

FTr 1 + (2Dr)2
= (2.13.5)
F0 (1 − r2 )2 + (2Dr)2
k must be increased in the same proportion so that (m + M)/k remains the same.
Thus as k → increases
F0
k
x= sin (ωt − φ) (2.13.6)
(1 − r2 )2 + (2Dr)2
→ x will be reduced.
When damping is negligible
FTr 1
Transmissibility = = 2 (2.13.7)
F0 ω
ωn −1
√
where it is understood that ω/ωn to be used is always greater than 2.
2.13.2 Passive isolation

Consider the case in which, there is a vibration of the ground in Figure 2.11.7 instead
of the force, F.
Following Equation (2.11.9), we can write
mẍ2 + cẋ2 + kx2 = cẋ1 + kx1 (2.13.8)
If x1 (t) = X1 sin ωt, Equation (2.13.8) reduces to
mẍ2 + cẋ2 + kx2 = X1 [k sin ωt + cω cos ωt]

= X1 k2 + c2 ω2 sin (ωt + φ) = Fx sin (ωt + φ) (2.13.9)
cω
where tan φ = = 2Dr. (2.13.10)
k

(a) (b)
Figure 2.13.4 Passive isolation.
The steady state solution is

x2 = X1 1 + (2Dr)2 κ sin (ωt − ψ + φ) (2.13.11)
in which
2Dr
tan ψ = (2.13.12)
1 − r2
The passive isolation is used to protect instruments and precision machines against
vibrations transmitted from the supporting structure. The amplitude of vibration of
an isolated object is expressed in terms of the amplitude of vibration of the base by
Equation (2.13.10). Thus a passive isolation should use the same idea of making the
mounting soft, as in the case of an active isolation. It is generally required that the
natural frequency of the isolated object shall not exceed one-fourth of the frequency of
vibration of the base. If the frequency of the disturbance is not known, it is necessary
to introduce in elastic pads in the mounting system. Thus the irregularities of a road
may have the shape of a sine curve with the wave length varying over a wide range.
Therefore, there is a real danger that the body of a moving vehicle may be in a state
of resonance; to limit resonant amplitudes the vehicle suspension is always provided
with hydraulic shock absorber which dissipate a considerable amount of energy during
vibration [Figure 2.13.4(a)]. This absorber system has a disadvantage: it does not
afford sufficient comfort of passengers when subjected to shocks which are transmitted
to the automobile body with almost no relief. To obtain the necessary softness of
the suspension it may be provided with additional flexible elastic damper shown in
Figure 2.13.4(b).
2.13.3 Isolation by trench

An exhaustive field study was carried out by Woods to examine the effectiveness
of open trenches as barrier for vibration isolation. Lamb analysed the problem of
the propagation of tremors over the surface of an elastic solid. He solved the problem
of spreading out of a symmetrical annular wave disturbance around a point source.

These waves consist of body waves, namely longitudinal wave (Primary wave or
P-wave), transverse wave (Shear wave or S-wave) and surface waves, namely, Rayleigh
surface waves (R-wave). At any point on the surface P-wave arrives first and it under-
goes an oscillatory displacement. It is followed by a relatively quiet period till another
oscillatory displacement owing to the arrival of the S-wave. Lamb termed this phase
of motion as minor tremor. A much larger oscillatory movement is followed subse-
quently due to the arrival of Rayleigh waves termed as major tremor. P-wave travels
faster then S-wave and R-wave is slightly slower than the S-wave. As to the nature of
wave propagation, a compressional wave (i.e. P-wave) propagates radially outward
from the source as hemispherical wave front whereas Rayleigh waves propagate radi-
ally outward in a cylindrical wave front. As waves travel outward the energy density
decreases with the distance from the source of disturbance. This is known as radiation
or geometric damping. The amplitudes of compressional wave attenuates approxi-
mately in proportion to 1/r, r is the distance of the source of disturbance. Along the
surface of an elastic halfspace, the attenuation is proportional
√ to 1/r2 . For the Rayleigh
waves, the decrease in amplitude is proportional to 1/ r. Again, about two-third of
the total energy of vibration is normally carried through the Rayleigh wave and its
smaller decay with the distance in comparison to other waves. Thus, the Rayleigh
wave is more important for structures near the surface, particularly in the event of
earthquakes, blasts and other dynamic operations.
2.14 MACHINE FOUNDATION SUPPORTED ON FRAMES
2.14.1 Introduction
In this section we will deal with machines supported on frames. These are also
sometimes termed as frame foundations. These type of foundations usually support
equipment like steam turbines (ST), boiler feed pumps (BFP), in power plants, com-
pressors in petroleum refineries, air blowers in automobile industry etc. Though the
basic analytical principle remains the same, for the present chapter however, we will
restrict our discussion mainly to foundations related to steam turbines and Boiler feed
pump only.
A pre-requisite to this section is again Chapter 5 (Vol. 1) (basic concepts in structural
and soil dynamics) and you should have gone through the previous section on analysis
and design of block foundation.
We also envisage that you have some basic concepts on Matrix Analysis of Structure
whose concepts we are going to use quite in detail.
Turbines and Boiler feed pumps form the heart of any power plant. Thus for any
developed and developing nation, capacity of supplying unhindered energy not only
ensures a steady industrial growth, but also goes on to improve the quality of life in
a long way. The main source of this energy is obviously electricity and this is what
a turbo-generator generates, based on the electro-mechanical process. Thus if the
foundation which supports these critical machines misbehave and the machine trips
during operation, the cascading effect on the end users and the industry dependent on
the power generated could suffer severe losses. If the shortage is severe in nature, this
could even have a very adverse effect on the economic growth to a complete part of a
country.

Thus for successful operation two aspects become critical for these machines
• The machine itself should run smoothly (round the clock).

• The foundation supporting the equipment is capable of sustaining the various
loads coming from the turbine under operation as well as those that could develop
due to the vagaries of nature or otherwise like earthquake, thermal, electrical
faults, short circuits etc.
In factory, quality and performance of the machine is controllable since materials

used are all man made and all requisite appurtenances are manufactured under a
careful controlled condition.
Thus, it is not difficult to arrive at a condition in the manufacturing process where
quality of output for two machines coming out of a factory could be stated to have
identical mechanical characteristics. However, for a civil engineer designing its foun-
dation the situation is completely different. He neither has control on the subsoil on
which it is being built nor he has any control on the vagaries of nature like earth-
quake, wind etc. In addition to this he has to cater to a number of uncertain loads at
the start of his design like piping loads, stator loads, and electrical fault loads etc and
still make sure that the foundation functions within acceptable limits of engineering
norm. Considering the difficult natural parameters, enormity of the machines and risk
involved in terms of public outcry, turbine foundations still remain one of the most
difficult and challenging task in civil engineering profession.
The engineer not only needs a very specialized knowledge in various aspects of
civil engineering like structural mechanics, dynamic theories related to structures and
soil, he should also have some interdisciplinary appreciation about mechanical and
electrical aspect of the machine itself. Though advent of digital computer has made
the life much simpler in terms of accurate calculations and analyzing the output results
visually, for turbine foundations this should be supplemented with some engineering
judgment and experience. For this is a case where the computer output numbers only,
do not reflect the actual picture. The engineer has to carefully weigh the effect of the
idealization in his mathematical modeling that has created these numbers and take
design decision using his engineering judgment. So before we get into the main topic
itself. . .
We plead with our readers to be cautious with this type of foundations and not hesi-
tate to take help of engineers who are experienced and also the construction people who
has constructed and commissioned such turbines and monitored their performance36 .
2.14.2 Different types of turbines

and the generation process. . .
Before we go into the analysis and design aspect of such a turbine foundations it would
be useful to know something about the machine itself, how it behaves and why we
36 He may be an old man not so expert with computers as our modern day engineers but remember that
his experience is worth more than a million dollar software you may write for he has a feel of this giant
who if starts misbehaving can have a very serious consequence.

take the trouble of mounting it on a frame when putting it on a massive block resting
on ground would have made our life much easier.
Irrespective of the nature of fuel like fossil fuel (coal fired plant), LPG/ Naptha (like
in Combined cycle or open cycle plant), processed uranium rods (in nuclear power
plant) basic principle of operation of turbine remains the same.
The fuel is used to generate steam to a pre-designed temperature in boiler and is
allowed to expand within a turbine under pressure. This generates a mechanical energy
which makes the turbine rotate.
The turbine shaft in turn is connected through a coupling or a synchronous clutch
to a generator rotor, that is rotated by the turbine and generates electrical forces due
to mechanical movement of the generator shaft in a magnetic field. The electric power
thus generated is transferred through bus duct connections to a primary transformer
where after stepping up the voltage it is supplied to power grid through a switch yard.
This in essence is the simplified process of electricity generation.
The machine itself is a centrifugal machine and are usually of two types
• Gas driven
• Steam driven
The gas driven one basically uses Naptha or natural gas as the base fuel and even at
exhaust, it contains substantial thermal energy. This is usually recycled through a heat
recovery system to further heat water into steam and is passed off to a steam driven
turbine to generate further electricity.
While gas driven turbine does not require any condenser at the gas exhaust, steam
turbines will always have a condenser connected to the steam exhaust to condense
off steam coming out of the turbine. This is collected in a hot-well from where it is
Generator on bearing Coupling Turbine on bearing

Shaft
Condenser
(spring mounted)
Figure 2.14.1 Longitudinal profile of a turbine foundation with the equipment.

further recycled back to boiler through a condensate extraction pump. For steam to
condense, usually advantage of gravity force is taken and also from plant layout and
pipe routing consideration, the condenser should preferably be positioned at a level
which is lower than the turbine operating floor level. Based on this concept the best
location for placing the condenser is usually at a location directly below the turbine.
It is for this steam driven turbines are usually mounted on frames to take advantage
of the space beneath it, while for gas turbines, as no such requirements are essential,
are usually mounted on block foundations.
Besides this, the frame mounted machines also provides easy access to electrical
connections to generator and main steam pipes. Connecting the steam pipe from the
bottom is preferable for it avoids dismantling of pipe work during maintenance; this
also prevents pipe work draining into the turbine.
A typical schematic sketch of a turbine foundation is as shown in Figure 2.14.1.
2.14.3 Layout planning

For turbines placed in a power house typical layout which is most common is as shown
in Figure 2.14.2.
Spring mounted
Boiler feed pump
(Turbine driven)
Condenser
Spring mounted
Figure 2.14.2 Typical cross section of turbine pedestal and power house.

For a typical 900 MW power plant this is usually about 16/17.00 m above the power
house floor level with condenser mounted on springs.
The 17.00 m level is also known as the operating floor level of the power house.
In such case, the boiler feed pump (BFP) which feeds the water to the boiler is driven
by turbine itself and usually rest on RCC block foundation mounted on springs over
steel structure.
For plants of lower capacity when the top deck height is much lower, BFP is also
sometimes mounted on RCC frames similar to the turbine.
2.14.4 Vibration analysis of turbine foundations

We stated earlier that analysis of turbine foundation is a complicated job and requires
a lot of ingenuity and deep insight to the problem of dynamics for these are machines
which are massive and rotates at a very high speed37 .
We present here the following methods of analysis of the framed foundation:
• Resonance or Rausch’s (1959) method

• Amplitude or Barkan’s (1962) method
• Combined or Major’s (1980) method
• A 2D soil structure interaction model of framed foundation that can take into
cognizance the effect of underlying soil/pile as frequency independent springs
(Chowdhury 1984).
• The 3D finite element model of the foundation considering the underlying soil into
cognizance. This is analyzed through a computer.
2.14.4.1 Rausch’s method

Rausch proposed a method where the basic criteria that needs to be satisfied is that the
fundamental natural frequency of the foundation should be out of tune to the operating
frequency of machine by about (±) 20%. He suggested a mathematical model where
for natural frequency in vertical direction for the individual cross frames self weight
and superimposed load on longitudinal girders and the load coming from the machine
is considered as lumped mass over the columns having single degree of freedom (Figure
2.14.3)38 .
For horizontal direction he assumed the bottom raft to be infinitely rigid and again
proposed a mathematical model having single degree of freedom. He also assumed
that in vertical direction the average of natural frequency of the frames is the natural
frequency of the system in that direction.
37 For 50 Hz power grids the typical RPM of machines are 3000 RPM. For 60 Hz grids the speed is about
3600 RPM. For Nuclear power plants these are about 1500–1800 RPM. 50 and 60 Hz are standard
Power grid cycles available globally.
38 This is surely an over simplification of the problem.

N P N
m=(P+Q+2N)/g
KV
Typical cross frame Equivalent mathematical model
Figure 2.14.3 Idealization in Rausch method.
1 Frequency in vertical direction
As per Rausch if there is n number of frames in the foundation and if fv is the natural
frequency of the structure in the vertical direction, then

n
fv = fi /n (2.14.1)
i=1

Knowing, ωn = Kv /m rad/sec

we have, ωn = (Kv g)/W
where, g = acceleration due to gravity; W = weight acting in the vertical
direction.
If δst is the static deflection of the frame then, δst = W/Kv i.e. ωn = g/δst rad/sec.
Using T = 2π/ωn ,
√
δst 1 g 60 × 9.81 ∼ 30
we have, T = 2π secs; f = cps, ➔f = √ =√ cycles/min
g 2π δst 2π δst δst
(2.14.2)
The vertical frequency of the of individual frame in vertical direction is thus given by

fv = 30/ δv cpm (2.14.3)
where δv = the total vertical deflection at mid-point of the cross beam in meters.
Hence, for different types of loading as shown above,
δv = δ1 + δ2 + δ3 + δ4 (2.14.4)


PL3 2ψ + 1
where, δ1 = is deflection due to concentrated load;
96EIb ψ +2

QL3 5ψ + 2
δ2 = is the deflection due to uniformly distributed load;
384EIb ψ + 2

3 L Q
δ3 = P+ is the deflection due to shear; (2.14.5)
5 EAb 2

h P+Q
δ4 = N+ is the axial deflection of column due to the
EAc 2
concentrated load transferred from the longitudinal girder (N)
in which, P = concentrated load from the machine; Q = UDL of the cross beam (qL);
q = self weight per unit length of the cross beam; N = concentrated load on the
column; Ab = area of cross section of the beam; Ac = area of cross section of the
column; Ib = moment of inertia of the beam; Ic = moment of Inertia for the column;
E = dynamic modulus of elasticity of the frame; h = effective height of the column;
L = effective length of the cross beam, and ψ = (Ib h)/(Ic L).
2 Frequency in horizontal direction

Again considering single degree of freedom the natural frequency fh is given by

Kh1 + Kh2 + · · · · · · · · · + Khn
fh = 30 (2.14.6)
W

12EIc 6ψ + 1
where, W = total load of machine plus the top deck and Khi = .
h3 3ψ + 2
This method does not have any provision of calculation of amplitude and suffers
from following drawbacks:
• Over simplification of the mathematical model based on single degree of freedom.

• A resonance check does not necessarily ensure that the design is safe and the ampli-
tudes are within acceptable limits, especially for low tuned foundation which has
been observed to undergo significant displacement when the machine speed passes
through the natural frequency value during start and stopping of the machine.
• It considers the bottom raft as stiff and finds frequency in translational mode only,
no rocking mode frequency has been calculated, and this could have significant
contribution to the overall dynamic response (which of course depends on the
geometry of the foundation system).

2.14.4.2 Amplitude or Barkan’s method

Barkan improved upon Rausch’s method by taking into consideration the following
steps for the analysis:
• In vertical direction he considered a two mass lumped model for analysis.

• He derived the translational amplitude by taking coupled rotation of the top deck
plain considering the top deck as rigid mass supported on a series of leaf springs
which represented lateral stiffness of the columns.
• However, like Rausch he also assumed the frames to be supported on slab that
is infinitely stiff and thus ignoring the effect of elastic base (soil) supporting the
bottom raft.
1 Calculation in the vertical mode

Barkan argued that under vertical mode, the transverse frame will take the deformed
shape as shown in Figures 2.14.4 and the mathematical idealization may be showed
as given in Figure 2.14.5.
m1/2 m1/2
m2, k2
k1/2 k1/2
Figure 2.14.4 Transverse frame of the foundation.
m2 z2
k2
m1
z1
k1 Fixed at base
Figure 2.14.5 Mathematical model in vertical direction.

Since the columns are stiff and have similar inertia they would deform uniformly
under axial compression while the longitudinal girders will try to resist the flexural
deformation of the transverse beam based on their torsional stiffness.
As torsional stiffness of the longitudinal girder is much less than axial stiffness of
the columns or flexural stiffness of the transverse girder, its effect on overall dynamic
response of the system is marginal and can be neglected.
Similar to the proposition of Rausch he also suggested that the transverse frames
can be treated independent of each other in the vertical direction39 .
Based on the above he defined the various analytical parameters for each transverse
frame as follows:
2EAc L3 (1 + 2ψ) 3L 1
k1 = ; δv = + ; k2 = (2.14.7)
h 96EIb (2 + ψ) 8GAb δv
where, G = dynamic shear modulus of concrete @ 0.5E.
Calculation of mass m
m2 = m0 + 0.45mb
where, m0 = P/g is the concentrated mass of the machine carried by the beam and
mb = the mass of the transverse girder and
m1 = mL + 0.255mb + 0.35mc .
in which, mL = mass from longitudinal girder transferred to the frame; mc = mass of

the column.
The natural frequency of each frame is then obtained from the equation

m1 0 z̈1 k1 + k2 −k2 z1
+ =0 (2.14.8)
0 m2 z̈2 −k2 k2 z2
Similarly the amplitude of each frame can be obtained based on the method we have
explained earlier40 .
For amplitude calculation, the vertical dynamic load was assumed as
Pv = Ci sin ωm t (2.14.9)
where, Ci = (R/g)eωm2 , in which, R = weight of the rotor; e = eccentricity of the
rotor, and ωm = operating frequency of the machine.
39 For a modern day engineer this might appear as Barkan was trying to simplify the case but what we
should realize was that he did not had a desk top computer readily available on his desk nor were
computers so easily available. It was an era when most of the calculations were done manually. What
is most appreciable was that he idealized and modeled an extremely complex problem to a level which
was amenable to manual calculation and in-spite of the simplification gave results which were very
reasonable.
40 We have explained the method of calculation of natural frequency and amplitude of vibration for
harmonic load for system with two degrees of freedom quite in detail in Chapter 5 (Vol. 1).


In this case it was assumed that the top deck and bottom slab as infinitely rigid in
its own plane and the columns act as leaf springs where the stiffness of the springs
tantamount to transverse stiffness of the individual columns. Figure 2.14.6 shows the
mathematical model perceived by Barkan.
Mass on each horizontal spring is given by
mi = m0i + mbi + 0.3mci + mLi (2.14.10)
Here the horizontal displacement is given by
h3 (2 + 3ψ) 1
δhi = and Khi = (2.14.11)
12EIc (1 + 6ψ) δhi
Here the term i represents the ith cross frame of the system.
Wb Wc Wd
Wa Xgb Xgc
Xga Xgd
H G
C/L axis H
G
Ka Kb Xhb Kc Kd
Xha B Xhc D
A C
Wa Wb Wc Wd
Ka Kb Kc Kd
Figure 2.14.6 Mathematical model in horizontal vibration.

It was argued that due to difference in geometry and shape, there will be some
difference between the center of gravity (G) andcenter of stiffness (H).

N
While the resultant of all the masses i=1 mi will pass through the point G, there
as thecenter of elasticity (H) through which resultant of

exists another point known
N
all the column stiffness i=1 Khi will pass.
As these two points do not coincide, as such other than translation the top deck
will also undergo a rotation in the horizontal plane (φ) which will be coupled with the
translation (x).
Taking the center of gravity as the reference co-ordinate point he obtained the
following differential equation of motion
Mẍ + Kh x + Kh eφ = Ph cos ωm t and Jφ φ̈ + Kh ex + (Kh e2 + γ )φ = Mh cos ωm t

(2.14.12)

N N
in which, M = for N number of frames; Kh =
i=1 mi i=1 Khi for N number of
N
frames; e = distance between the points H and G; Jφ = Ni=1 mi Xgi ; γ =
2 2
i=1 Khi Xhi ;
N N
Phi = Ci cos ωm t; Ph = i=1 Phi and Mh = i=1 Phi Xgi .
Writing the above equation in the matrix form, we have

M 0 ẍ Kh Kh e x Ph cos ωm t
+ = (2.14.13)
0 Jφ φ̈ Kh e Kh e 2 + γ φ Mh cos ωm t
The coupled natural frequency of the system can be obtained from the equations
f (ωλ2 ) = ωn4 − (αωx2 + ωϕ2 )ωn2 + ωx2 ωϕ2 = 0 (2.14.14)

N
Here ωx = Kh /M , ωφ = i=1 Khi Xh /Jφ and α = (1+e )/r where r = Jφ /M
2 2 2 2 41
i
3 Amplitude of vibration
The amplitude of vibration is obtained from the expression

e2 Ph e2 2
r2
+ ωx2 + ωϕ2 − ωm
2
M − eωx2 M
Jφ
h
ω
r2 x
Ph
M
2 ) Mh
− (ωx2 − ωm Jϕ
x= ; φ=
f (ωλ2 ) f (ωλ2 )
(2.14.15)
The net amplitude of horizontal vibration is given by
xnet = x + X φ (2.14.16)
41 Alternatively this can also be calculated based on the eigen value technique for two degree of freedom
showed in Chapter 5 (Vol. 1).

where, X = is the farthest point from the center of gravity point G.

We can also apply the modal technique shown earlier in the chapter 5 (Vol. 1) and
directly find out the amplitude of vibration in matrix notation too.
2.14.4.3 Combined or Major’s method of analysis

Major actually combined the above two methods (Barkan & Rausch) to arrive
at a method which is usually known as the combined method of analysis.
He realized that the resonance method of Rausch and amplitude method of
Barkan are actually mutually complementary and blended the better of the two
approaches42 .
Thus it would be worth knowing as to what improvements Major did with respect
to the previous two models.
The improvements may be summarized as follows:
• Both Rausch and Barkan neglected the effect of underlying soil from their cal-
culation43 , Major did try to cater for the effect of soil at least in vertical
mode of vibration by adding the soil deformation to elastic deformation of the
frame.
• As stated earlier that resonance check does not always prove to be an ade-
quate design especially for under-tuned foundation which are found to show
significant vibration during start and stop of the machine, Major did devise
a model where the foundation behavior under this transient can also be
checked.
These, in essence, are the two significant contribution of Major in his combined
method.
The methodology applied in this method is explained hereafter44 .
For vertical frequency analysis Major followed in essence the method proposed by
Rausch except that he took Barkan’s two-mass model as shown in Figure 2.14.5.
Here, m2 = mass of the (upper slab + machine) + 0.5 times the mass of the column;
m1 = mass of the bottom slab + mass of the condenser + 0.5 times the mass of the
column; k2 = equivalent spring constants for the columns, and k1 = equivalent spring
constants of the soil.
42 IS 2974 also recommends Major’s method for design of the Turbo-generator foundations.
43 Though Barkan acknowledged that this might affect the response but conceded that the analy-
sis was too complex to be done manually and for very thick bottom raft, the effect of soil was
negligible.
44 We apologize, for there would be some repetition with respect to earlier method of Rausch and Barkan.
But we would still like to repeat it for firstly- the clarity and secondly to highlight what is the difference
in approach with respect to the previous two methods.

He proposed that total vertical deflection is given by45

δv = δ1 + δ2 + δ3 + δ4 + δs (2.14.17)
Here,

PL3 2ψ + 1
δ1 = is deflection due to concentrated load;
96EIb ψ +2

QL3 5ψ + 2
δ2 = is the deflection due to uniformly distributed load;
384EIb ψ +2

3 L Q
δ3 = P+ is the deflection due to shear; (2.14.18)
5 EAb 2
!
h P+Q
δ4 = EAc N+ 2 is the axial deflection of column due to the concentrated
load transferred from the longitudinal girder (N).

(P + Q + 2N) + Wf
δs = (2.14.19)
L f B f cu
is the elastic deformation of soil in vertical mode.

Here, P = concentrated load from the machine; Q = UDL of the cross beam (qL);
q = self weight per unit length of the cross beam; N = concentrated load on the
column; Wf = weight of bottom slab + half the weight of the columns; Ab = area of
cross section of the beam; Ac = area of cross section of the column; Ib = moment of
inertia of the beam; Ic = moment of Inertia for the column; E = dynamic modulus of
elasticity of the frame; h = effective height of the column; L = effective length of the
cross beam; Lf = length of the foundation; Bf = width of the foundation and cu =
co-efficient of elastic compression of the soilψ = Ib h/(Ic L).
The fundamental frequency in vertical direction is then given by

fv = 30/ δv cpm. (2.14.20)
Considering n number of cross frames, in horizontal direction, Major followed the

same procedure of Barkan, as shown in Figure 2.14.6 like idealizing the top deck as
45 This is a very interesting proposition of adding elastic deformation of the soil directly to the structure
just note it for the time being we will discuss more about it later at appropriate time.

rigid in its own plane and considering an eccentricity e between the center of mass and
center of stiffness he arrived at an expression
1
2 n
2 Khi Ih
3
fn h = 30 α0 ± α02 − i=1 cpm (2.14.21)
n
i=1 Wi Jφ
h (2+3ψ) 3
where, Khi = lateral stiffness of the ith frame i; and Khi = δ1hi where δhi = 12EI c (1+6ψ)
;
Wi = total weight of the ith frame plus weight of the machine plus weight of
the
n transverse beam and the longitudinal beams; Jφ = mass moment of inertia ∼ =
W X 2 ; X = distance of weight W from the resultant center of mass point G46 ;
i=1 i g
gi
Ih = ni=1 Khi Xhi 2 ; X = distance of each frame from the center of rigidity H,
h
n
1 2 ni=1 Khi i=1 Khi Ih
and α0 = e + n + . (2.14.22)
2 Jϕ i=1 Wi Jφ
3 Calculation of amplitude
We had seen earlier in Chapter 5 (Vol. 1) that under harmonic load the amplitude of
vibration is given by the expression
P0
k
sin ωm t
x̄max = (2.14.23)
(1 − r2 )2 + (2Dr)2
where
√ r = ωm /ωn and D = c/cc with, cc = Critical damping of the system and is
2 mk.
For sin ωm t = 1, we have
P0
k δst
x̄max = ➔ x̄max = (2.14.24)
(1 − r2 )2 + (2Dr)2 (1 − r2 )2 + (2Dr)2
Here, we introduce a term called logarithmic decrement given by ∇ = 2πD, where

∇ = Logarithmic decrement; D = damping ratio.
Major replaced the 2D by ∇π and defined amplitude of vibration as
δst
➔ x̄max = ∇ 2 . (2.14.25)
(1 − r2 )2 + π (r)2
46 Refer to Figure 2.14.6.

4 Under-tuned foundation
For under tuned foundation when ωn < ωm during starting and stopping of machine,
there will be a case, when for a fleeting moment ωn = ωm and as such the frequency
ratio (r) will be equal to 1.0 for that instant.
During this point considering r = 1 the amplitude of vibration reduces to
x̄max = π δst /∇ (2.14.26)
Major has suggested that the logarithmic decrement (∇) be taken for concrete as
0.4 when the maximum amplitude becomes, x̄max = 7.85 δst .
5 Over tuned foundation

For over tuned foundation when ωn > ωm the maximum amplitude can be found out
from the expression
δst
x̄max = ∇ 2 (2.14.27)
(1 − r2 )2 + π (r)2
where δst = δv or δh as the case may be.
6 Calculation of unbalanced centrifugal force

For under tuned foundation (ωn < ωm ) the centrifugal force Ci is given by
2
ωn
Ci = αR (2.14.28)
ωm
For over tuned case (ωn > ωm ) the centrifugal force Ci is given by
Ci = αR (2.14.29)
where the value of α is as given in Table 2.14.1.

While in the vertical direction, Major considered the deflection of individual frame
which when multiplied by the above factors gives the dynamic amplitude under
transient condition.
In horizontal direction a stick model has been considered, where the stiffness of all
frames are clubbed together to arrive at a unique value of amplitude.
Table 2.14.1 Values of α.
Sl. No. α rpm rating of machine
1 0.2 ≥3000
2 0.16 1500
3 0.1 750

Since individual amplitude of each frame is also necessary he approached the

problem in the following manner.

If C = N i=1 Ci denotes the total centrifugal force in horizontal direction then the
centrifugal force on each individual frame is distributed in terms of their individual
stiffness
K CK X
Ci = C N hi + e N hi hi (2.14.30)
2
i=1 Khi i=1 Khi Xhi
Here e is the distance between center of rigidity Xhi and center of the resultant of

the horizontal dynamic forces, C = N i=1 Ci .
Once Ci is obtained the deflection of the ith frame is obtained from the expression
Ci
δhi = (2.14.31)
Khi
with the value of δhi , the amplitude of vibration in horizontal direction is obtained
from the expression
δhi
ahi = ∇ 2 (2.14.32)
(1 − r2 )2 + π (r)2
Major states that since the structure is usually more flexible in transverse direction
and considering the high speed of the machine is usually under tuned in this direction
and as such it is a common practice to consider for horizontal mode
ahi = 7.85δhi . (2.14.33)
7 Dynamic forces
The dynamic forces to be accounted for in structural design of the frame have been
expressed by Major as follows:
To account for idealization made in calculation of natural frequency it is suggested
to correct the calculated natural frequency by a term, fn = fn (1 ± α), where α is a
correction factor and may be considered as 0.2.
For under-tuned foundation (fn < fm ) plus signed should be considered while for
over tuned foundation minus sign to be considered47 .
When fn lies between 1+αfm fm
and 1−α , then fn = fm .
47 This actually means Major is assuming that the frequency calculation could be out from actual by (±)
20% and based on the correction factor is actually trying to develop a conservative estimate of the
dynamic force.

Table 2.14.2 Values of dynamic force under various conditions.
Operating frequency
Case of machine (rpm) Dynamic force Remarks
2
fn < fm
fn
3000 F = 16R f Fmax = 16R
m 2
fn
1500 F = 12R fm Fmax = 12R
2
f
750 F = 8R fmn Fmax = 8R
fn > fm 3000 F=
2Fmax
Fmax = 1.0R
2 2 2 2
1− 2 + ∇
fm fm
f n π f 2
n
1500 Do Fmax = 0.8R
750 Do Fmax = 0.5R
fm
1−α < fn < fm
1+α 3000 F = 16R
1500 F = 12R
750 F = 8R
R = rotating weight on the frame.
Based on the above, Major suggested Table 2.14.2 for calculating the dynamic
forces. For vertical dynamic force that acts on the center of the transverse beam the
rotating weight on the beam only should considered as the expression R.
For calculation of the horizontal dynamic force in transverse direction total rotat-
ing weight on the transverse beam plus rotating weight on the longitudinal girder
transferred to the column shall also be considered while calculating the term R.
2.15 DYNAMIC SOIL-STRUCTURE INTERACTION MODEL

FOR VIBRATION ANALYSIS OF TURBINE
FOUNDATION
We present here a further modification of Major’s method considering the effect of

underlying soil on the vibration analysis of turbine foundation. The history behind
its evolution is quite interesting and would not be possibly out of context to share
the background with you. Till 1980’s Indian power industry was mostly restricted to
Turbine units having capacity up to 210 MW. These turbines were all supplied by
BHEL48 and were prototype of LMW models used in the USSR. While the machines
were quite massive and sturdy, the foundation system for these types of machines was
usually wall mounted and not frame type. In reality they were actually massive RCC
blocks having cutouts in it for laying the piping and fixing other sundry fixtures. They
were generically short in height and because of their massiveness and immense rigidity
these foundations were mostly over tuned. Thus conventional theories as proposed by
Barkan/Major justified their analysis quite well.
48 Bharat Heavy Electrical Limited, they are the premier Turbine manufacturing company in India.

The first 500 MW turbines supplied in India were from Siemens KWU49 and had
a complete different structural configuration from the erstwhile models that were in
vogue in the industry. The top deck level was much taller (about 16.0 m); they were
much slicker and called for much more space below the turbine and truly represented
a framed structure having under tuned characteristics.
In the meantime a number of turbines in operation in various parts of India were
monitored for vibration and it was observed that some of them which were designed
as over-tuned system still showed transient excitation during start and stop of the
machine (meaning thereby that they were showing under tuned characteristics).
The question was why it was happening so? It was realized that it was possibly the
soil below the bottom mat which was participating in the vibration and changing the
characteristic behavior of the foundation.
Wedpathak, Pandit and Guha (1977) conducted vibration monitoring on various
TG foundations at different power plant in India and showed that there existed
a considerable variation in amplitudes observed in the field and those calculated
theoretically.
The above discrepancy suggested that there was definitely a necessity to arrive at a
more realistic mathematical model to predict the response of the turbine foundations.
It also proved that the assumption made in conventional analysis by Barkan and
Major, that making the bottom raft thick- nullifies any participation of the under-
lying soil in the vibration may not be true in all cases. Especially for 500 MW
class of turbine where to suppress the vibration of the underlying soil the thickness
of the bottom, mat would have to be so thick that the foundation could become
prohibitively expensive.
Moreover, due to their height and slenderness in transverse direction it was realized
that translation in this direction will also induce a coupled rocking mode in the trans-
verse plane which was not accounted for in the conventional method. Considering the
inadequacy in the conventional method in the context of present day class of turbines,
we started our investigation into this problem to arrive at a more rational model where
the contribution of the soil in vibration of such frame foundations can be catered for.
While it was always possible to solve this problem based on FEM50 , we realized that
prior to that one should have the feel as to how the system is behaving and moreover
considering the expense incurred for doing a major FEM analysis in terms of man hour
spent in data generation, data input, checking the output and result interpretations,
was there an alternative model which would give reasonable results if needed to be
done manually or use computer to a minimum?
That was the philosophy based on which we started our quest for a solution and
the outcome is what we would like to share with you.
Unlike Major’s model we consider here a three-mass lumped system as shown in
Figure 2.15.1. We use here a judicious mixture of Barkan and Major’s method and
couple the soil springs based on Richart or Wolf’s formulation.
49 The first Siemens machine of 500 MW was supplied to Trombay (Tata Electric) and the second to
Singrauli NTPC.
50 This we had tackled too and will be presented at a later stage.

m3
z3
k3
m2
z2
k2
m1
k
z1
Figure 2.15.1 2D-Mathematical model for soil-structure interaction (Vert. mode).
Here for n number of cross frames, we have

• m3 = ni=1 (Concentrated mass of the machine carried by the transverse girder
+0.45 times self weight of the transverse girder)
• m2 = ni=1 (0.25 times the Mass of the transverse girder + mass from the longitu-
dinal girder including machine weight if any transferred to the cross frame + 0.30
times the mass of the column)
• m1 = mass of the bottom slab + mass of the condenser + ni=1 (0.3 times the
mass of the column) + mass of the soil participating in the vibration.
For spring k3 , for the beams we have

n
L3 (1 + 2ψ) 3L
n
1
δv = + and k3 = (2.15.1)
96EIb (2 + ψ) 8GAb δv
i=1 i=1
n 2Ac Ec
where, k2 = equivalent spring constants for the columns @ i=1 h
; k1 =
equivalent spring for the soil obtained from Richart or Wolf’s formulation51 and G =
dynamic shear modulus of concrete @ 0.5E.
Applying D’Alembert’s equation free vibration of the system can deduced as
⎡ ⎤⎧ ⎫ ⎡ ⎤⎧ ⎫
m1 0 0 ⎨ẍ1 ⎬ k 1 + k2 −k2 0 ⎨x1 ⎬
⎣ 0 m2 0 ⎦ ẍ2 + ⎣ −k2 k2 + k3 −k3 ⎦ x2 = 0 (2.15.2)
⎩ ⎭ ⎩ ⎭
0 0 m3 ẍ3 0 −k3 k3 x3
51 Refer section on block foundation for the values of the soil springs.

Calculation is now quite straight forward for frequency analysis based on eigen
value solution.
Instead of soil, if the foundation is resting on piles then we can straight use pile
springs based on Novak’s formulation or other methods as cited previously and use
this spring as the spring k1 .
The above matrix on expansion will give an equation of third degree whose
characteristics roots will give the eigen values of the above problem.
2 Calculation of horizontal frequency

Based on the discussion in the preceding page, it was highlighted that major lacunae
lies in this mode. While conventional analysis considers translation and rotation in
plan it does not consider the rotation in elevation which will also get coupled when
the height of turbine deck is high. We consider all these aspects in our formulation
and develop a matrix of order 4 × 4 that we feel takes into cognizance all the short
comings of the conventional method we discussed. We show hereafter an analytical
model conceived to cater to all the above aspects.
While the conventional analysis considers lateral translation x and rotation φ in
plan it considers the bottom raft to be completely rigid and the soil has no effect on
the vibration.
Since the major horizontal motion of the machine is in the transverse direction we
have added additional degrees of freedom
• u which is the translational displacement of the foundation.

• For turbines of capacity 500 MW and above as the height h of the column is quite
large this will also induce a rocking of the foundation (in transverse plane) and
assigned a value θ .
Thus while the conventional analysis has two degrees of freedom x and φ, in our
model shown in Figure 2.15.2, we have four degrees of freedom, namely x, φ, u, θ.
Here, Kx = translation spring value of soil; Kθ = rocking spring value of soil; m0 ,
Jφ = mass and mass moment of inertia of top deck + Machine; mf and Jθ = mass and
mass moment of inertia of the bottom raft.
To arrive at the equation of motion based on D’Alembert’s principle will be quite
difficult as the coupled motion is quite complicated.
So to derive the equations we use the famous Lagrange’s equation from the energy
principle when
n

d ∂T ∂T ∂U
d(T + U) = − + dqi = 0 (2.15.3)
i=1
T = f (q1 , q2 , q3 . . . . . . . qn ; q̇1 , q̇2, q̇3 , . . . . . . . . . . q̇n ) and

U = f (q1 , q2 , q3 , . . . . . . . . . , qn ) (2.15.4)
The kinetic energy, T for the system is given by
1 1 1 1
T= m u̇2 + Jθ θ̇ 2 + m0 (u̇ + ẋ + hθ̇ + eφ̇)2 + Jφ φ̇ 2 (2.15.5)
2 f 2 2 2

m0, Jφ
Kh
h
Kx K
u
mf, J
Figure 2.15.2 2D-Mathematical model for soil-structure interacton (Horz. mode).
The potential energy, U is given by
1 1 1 1
U= K x u 2 + K θ θ 2 + K h x 2 + Iφ φ 2 (2.15.6)
2 2 2 2
Differentiating,
∂T
= mf u̇ + m0 (u̇ + ẋ + hθ̇ + eφ̇) and
∂ u̇
d ∂T
= mf ü + m0 (ü + ẍ + hθ̈ + eφ̈)
dt ∂ u̇
∂T (2.15.7)
= Jθ θ̇ + m0 h(u̇ + ẋ + hθ̇ + eφ̇) and
∂ θ̇

d ∂T
= Jθ θ̈ + m0 h(ü + ẍ + hθ̈ + eφ̈)
dt ∂ θ̇
Similarly

d ∂T
= m0 (ü + ẍ + hθ̈ + eφ̈) and
dt ∂ ẋ
(2.15.8)
d ∂T
= Jφ φ̈ + m0 e(ü + ẍ + hθ̈ + eφ̈)
dt ∂ φ̇

For potential energy, we have
∂U ∂U ∂U ∂U
= Kx U; = Kθ θ ; = Kh x and = Kh e2 φ + Iφ φ
∂u ∂θ ∂x ∂φ
Substituting the above values in the equation
n

d ∂T ∂T ∂
d(T + U) = − + dqi = 0
i=1
and writing in matrix form we have
⎡ ⎤⎧ ⎫
m0 m0 e m0 m0 h ⎪
⎪ ẍ ⎪
⎢ m0 e ⎨ ⎪ ⎬
⎢ Jφ + m0 e2 m0 e m0 eh ⎥ ⎥ φ̈
⎣ m0 m0 e m 0 + mf m0 h ⎦ ⎪ Ü ⎪
⎪
⎩ ⎪ ⎭
m0 h m0 eh m0 h J θ + m 0 h2 θ̈
⎡ ⎤⎧ ⎫
Kh 0 0 0 ⎪ ⎪ x⎪
⎢0 ⎨ ⎪ ⎬
Kh e2 + Iφ 0 0⎥ ϕ
+⎢
⎣0
⎥
⎦ =0 (2.15.9)
0 Kx 0 ⎪ ⎪U⎪
⎩ ⎪ ⎭
0 0 0 Kθ θ
Equation (2.15.9) gives the complete free vibration equation of motion for the
turbine foundation system considering the soil springs the translation and rocking
modes.

N
N
1 h3 (2 + 3ψ)
Here, Kh = Khi = where,δhi = ;
δhi 12EIc (1 + 6ψ)
i=1 i=1

N
N
2 2
Jφ = mi Xgi and Iφ = Ki Xhi (2.15.10)
i=1 i=1
Before we go further a few things needs to be noticed
• The matrix is real and symmetric.

• The equations are dynamically coupled thus the reference co-ordinate is the center
of rigidity and not center of mass as is the case with D’Alembert’s equation.
• Due to dynamic coupling, the mass matrix is a full matrix while stiffness and the
damping matrix would remain in uncoupled form.
Expansion of the eigen value matrix will give a fourth order polynomial whose roots
can be found based on Bairstow’s method or else can be very easily solved based on
software tools like MATH CAD/ MATLAB etc.

3 Calculation of amplitude of vibration

We will use here the generic modal response technique to obtain the amplitude of
vibration using the orthogonal property of the matrix.
Moreover since the Turbine once started will continue to operate for a long time as
such the steady state response is critical and we shall ignore the transient part. Thus
in the vertical /horizontal direction, we have
[M]{Ẍ} + [C]{Ẋ} + [K]{X} = {P(t)} (2.15.11)
where, [M] = mass matrix of the system; [C] = damping matrix of the system; [K] =
stiffness matrix of the system, and {P(t)} = P sin ωm t/P cos ωm t the dynamic force
with sine or cosine function for the vertical or horizontal case respectively.
Now considering the operation,
[φ]T [M][φ]{Ẍ} + [φ]T [C][φ]{Ẋ} + [φ]T [K][φ]{X} = [φ]T {P(t)}[φ] (2.15.12)
If the total numbers of degrees of freedom is j say then we have j numbers of

uncoupled equation depicted by

j
ξ̈i + 2Di ωi ξ̇i + ωi 2ξi = p0i (t) (2.15.13)
i=1

j=3 p0i sin ωm t
when ξi = in the vertical direction. And
i=1 (1 − r2 )2 + (2Dr)2
(2.15.14)

j=4 p0i cos ωm t
ξi = in the horizontal direction.
i=1 (1 − r2 )2 + (2Dr)2
Once the displacement in uncoupled form are known the global amplitude is found
out based on the expression, {X} = [φ]{ξ }.
The net amplitude at the top deck, is given by the expression
xinet = xi + Ui + Xhi ϕ + hθ 52 (2.15.15)
It would be worth now to objectively evaluate the advantage of this method.

Some of the advantages that can be attributed to this model are:
• It takes all the fundamental degrees of freedom considered by the conventional

method and also takes into consideration the effect of the soil in vibration analysis.
52 We are not trying to take a short cut. We will further elaborate the whole technique including the
complete design based on a suitable problem hereafter.

• It can take both soil or pile springs (with and without embedded effect) as an input
to the overall matrix.
• The calculation though more intense than conventional method it is yet amenable
to manual computation and gives the engineer a first order feel as to how the
coupled soil-structure is behaving under dynamic loading.
• It will surely give quantitatively a clear idea as to how much is the effect of soil
on the overall vibration vis-a-vis fixed base frequency when the effect of soil is
neglected53 .
• It will also help in taking a better decision if further elaborate analysis based on
3D space frame model is envisaged or not.
To people of orthodox school as well as the computer buffs54 we can assure that
this technique works quit fine.
This technique has been put into practice for a boiler feed pump framed foundation
for a power plant in India and we are happy to inform that it has been operating
smoothly without any problem for more than 15 years (Chowdhury and Som 1993).
2.16 COMPUTER ANALYSIS OF TURBINE FOUNDATION

BASED ON MULTI DEGREE OF FREEDOM
In this section we discuss the method of analysis and design of turbine foundations
considering it as a frame having multi degree of freedom through computer55 .
In this case the steps followed for analysis of the frame foundation is as follows:
The system is broken up into three parts as shown
• The super structure

• The raft
• The soil
We basically use here the concept of finite element to solve the above problem.
Though application of finite element is more appropriate for continuum, however
basic principle of its application is well valid for this case also.
Shown in Figure 2.16.1 is a typical conceptual model of a turbine foundation resting
on a bottom raft which in turn is resting on soil.
53 If you are solving the problem in MATHCAD/MATLAB just put Kx = 1020 and Kθ = 1020 this
will effectively make the problem a fixed base one. Else delete the rows and column in the matrix
pertaining to the soil degrees of freedom and reduce it to a 2X2 matrix having x and φ as the active
degrees of freedom.
54 Whose staple diet is a problem having 1000 degrees of freedom. Anything less than that is
surely crude!
55 It is not that we would like to continue our designs based on a paper, pencil and a calculator at best. At
the door of the 21st century we do not want to carry the stigma of being Rip Van Winkle though we
confirm that we discourage the use of software as a black box.

Figure 2.16.1 3D computor model of a turbine frame with bottom raft and soil spring.
We discuss below step by step the concepts underlying the development of its
mathematical model for analysis in computer.
1 The super structure
What element do I choose and how many nodes do I consider?

The intuitive choice for the super-structure is obviously to model it as a space frame
where the beams and columns are idealized as beam elements having six degrees of
freedom at each node. But for modeling a turbine foundation frame there is a difference
with normal building frames.
Mathematical model for the beam and column are usually taken at the center line
of the element as shown in Figure 2.16.2.
Based on Figure 2.16.2, during computer analysis, the moment and shear output
will be given at the chosen nodes. For normal building frame this does not digress
from the reality much for the dimension of the columns are small. However for turbine
foundation the columns are of large dimension (usually they are about 1500/2000 mm).
During design of beams since we know that the design bending moment at support
is to be taken at the face of the column, the large dimension of the column makes
a significant reduction in the design moment of the beam at the support. The major
advantage is that it helps in reduction of congestion of reinforcement at the beam
column junction.
As such to correctly predict this phenomenon the model should consist of three
nodes instead of one connected by rigid links as shown Figure 2.16.3.

Node (Typ.)
Actual Frame
Idealized Model
Figure 2.16.2 Idealized model of a normal frame.
Moment based on one node at C/L of beam column
Design moment at column face

Column Node Based on three node concept
Beam Node
Rigid Link
Bending Moment profile
Figure 2.16.3 Typical connection of beam column junction with rigid link.
Note: In some software packages this may also be input as master and slave option where
the beam node is usually taken as the master and the column node as the slave node.
For the beam elements as the span by depth ratio is significant it is preferable to
consider the shear deformation of the girder during the analysis.
The loads that are induced by the machine to the deck are mostly transferred through
the bearing/sole plate. The sole plates are not necessarily always co-aligned with beam
center line. Thus to simulate this situation two of the following techniques could
be used.

F
F
Sole plate Link
C/L
Beam
T=FxD d
Figure 2.16.4 Vertical load acting on sole plate eccentric to the center line of the beam.
• By defining the load with additional torsion about the center line of the beam
based on the eccentricity between the bearing plate and center line axis of the
beam
• Providing node at the point of incident of the load and connecting this point to
the mathematical model by a rigid link as shown in Figure 2.16.4.
2 How many nodes do I consider?

Intuitively the primary choice of nodes will be the beam column junctions. On identify-
ing these nodes we further break it up into two additional nodes based on the concepts
as mentioned above. Other than this points at which direct load is transferred to the
girders nodal points are to be considered also.
For members under complex loading in span the number of nodes to be provided
for each beam member should be sufficient to plot the bending moment and the shear
force diagram.
For dynamic analysis enough nodes should be considered along the length of the
beam and column so that all the modes having a natural frequency less than or equal
to the operating frequency of the machine are simulated. The lower rigid body mode
of the top deck as a unit is not affected significantly by the number of nodes along the
length of the beams.
However higher modes simulating the differential deflection of the top deck are
affected by the distribution of nodes. If not modeled with enough nodes these modes
may be entirely missed leading to an incorrect result.
The suggested number of nodes n, to be placed along the length of the span is given
by the larger of the following two values56 :
L √ m 14 1 Lωm m 12
n≥ ωm :n≥ + (2.16.1)
π EI 2 π EA
56 The expressions are derived from frequencies of a simple supported beam in flexural and axial mode.
The basis of this expression is that if the nth natural frequency of the beam is at or below the operating
frequency of the machine then at least n mid-span nodes will be required to calculate the n modes using
the discrete model.

where, L = span of the member; E = modulus of elasticity; m = mass per unit length;
ωm = operating frequency of the machine in rad/sec; I = moment of inertia about the
beam about its weaker axis of bending, and A = cross sectional area of the beam.
The nodal mass may be calculated either based on lumped mass approach or con-
sistent mass approach (Archer 1963). The consistent mass approach accounts for the
distributed mass and variation of deflection along the length of the beam.
However, one major disadvantage with the consistent mass matrix is that it is a full
matrix in contrary to lumped mass which is a diagonal matrix and thus calls for more
computational effort.
It has been observed that the natural frequency obtained by consistent mass
approach is more accurate than lumped mass approach though the difference may be
small for most of the practical problems. For practical analysis of Turbine foundation
considering masses lumped at the nodes is the common industrial practice.
Once the beam elements and the nodes are chosen and their properties like moment
of inertia and sectional area etc are provided as input, the computer generates the local
stiffness matrix of each beam (of size 12 × 12) and then based on their direction cosine
transfers the local stiffness matrix into the global axis and assembles them to form the
global stiffness matrix of the superstructure.
3 The foundation raft

The foundation raft usually consists of a slab resting on soil or pile which is about
2000/2500 mm thick.
What element to use which would be optimal as well as provide the best result is
still a debate among the finite element analysts.
Some literature (Design Criteria for Turbine Generator Pedestal, 1970) recommend
to model the raft as plate bending elements while the others (Arya et al. 1979) insist
on to model it as beam elements supported on soil springs. While some advocate to
use even 8 nodded brick element to model the raft. With such controversies prevailing
on this issue it would possibly be worthwhile to evaluate the pros and cons of each of
these elements.
Plate elements
Plate elements apparently look to be a good choice for physically, it best reflects the
continuum. But as far as mathematical formulation of plates based on Finite Element
formulation is concerned the best available element for plate bending considering its
numerical convergence is the Discrete Kirchoff Triangular (DKT) plate element. The
stiffness matrix formulation of DKT plate element is based on the thin plate theory
having three (two translation and one rotation) degrees of freedom per node. The
basic idealization is that the thickness of the plate is negligible in comparison to its
plan dimension and as such the effect of transverse shear acting along the edge of the
plate is neglected.
For the turbine raft having thickness of 2000/2500 mm it is evident that the thickness
of the raft is quite large and as such it would not be perhaps prudent to neglect the
thickness vis-à-vis the effect of shear strain energy contribution of the overall system.
Which catapults the problem from Kirchoff-type of thin plate to Mindlin-Reissner
type of thick plate where solution is sought taking into consideration the shear
deformation along the edge of the plate.

Though many researcher have tried to formulate these type of plate based on FEM
most of them suffers from one technical snag or other, namely
• Failing to pass important patch test

• Spurious zero energy mode
• Being sensitive to geometric distortion and meshing.
These can lead to poor solutions and even results which are unacceptable at times
(Kardestuncer 1987). Hence, without a proper mathematical formulation of the thick
plate in hand, specially the numerical problems it can create while seeking solution to
the problem, we would suggest not to use such elements in modeling this problem.
Brick elements
Brick elements could also become a plausible choice for modeling the turbine raft.
From convergence point of view brick elements are stable and have been successfully
adapted to solve different class of problems in fracture, rock and fluid mechanics.
However it has been observed that the eight nodded brick element usually have poor
approximation capability and higher order elements having 16 or 24 nodes are usually
used for efficient solution.
But use of such higher order elements calls for a much more expensive analysis in
terms of computer time, data preparation, input, output etc and is usually not essential.
Besides this brick element suffers from one serious lacunae in terms of design. Brick
elements in most of the commercially available software give output in terms of normal
and shear stress parameters. While this is fine in terms rock or fracture mechanics
problem where design check is done against allowable stresses, for the turbine raft
design we are basically looking for output in terms of moment, shear and torsion.
To back calculate these parameters from the computer out put and subsequent
interpolation to arrive at the design moments, shears etc can be extremely tedious and
chances are very high that the engineer assigned to perform this task gets lost in a maze
of numbers and gets totally confused.
For eigen-solution though use of brick element is OK we would however suggest
users the use of brick elements for design purpose with caution for the enormous
difficulty one could face in back calculating the stress output in terms of moment,
shear and torsion.
Beam elements
This brings us to the last of element in use, the beam element to model the turbine raft.
From convergence and correctness of results we had already discussed in quite detail
in Chapter 2 (Vol. 1) that if properly modeled beam elements gives results which is
very close to plate elements in simulating a raft problem57 .
Moreover for derivation of stiffness matrix irrespective of the methodology used like
moment area theorem, strain energy method or numerical methods like finite element,
the results converge to an exact solution.
57 Refer Chapter 4 (Vol. 1) on Static soil structure Interaction where we have discussed in detail the use of
beam vis-a-vis plate bending elements for simulation of rafts resting on soil.

Even when the stiffness formulation takes into consideration transverse shear defor-
mation unlike thick plate element the formulation is consistent and conforming. Finally
computer output is in terms of moment shear and torsion directly and may be directly
used for re-bar calculation without having recourse to deriving them from stress output
unlike brick elements.
Moreover if we take the elements with reasonable mesh refinements including the
transverse shear deformation into cognizance58 , we can approach a state where the
energy compatibility in terms of external work done and consequent strain energy
induced can be well satisfied.
Thus in terms of ease of use as well as convergence of results beam elements do
make a very attractive choice59 .
For the raft, as the thickness is significant considering the shear deformation
characteristics is a must for maintaining the strain energy compatibility.
4 The soil
The basic soil parameter which needs to be known to mathematically model the soil is
dynamic shear modulus (G)60 . The soil being a continuum itself can either be modeled
based on FEM as 3D brick elements61 , 2D plane strain elements or discrete springs.
For modeling the soil, the choice is again multiple. However as soil itself is an
infinite domain successful application of FEM has been mostly in cases where the
problem could be simulated by a two dimensional model where the soil itself has been
modeled as plane strain elements or infinite finite elements to arrive at a meaningful
result.
Rarely, we have come across cases where in practical problems pertaining to soil
has been modeled in 3D elements for the effort and cost in terms of man-hour and
output interpretation can make the analysis prohibitively expensive.
For the particular case of turbine foundation analysis as we are interested to know
more about the behavior of the frame and the bottom raft rather then the intricate
behavior of the soil itself, the common practice is to model the soil as frequency
independent linear springs based on Richart or Wolf’s springs as described in section
of block foundation.
For practical application this has been found to be quite adequate. More sophisti-
cated model based on frequency dependent complex stiffness is usually not warranted
in this case.
Depending on the soil stiffness and the stiffness of the raft a correction to the spring
needs to be done for correct evaluation of the response62 .
Once the spring values are evaluated they are connected to the node of the raft ele-
ment based on usual finite element procedure to arrive at the complete stiffness matrix
58 Whose contribution becomes significant as the ratio of span by depth reduces.

59 For protagonists of classical school this is to inform that many Turbines raft has been modeled as beam
elements which have been analyzed, designed and constructed and has successfully stood the vagaries
of nature and the test of time.
60 We have dealt this topic in detail in the Chapter 1 (Vol. 2).
61 Refer Chapter 4 (Vol. 1) where we have discussed such problems in detail in terms of static loading.
62 This we have dealt in detail in the chapter 1 (Vol. 2).

of soil foundation system. A typical example is shown in Chapter 2 (Vol. 1) where we

have solved a table top centrifugal compressor foundation based on this method.
5 The machine
Do we model the machine resting on the top deck also in our analysis?
A debate which has been in the profession for quite some time and we do not want to
pass a sacrosanct sermon on this issue.
However our objective analysis of this Shakespearean dilemma63 is as follows:
For Turbines of low capacity (<350 MW) the foundations are usually designed
having over tuned characteristics. Moreover as the overall dimension of the machine
is also relatively smaller, as such it is reasonable to consider the whole turbine and the
generator as a rigid mass whose inertial contribution as a lumped mass is taken into
cognizance in the analysis only.
However with increasing demand for energy, power manufactures are coming out
with Turbines having higher and higher capacities.
This has made the overall dimension of the turbine larger and the foundation size
have also increased and have made it flexible and more susceptible to dynamic excita-
tion. For the equipment, the main shaft which connects the turbine and the generator
has become longer, thus flexible, and with increase in the operating speed a slight
imbalance in the rotating mass can induce significant dynamic load on the shaft and
also the over all deformation of the soil, raft and the frame (specially in the flexural
mode) can generate a phenomenon which is know as the bowing of the turbine shaft.
Bowing or bending of the shaft about its center line axis can create damage to
the machine components, induce large forces at the bearing and can also reduce the
operating efficiency of the turbine.
Thus for larger turbines (>500 MW) it would be possibly justified to consider the
machine as an integral part of the analysis too.
For such consideration an elaborate Finite Element modeling of the turbine and the
generator is usually not warranted a simple mathematical model consisting of masses
lumped at strategic nodes connected by beams, springs, rigid links etc would usually
suffice64 .
2.17 ANALYSIS OF TURBINE FOUNDATION
2.17.1 The analysis

The analysis is usually done in the computer in four steps:
• Dynamic analysis to calculate the natural frequencies of the system to ensure that
it is out of tune to the operating frequency of the machine by ±20%.
63 To be or not to be . . . .
64 At this point we would strongly recommend you to take help of your equipment specialist while modeling
the equipment connected to the super-structure.

• Calculation of the dynamic amplitude to check that the same are within the
acceptable limits as prescribed in the code or as pre-defined by the equipment
supplier.
• Earthquake analysis if the same is perceived critical for the foundation.
• A pseudo–static analysis to obtain the design Moment, Shear and Torsion induced
in the members check the stresses induced in the different structural elements like
beam column and slabs.
2.17.2 Calculation of the eigen values

For calculation of the natural frequencies or the eigen values the first choice the user
has to make as to how many modes do I consider for the analysis?
First three modes, six modes or twenty modes . . . We have heard variety of such
numbers65 . Unfortunately, none of the answers are universally correct, for how many
modes are significant for the analysis varies from case to case and it also depends on
what we are looking for in terms of cases like checking the resonance, checking the
transient response or checking the response against earthquake.
The most rational basis of choice of modes would be based on modal mass participa-
tion factor66 which should always be the basis of arriving at the number of significant
modes to be considered for dynamic analysis when we are doing a resonance check.
As a first step start with say five or six significant modes check the frequency with
the operating speed of the machine and also at the same time check the modal mass
participation factor for these modes67 . If the mass participation is of the order of say
50 or 60% it is evident more number of modes need to considered.
Number of modes that excite at least 95 to 99% of the mass should be the basis of
number of significant modes to be considered in the analysis. The reason is as explained
hereafter.
Suppose for the first six modes we find the natural frequency of the system is below
the operating speed of the machine by 20% but it has only excited say 60% of the mass
while higher modes which are in the vicinity of the operating frequency has excited
say 89% of the mass (say the 9th or the 10th mode) it is obvious that these modes
will excite the structure much more and this we will completely miss if we restrict our
analysis to a preconceived six-mode analysis only.
The other advantage is that as the eigen values go on increasing with each mode
there will always be some value which would match or be very near to the operating
frequency of the machine. But, if nearly 100% of the mass has already participated
in the vibration in the earlier modes this will have no effect on the response of the
structure even though the frequency is in the vicinity of the resonance range.
However, this can only be predicted confidently provided you know exactly how
much mass has already participated in the vibration.
65 With comments such as “From my experience”, “Normal engineering practice”, and finally “From
previous experience” – from an engineer with 2 years of experience(!!!) etc. to name a few.
66 For details of modal mass participation refer to Chapter 3 (Vol. 2).
67 Most of the commercially available FEM and dynamic analysis software have this option as an output
for the user to check the mass participation in the X,Y and Z direction.

2.17.3 So the ground rule is. . .

Do not guess or start with the pre-conceived notion that “n” number of modes would
suffice check the modal mass participation factor and then decide.
To arrive at the eigen value vis a vis the natural frequency, though the basic equation
for the solution remains same i.e.
[K]{x} = ω2 [M]{x} (2.17.1)
computation of ω2 is surely not done in the way we have described in our ear-
lier chapter68 . For solution of eigen-values having large degrees of freedom special
numerical techniques are usually used.
When earthquake analysis is also critical, number of modes significant enough to
simulate the natural frequency to 33 Hz should be considered for the analysis.
2.17.4 Calculation of amplitude

Once the resonance condition is checked the next step is to ensure that the amplitude
of vibrations is restricted within the acceptable limits. The techniques explained earlier
based on modal analysis and orthogonal transformation69 is usually used to obtain
the amplitude of vibration under operating conditions.
It has been mostly seen that the response of the turbine foundation, especially con-
sidering the soil effect is usually not critical under the normal operating condition. It
is only during the start and stop of the machine when the system goes on transient
resonance that it shows significant excitation. As explained and shown earlier, in the
previous example of the 2D soil-structure interaction model, the best technique to find
such responses would be based on time history analysis where both the transient and
steady state response needs to be checked, to ensure that such fleeting response are also
within the acceptable limits as prescribed by the manufacturer or the code of practice.
2.17.5 Calculation of moments, shears and torsion

If earthquake load is not a governing case usually an equivalent static analysis will
suffice where an equivalent static load for the induced dynamic loads is obtained,
based on magnification factors as suggested in the code.
The table suggesting such factors has already been shown earlier while describing
Major’s combined method. IS 2974 usually recommends the use of this table to obtain
an equivalent static force for the rotating mass and advocates to add these loads to
other loads for an equivalent static analysis and structural design of the members.
68 Different techniques used for calculation of eigen values of the system having large degrees of freedom
has been dealt in detail in Chapter 5 (Vol. 1) and may please be referred to.
69 Refer the calculations for 2D model we have derived earlier or Chapter 5 (Vol. 1) for the details of such
analysis.

2.17.6 Practical aspects of design of Turbine foundation

Here we digress from the theoretical contemplation and enter the real world of prac-
ticing engineer to evaluate further what other parameters and decisions form the back
bone of a successful design of a Turbine foundation. As a first step we start with a
check list to see what inputs we require to start a design.
2.18 DESIGN OF TURBINE FOUNDATION
2.18.1 Check list for turbine foundation design

1 Does the drawing furnish the overall dimension of the machine?
2 Are the anchor bolt locations, size of the bolts (both diameter and length) and
details of how it should be anchored to the foundation furnished by the vendor?
3 Does the drawing supply the height at which the centre line of the shaft of the
machine is located from the bottom of the machine frame (which will be the top
of concrete or top of grout for you).
4 Does the drawing supply you with the operating speed of the machine or the range
which should be cleared during the design of the foundation?
5 Does the top deck need to support any pipes or valves on it other than the machine
itself?
6 If so are all the loads and locations of these valves and pipes are mentioned in the
drawing?
7 Does the drawing clearly mention the unbalanced mass, eccentricity or the
dynamic loads generated during the operation of the machine?
8 Are all the cut outs in the top deck including its size and location has been made
clear in the drawing?
9 Is the location of all embedded part on the top deck including their size, location
and thickness has been made clear in the drawing?
10 Is the location of the condenser support including the load coming from it is
available to you?
11 Is the Plan area of the working platform for accessing valves and for maintenance
is made clear?
12 Different load combinations for which the turbine foundation has to be designed
specially from mechanical considerations like short circuit moments, breaking of
impeller, Thermal differential etc has been furnished?
13 Finally has the equipment supplier defined any performance criterion which needs
to be met in terms of amplitude, frequency etc.
The importance of this has already been made clear previously in the chapter of
block foundations.
14 Allowable bearing capacity of the soil.
15 Dynamic shear modulus of the soil.
16 Grade of concrete to be used.
Once the above check list is satisfied the engineer starts his analysis with the tentative
sizing of the geometry of the super-structure.
The guideline furnished below, are suggestive as a first trial and the adequacy of the
same shall be checked against a thorough dynamic analysis.

• The designer should give enough thought to the sizing of the equipment, its size
and clearance requirements in terms of maintenance and access during operation.
The size of such access corridor should be clearly discussed with the equipment
vendor and also with the plant operation people to finalize the overall dimension
of the top deck.
• All columns should be sized in such a way that they are almost equally stressed
under vertical loads (i.e. σ = P/A shall be constant for all the columns as far as
possible). As a rule of thumb, the columns shall have load carrying capacity of
about six times the vertical load and shall be placed not less than 3.6 meters center
to center.
• The depth of the longitudinal and the transverse beam shall be one fifth the clear
span with the width equal to the width of the column. Care should also be taken
that if some anchor bolts are embedded in the beam the depth of the beam is
adequate for generating the full strength of the anchor bolts. The deflection of the
beam under static load shall be restricted to 0.5 mm.
• The turbine frame should in principle act as a rigid shear frame as such the flexural
stiffness of the top deck beams shall be two times the flexural stiffness of the
columns.
• The bottom of the raft shall not be placed above the level as suggested by the
geo-technical consultant where the thickness (t) of the slab shall not be less than,
t = 0.07L4/3 , where L is the average distance between columns.
• The mass of the top deck plus mass of half the length of the column shall not be
less than the mass of the supported turbine and its auxiliaries on the top deck.
• The total mass of the frame plus the raft shall not be less than three times the mass
of the machine.
• The stress induced in soil shall not exceed 50% of the allowable bearing capacity
of the soil.
• For foundations supported on piles the most heavily loaded pile shall not carry
50% of its allowable load.
• The center of resistance for the pile group or the soil shall not be more than 300 mm
from c.g. of the superimposed loads.
• The center of rigidity of the columns shall coincide with the c.g. of the equipment
plus the top half of the structural loads both in the transverse and longitudinal
direction. This shall be done based on the equations:

n 4
n
n 4
n
x̄ = x i Ix i Ixi , z̄ = zi I z i I zi (2.18.1)
i=1 i=1 i=1 i=1
where, x̄ = co-ordinate of the center of rigidity in longitudinal direction; z̄ =

co-ordinate of the center of rigidity in transverse direction; Ix and Iz = moment
of inertia of the columns, and n = number of columns.
• All columns should deflect equally in vertical, transverse and longitudinal direc-
tions as far as possible when subjected to equivalent static load with a limit on
deflection for all cases as 0.5 mm.

• Intermediate platforms are some times provided below the turbine deck for access
from bottom and maintenance. These platforms should preferably be placed below
the high pressure turbine and should be of RCC. The beams are usually of depth
varying from 0.9 to 1.2 meter with a slab thickness not less than 300 mm. During
computer analysis stiffening effect of such platforms on the superstructure shall
be considered in the analysis and it should also be ensured that the platform itself
is not in resonance with the operating speed of the machine.
1 Loads and load combinations for analysis
This we are going to deal in some detail. For unlike normal civil engineering structure
the turbine foundation is a very specialized structure where different types of loading
arise from the mechanical and electrical aspects of the machine.
If the engineer analyzing the foundation does not have a clear idea about these loads
he may land up with an analysis which could be deemed useless.
Irrespective of how sophisticated FEM package you use or use the most comprehen-
sive mathematical model if the loading input is not correct the result output is always
useless.
The different loads which come on the turbine are as discussed hereafter. While civil
engineers are quite comfortable with loading like Dead Load (DL), Live Load (LL),
Seismic load (SL) etc., our observation is that many of them are not very clear about
the typical loads which come on a turbine foundation like condenser vacuum loading
(CVL), normal torque loading (NTL) etc and how they could effect the behavior of
the foundation.
We break up the loading in three different categories: Civil Loads; Mechanical loads,
and, Electrical loads.
a Civil Loads
This is constitutes of the following:
1 Dead Load (DL)

As the name suggests this combines the self weight of all the frame members and weight
of the foundation.
2 Live Load (LL)

The live load includes those loads that vary in its magnitude and occurrence. The
normal practice is to consider a Live Load @10 kN/m2 , on the top deck for the analysis
and design.
If based on the maintenance concept it is expected that maintenance load and lay
down load shall also come on the top deck then they shall be considered as live load
in the design.
3 Wind Load (WL)

This is usually not considered in the analysis of Turbine foundations for in most of
the cases the TG foundation is placed inside a building (the power house) where all

the wind load is transferred to the ground through the power house structure itself.
There are exceptional cases only when it needs to be taken into cognizance70 .
4 Earthquake/Seismic Load (SL)

This could be the major design load if the power plant is being built in area prone to
major earthquakes.
Earthquake is itself a major topic of study and we will not go in detail here on this
issue for we have dealt this issue separately71 .
5 Loading due to Creep and Shrinkage (SCL)

This usually applies to RCC frames where after initial deflection the structure
undergoes deformation under sustained loading.
This time dependent deflection at the bearing location can be two or three times more
than the short term elastic deflection. However this phenomenon was not considered
earlier for design due to the following reason. For a typical coal fired power plant the
initial machine alignment use to occur at about 24 to 36 month after the foundation
has been constructed. By this time most of the deformation due to shrinkage and creep
would have taken place thus further deformations were negligible and had practically
no effect on the shaft alignment.
However under present scenario with demand in power on the rise globally the
turnkey contractors are expected to finish and hand over one whole plant in 20 to 29
months only. As such it is obvious now the loading on the turbine frame would come
much earlier when the secondary deformation effect of creep and shrinkage could be
significant at the bearing level and should be carefully evaluated.
b Mechanical Loads
1 Machine Dead Load (MDL)
This constitutes of the weight of the various turbine components and is usually termed
as the machine dead load. The turbine manufacturer in their equipment layout drawing
usually supplies these loadings and their locations.
2 Condenser Dead Load (CDL)

We had already explained earlier that the condenser is normally mounted below
the turbine top deck. Depending upon the supporting system used for installing the
condenser the loading induced on the foundation varies. The common practice for
installing the condenser is either of the two systems as discussed hereafter:
The condenser is spring mounted on the bottom raft while the top neck is rigidly
connected to the turbine exhaust nozzle. The springs are of adjustable type enabling
them to transfer specified loads to the turbine exhaust nozzle. They are also sometimes
70 There are cases where the turbine deck is spring mounted and rests on steel columns, which in turn is
connected to the power house structure. In such cases WL load has to be taken in consideration in the
analysis specially the load combinations. In such case usually a combined power house and TG frame
analysis is carried out.
71 For more detail on this issue refer to Chapter 3 (Vol. 2).

used to balance the loading eccentricity that can develop due to circulating water
pressure loads.
The condenser bottom is mounted on a rigid frame and an expansion joint is pro-
vided between the condenser and the turbine exhaust nozzle to relieve the thermal
expansion and variations in the condenser loads.
For condenser mounted on rigid frame the total weight of the condenser is trans-
ferred to the bottom raft. For spring-mounted condensers, it is mostly welded to the
turbine exhaust nozzle when the proportion of load that will be distributed between
the top deck and the bottom raft depends upon the stiffness of the spring and their
alignment. The equipment supplier usually supplies this loading.
3 Condenser Vacuum Load (CVL)

For condensers mounted on rigid frame we had already mentioned that an expansion
joint is provided between it and the turbine exhaust nozzle, for this the difference
between the atmospheric pressure on the casing of the turbine and the vacuum pressure
inside the condenser develops a force on the turbine. This load can be several times in
magnitude to the weight of the condenser itself and is transmitted to the foundation
through the turbine soleplates. The turbine manufacturer provides the distribution of
this loading.
For spring mounted condensers when the condenser is rigidly connected to the
turbine exhaust nozzle no vacuum load is transmitted to the turbine top deck.
4 Normal Torque Load (NTL)

The steam expanding within the turbine imposes a torque on the stationary casing
in the opposite direction of the rotation of the rotor. The magnitude of the torque
depends on the angular speed and the power output of the turbine. The equipment
vendor usually supplies this load in the vendor drawing as equivalent vertical loads on
the sole plate.
5 Other Equipment Loads (OEL)

Other than the turbine itself the foundation may support other equipment such as
turbine stops, control valves, interceptor valves, main steam pipeline hangers etc.
Thus additional dead loads from these, which are not included under the heading
MDL, shall also be considered in the design.
6 Thermal Load (ThL)

During operation of the turbine, temperature change of the turbine and the generator
causes expansion and contraction to take place resulting in various parts to slide. As
the progressive heating of the machine take place the turbine shaft expands, however
the expansion does not induce any loading on the foundation for the shaft is fixed
longitudinally by single thrust bearing when the shaft slides freely across the journal
bearings which are adequately lubricated.
Unlike the shaft during the heat build up in the system during operation the turbine
casing also gets heated and imposes thermal loading on the foundation. The transverse
beams usually support the sole plates supporting the high pressure and the intermediate
pressure turbine casing. The low-pressure turbine casings, the generator casing and the
exciter are supported on the sole plates of the longitudinal and the transverse beams.

During the heat built up the casing expands from their anchor points thus producing a
friction load. Though exact calculation of such forces is very difficult for it depends on
a number of factors however the common practice is to use the following simplifying
analysis in lieu of an exact analysis has been found to be adequate.
The total thermal loading in longitudinal or transverse direction is considered as the
vector sum of the forces acting on that direction.
The magnitude of the force on any sole plate is calculated as:
Force = μx
(Here x is sum of machine dead load, condenser load, normal torque load and piping
load if any), where, μ = coefficient of friction which varies from 0.2 to 0.5. The value
has to be confirmed with the turbine manufacturer.
At the startup condition the expansive load shall be taken as acting away from the
center line of the turbine while during shutdown it will considered acting towards the
center line of the turbine.
In case of the longitudinal expansion an approximation is made to the direction of
the force and the unbalanced force between two anchorage points, which prevent the
movement of the turbine, is applied as the concentrated load at the anchorage points.
7 Turbine Casing Pipe Load (TCPL)
The pipes connected to the turbine casing also induce loads to the foundation. The
turbine generator manufacturer to prevent distortion or overturning of the turbine
components specifies maximum loads. The turbine casing may be assumed to be rigid
and the forces are then calculated at the support points on the foundation. The types
of piping that generate most of the loads are: main steam inlet piping; reheat steam
piping, and extraction steam piping.
8 Piping Load from Equipment Attached to the Foundation (PEL)
As we had stated earlier that various auxiliary equipment are also supported on the
turbine deck. Positioning and aligning piping for this equipment creates erection forces.
Turbine piping is assembled and welded to these equipment and is anchored to the
foundation. The remainder of the steam inlet pipes is then welded to the assembly inlet
connections. Different forces are created due to thermal expansion during operation.
Erection forces, static and dynamic forces should be evaluated to check if they have
any significant contribution or not. For instance a rapid closing of the steam stop valve
attached to the foundation can induce a major loading.
9 Load due to Machine Unbalance (MUL)
Irrespective of however care is taken in balancing the turbine generator rotor it practi-
cally impossible to do away with some imbalance in force which it will generate during
its rotation.
The magnitude of this imbalance depends on a number of factors like design
considerations, installation and maintenance procedures. The factors which usually
contribute to such imbalanced dynamic load can be summarised as follows:
i Axis of rotation eccentric to the center of mass of the rotor;

ii Deflection of the shaft due to gravity load;

iii Uneven thermal expansion;

iv Misalignment during installation;
v Normal wear and tear during operation and,
vi Corrosion.
The combined or few of the reasons as mentioned above contribute to the dynamic
imbalance in the in the rotating shaft which is synchronous with the shaft rotational
speed. These forces are transmitted through the bearing shaft to the foundation.
The dynamic load is defined by
Pdyn = m · e · ω2 (2.18.2)
Here, m = unbalanced mass of the rotor; e = eccentricity of the rotor shaft, and
ω = operating frequency of the machine.
10 Load due to Bowed Rotor (BRL)
A bowed rotor can impose large dynamic loads on turbine generators foundation.
The bowed condition of the rotor will create unbalance force which are transmitted
through the machine bearings to the sole plates. The magnitude of the force will vary
with the unbalanced dynamic force as mentioned above.
The phenomenon can happen due to:
i Failure to put the rotor on turning gear when the machine is shut down;
ii Deflection of the raft, soil and the frame in flexural mode;
iii Water Induction and
iv Very severe packing rub.
The largest bowed rotor response occurs at the first critical speed for the rotor. The
time taken by the turbine rotor to pass through the critical speed is shorter when going
on-line.
However it takes much longer time when it goes off-line and the rotor coasts through
the resonant speed. Since this is a condition that usually requires turbine generator shut
down it will exist only for the time required for the rotor to coast down to rest. Thus
it is sufficient to ensure that the foundation stresses are low enough to eliminate the
chance of any permanent damage to the structure during the shut down period.
The magnitude and the location of the bowed rotor is usually supplied by the man-
ufacturer of the turbine in question and is dependent on the specific assumption made
by the vendor.
The force due to bowed rotor is function of the unbalanced dynamic force
Pdyn = m · e · ω2 (2.18.3)
The loading is normally provided in the form of a sinusoidal function for the dynamic
analysis or an equivalent static load for simplified analysis.
It is to be noted that, some turbine manufacturer may not supply this load for
depending on their own design some consider bowed rotor as worst case of accidental

loading for the high pressure and intermediate pressure turbine while others consider
loss of turbine blade as worst case of accidental load that can come on the foundation
as an emergency load.
11 Load due to Missing Rotor Blade (MRBL)

A turbine rotor must be balanced dynamically to ensure satisfactory operation and
cause no adverse effects on the turbine equipment or the foundation.
However it has been observed from previous operational experience that in some
cases (though very rare) that the last row of blade in a low pressure rotor breaks loose
from the rotor and causes a severe imbalance in the system.
The lose of blade which can vary in size from 500 mm to 1000 mm in length
can cause substantial force on the rotor, the bearing and the foundation system. The
magnitude of this unbalance is a function of the rotor blade weight, its center of gravity
with respect to the rotor and rotational speed of the rotor. As this can happen with
any of the several rows of last row blades a separate analysis should be made with a
single unbalance equivalent to the loss of one last row blade applied to the mass point
corresponding to each of the last row blade in each low pressure turbine.
Since this is an emergency situation and will require the turbine to be shut down it
will only exist only for the time period required for turbine to come down to a stop.
Thus it is sufficient to ensure that the stresses in the foundation are low enough to
preclude any permanent damage during the coast down period.
The loading is normally provided by the vendor in the form of a sinusoidal function
for the dynamic analysis or an equivalent static load for simplified analysis.
12 Electrical Loads
1 Generator Emergency Torque (GET)
Of all the loads that can occur a line-to-line short circuit at the generator terminal
causes the most severe loading of the turbine generator loading. Such a fault occurs
when any two of the three generators phase are shorted. The calculation of the maxi-
mum generator air gap torque during symmetrical three phase and unsymmetrical line
to line or line to ground terminal short circuits is normally performed assuming no
electrical damping in order to obtain greatest possible forces that can be transmitted
to the foundation under different fault condition.
Experience and previous data shows that the maximum torque resulting from a line
to line short circuit is about 25% greater than that caused by a single terminal to
ground fault and roughly 30% more than that with a symmetrical 3 phase fault at the
terminal of the generator.
The vendor in the form of a forcing function or an equivalent static force normally
provides the loading due to generator short circuit.
The use of equivalent static force for the maximum short circuit torque assume that
the foundation is infinitely rigid and thus must directly absorb the full impact of the
severe shock forces.
Since this assumption may result in over designing the foundation the more realistic
approach of a dynamic analysis is on the basis of the short circuit moment as a time
dependent function is usually preferred.

In view of the very severe transient nature of the maximum short circuit loading the
foundation in the vicinity of the generator, the designer should perform an appropriate
dynamic analysis of this abnormal load case.
2 Load combination for design

The following load combination is generally considered for the design as per American
practice72 .
i Operating conditions
The loading condition for which the foundation has to checked for and designed is
= 1.4(DL+MDL+OEL+PEL +CDL+CVL)+1.7(LL+NTL+ThL+TCPL+MUL)
ii Accident Conditions
Generator Emergency −
= DL+MDL+OEL+PEL+CDL+CVL+LL+GET+ThL+TCPL+MUL
Bowed Rotor case −
= DL+MDL+OEL+PEL+CDL+CVL+LL+NTL+BRL+ThL+TCPL
Missing Rotor Blade −
= DL+MDL+OEL+PEL+CDL+CVL+MRBL+NTL+ThL+TCPL
Seismic load −
= 0.75 [1.4(DL+MDL+OEL+PEL+CDL+CVL)+1.7(LL+NTL+ThL+TCPL
+MUL+1.1 SL)
It is to be noted that 1.4 and 1.7 are load factors for design of concrete section based
on ACI-318. For design of sections based on other codes like IS or BS appropriate load
factors in place 1.4 and 1.7 has to be taken.
2.18.2 Spring mounted turbine foundation

In this section we discuss the method of analysis and design of turbine foundations
mounted on springs.
This is a practice which is quite common in European countries and is being put to
increasing use in this part of the world now a days specially for foundations supporting
Turbines of high capacity.
In this case the top deck is usually mounted on springs of pre-designed specification
and is supported in turn on a frame as shown in Figure 2.18.1.
From the conceptual Figure 2.18.1, the obvious question that comes to mind is why
do we do such a thing and what advantage we gain from it?
To understand this we take up hereafter a concept which is otherwise known as
vibration isolation.
72 IS 2974 Part III though discusses the vibration analysis in detail it is silent on how and what load
combinations should be considered for design.

Top Deck
C.G of top deck
Mechanical Springs
Frame supporting
Top Deck
Bottom Raft
Figure 2.18.1 Spring mounted turbine foundation.
2.18.2.1 Theory of vibration isolation

We had seen in the section for analysis of block foundation that under dynamic load
the amplitude of vibration is expressed by the formula
P0
Kz sin ωm t
δz = (2.18.4)
(1 − r2 )2 + (2Dz r)2
where r = ωωmn ; Dz = damping ratio, and Kz = equivalent spring of the soil.

Now instead of soil spring if we support the block on some mechanical springs (Ks )
only the amplitude of vibration of the spring can be expressed as
P0
Ks sin ωm t ωm
δs = where r = . (2.18.5)
1 − r2 ωn
P0 sin ωm t P0 sinωm t
Now, Ks δs = = Ps = where Ps = Ks δs (2.18.6)
1 − r2 1 − r2

Considering, TF = Ps /P0 = transmissibility factor, we have
1
TF = (2.18.7)
1 − r2
The transmissibility factor is thus a measure of how much of the dynamic force is
transmitted to the supporting springs.
For transmissibility in the range less than unity the above equation is written in
the form
1
TF = (2.18.8)
r2 − 1
√
Considering the limiting case of TF = 1 we have, r2 − 1 = 1 → i.e. r = 2.
Thus it is seen that√the transmissibility factor T F shall have a value less than unity
for all values of r ≥ 2.
For damping prevalent in the system the transmissibility factor is given by
expression

1 + (2Dz r)2
TF = (2.18.9)
(1 − r2 )2 + (2Dz r)2
ωm
where r = and Dz = damping ratio.
ωn
If we plot the above equations for different values of frequency ratio and TF we
have curves as shown in Figure 2.18.2.
Observing the curves, it will be seen that even with√
damping existing in the system
TF value is less than 1 when the frequency ratio r ≥ 2 i.e. the force transmitted to
the support is less than the induced dynamic force.
To get a further insight into how the frequency ratio affects transmissibility factor
we study an expression called isolation efficiency expressed as
I = rr2 −2 × 100 in % where r = ωωmn and is the measure of the reduction of Trans-
2
−1
missibility factor of the system (Crede 1951).
We plot a curve, shown in Figure 2.18.3, based on the above
√ expression.
Based on this figure we find that when frequency ratio is 2 the isolation efficiency
is 0% i.e. 100% of the dynamic load gets transmitted to the support.
However when r = 2.45 the reduction efficiency increases to 80% i.e. a significant
amount of reduction of force transmittal to the support system is obtained.

12
Transmissibility Factor
10
Damping ratio @ 5%
8
Damping ratio @ 10%
6 Damping ratio @ 15%
Damping ratio @ 20%
4 Damping ratio @ 25%
Damping ratio @ 30%
2
0
25
75
25
75
25
75
25
5
5
0
3
0.
1.
2.
0.
0.
1.
1.
2.
2.
3.
Frequency Ratio
Figure 2.18.2 Variation in transmissibility factor.
Isolation Efficiency(%)
120
Isolation efficiency(%)
100
80
60 Isolation
Efficiency(%)
40
20
0
2.1
2.8
3.5
4.2
4.9
1.41
1.75
2.45
3.15
3.85
4.55
Frequency Ratio
Figure 2.18.3 Isolation efficiency (%).
We give below some data showing variation of Isolation efficiency with respect to
the frequency ratio
Frequency ratio 1.414 1.5 2.0 2.5 3.0 3.5 4.0 5.0
Isolation efficiency(%) 0.0 20 66.66 80.95 87.5 91.11 93.33 95.8
It will be observed both from the above figures as well as from Figure 2.18.3 that
up to a frequency ratio of 3.0, the reduction in transmitted force to the support is
significant but beyond that as the curve flattens asymptotically not much reduction in
transmissibility is obtained.
For instance if we increase the frequency ratio from 3 to 5 say the variation in
isolation frequency is only 8.6% however the manufacturing cost for such mechanical
springs as per some vendors nearly gets doubled.

Thus the common practise is to restrict the frequency ratio to maximum between 3
and 4 in practical engineering design.
Hence it is seen that if we can provide elastic supports like springs below a foundation
and can maintain a separation ratio of 3 to 4 with respect to the operating frequency
of the machine following advantages may be obtained
• The dynamic force transmitted to the supporting system for the springs could be
significantly reduced.
• Based on the reduced dynamic force it is possible to restrict the amplitude of
vibration to manageable limits.
• The foundation remains isolated/de-coupled to the surrounding and does not
transfer any dynamic load.
The above points are in a nutshell major advantage gained by providing springs for
vibration isolation.
Moreover as the springs are man made (unlike soil where we do not have any
control on its property) under a careful controlled condition, it is possible to design
these springs in such a way that they do have a frequency ratio between 3 to 4 with
the operating frequency of the machine.
2.18.2.2 Effect of damping on the transmissibility factor

Since any physical system in this world has some amount of damping (even air) it
would be worthwhile to evaluate how damping affects the transmissibility coefficient.
Freq. ratio Damping ratio 0.0 0.05 0.10 0.15 0.20 0.25 0.3
2.0 TF 0.333 0.339 0.356 0.381 0.412 0.447 0.483

3.0 TF 0.125 0.130 0.145 0.167 0.193 0.221 0.251
4.0 TF 0.066 0.072 0.085 0.103 0.125 0.147 0.171
Studying the above table it will be observed that having high damping value in the
system is counter productive to transmissibility. On the contrary a little amount of
damping in the system is advantageous in terms of transmittal of dynamic forces to
the foundation.
On the other hand, as we know that amplitude gets reduced due to the effect of
damping in the resonant zone the most ideal damper that can be introduced in a
system should thus have the following properties:
• High damping value when the frequency of the machine is passing through the
resonant range.
• Nominal damping value when the machine is operating at its normal speed.
Vendors specialising in supplying such viscous-dampers have their patented products

which exhibits such property as discussed above thus suppressing the dynamic effect
of the machine on the foundation and to its surrounding considerably.

2.18.2.3 How springs affect turbine foundation?

Based on the above theoretical discussion the query is but obvious.
For Turbine foundation as shown in the Figure 2.18.1, the top deck is usually
mounted on the spring and whole spring mounted assembly is then supported on a
frame which could be made of either RCC or steel.
The springs are supported in such a manner that the support points match with the
c.g. of the top deck and machine. This helps in suppressing the coupled translation and
rocking mode of the top deck. The technique of providing spring mounted Turbine
top deck is mostly in vogue in Europe where the Germans pioneered this technique
about 40 years ago.
Surprisingly in spite of certain advantage it provides (specially for turbines oper-
ating in Nuclear power plants) in terms of cost, plant layout etc it has not
been a popular concept in USA where engineers still opt for conventional framed
foundations.
Conventional turbine frame foundations, usually calls for columns of heavy section
and also a huge base mat to suppress the dynamic effect.
When the top deck is mounted on springs the major advantage is that the dynamic
effect of the machine is restricted up to the spring part only and the rest of the
foundation needs to be only designed for static loading.
The obvious advantage is that it calls for much slicker frame resulting in considerable
saving in material cost and as far as analysis is concerned, uncertainties prevalent with
a comprehensive dynamic soil structure interaction analysis (specially if resting on
piles) for such complex system is not required.
For nuclear power plants, the operating frequency of turbine is usually around
1500 rpm unlike conventional power plants (where it is about 3000 to 3600 rpm)
thus while designing the pedestals for these foundations engineers faced difficulties
to keep them significantly away from the operating frequency of the machine as they
were becoming far too flexible to their discomfort. The obvious choice was then to
mount them on springs and isolate the rest of the foundation from the dynamic effect.
Though the above was a starting point of such concepts, spring mounted
turbine foundations are now quite common in conventional fossil fuel power
plant also.
The major advantages gained in this case can be summarised as given hereunder:
• The top deck remains dynamically uncoupled with respect to the supporting frame,
thus the supporting frame is only subjected to static load and needs to be designed
accordingly.
• This makes the supporting framed structure slicker and also does away with
the necessity of providing a heavy bottom mat which is otherwise essential for
a conventional frame foundation.
• The springs are capable to certain extent adjust themselves to cater to the
differential settlement, if any.
• Even due to the overall settlement of the foundation which can cause additional
stress to the critical pipe connection, adjustment can be directly made using the
springs to adjust the levels and that too without interrupting the operation of the
machine.

• More space is usually available below the foundation thus maintenance and laying
of piping and cables become more accessible and easy.
• Substantial gain in material and cost is evident. Some vendors claim that with
spring mounted turbine foundations saving in cost could be as high as 45% when
compared to conventional frame foundation.
• Cost of piling is reduced as there is a significant reduction in weight.
• No dynamic loads need to be considered for the piles.
• The structural uncoupling of the top deck allows for the use of even steel structures
for the supporting frames.
• Use of steel structures gives additional advantage in terms of construction sequence
for they can be installed parallel to the power house structure which gives a
significant saving in construction time.
• Differential settlements can be easily measured based on the variation of spring
heights. Instrumentation techniques are available which monitors these spring
heights and when it exceeds preset-values automatically give visual signals or sends
alarms.
The advantages as mentioned above are making this concept progressively pop-
ular in the Industry. In many projects in India also this concept has been put to
practice and the turbines are found to be operating quite smoothly without any
hindrance.
2.18.2.4 Mathematical modelling of spring mounted turbines

The intuitive computer model that could be conceived for this case is to conceive
spring elements connected to the top deck directly supported in the bottom frame
[Figure 2.18.4].
However other than ANSYS most of the normally available structural engineering
package do not have the provision of adding springs directly between two members
(the basic pre-condition is one end of the spring should be fixed and not an active
node).
Top deck
Spring Elements
Supporting frame
Figure 2.18.4 Actual model of the top deck mounted on spring.

In such cases the most effective way to model the spring would be to represent
it by equivalent truss element having stiffness as AE/L, where AE/L shall have a
magnitude equal to the individual spring stiffness as considered in the vendor’s
catalogue.
2.18.2.5 Turbine foundations concrete versus steel structure

In USA as well as India conventional Turbine Foundation design is still dominated by
RCC structure.
However in many countries in Europe (especially Germany, France, Hungary etc.)
and Canada, Turbine foundation made of steel has been successfully implemented.
One of the major advantages with the RCC is its high damping property and not
requiring a very sophisticated construction technology to construct it. At the early
stage of advent of turbine foundations thus RCC made a very attractive choice.
However with turbine capacities increasing progressively the size of the turbine
foundations are also getting bigger and the construction technology is getting more
and more complex.
One of the major requirements of casting of RCC Turbine Foundation is that it
should be preferably cast in one go. Else additional steel has to be provided at such cold
joints and additional care has to be taken during construction to ensure its monolithic
property.
While for a foundation having concrete volume of 500 m3 this was not a very
difficult task, but when foundation capacity gets increased to 1500 m3 or more it
surely becomes a different ball game.
Firstly one needs a complete batching plant to be erected at the turbine foundation
site capable of supplying continuously concrete of same quality. This requires a very
elaborate arrangement to be made by the contractor at site including a building of a
make shift testing laboratory, where samples are collected and tested from different
batches continuously to ensure that the concrete is of the desired specified quality.
The next major difficulty encountered while casting is the heat of hydration which
RCC generates during hardening.
When the volume of concrete is large the heat of hydration can be substantial to cre-
ate cracks at the surfaces and needs to be carefully controlled at site (usually controlled
by using pre fixed quantity of Ice blocks in lieu of water) to nullify its effects.
The time consumed for laying reinforcement is substantial and needs to be
thoroughly checked with respect to drawing.
For large turbine foundations (>500 MW) at times client also insists in his contract
that the contractor to ensure based on non-destructive test that there are no voids or
honeycombing within the concrete. This calls for expensive ultra-sonic taste of the
foundation which is not only an expensive exercise but time consuming too.
As far as design aspect is concerned one of the major difficulties encountered is the
laying of the embedded plates and hangers in top deck for pipe supports which are
usually large in numbers.
Firstly when the turbine foundation is getting designed the critical steam piping
design is yet not finalised and thus the location of embedded plates and supports
furnished by the piping engineer is only tentative and could be subjected to change.
This surely makes the foundation designers task a difficult one.

For once the turbine deck is cast no perforation or anchoring to the deck is allowed
which could be detrimental to the concrete strength73 . Thus the engineer has to be
doubly careful with this drawing and check it very thoroughly to ensure that not a
single plate is missed, thus time taken for engineering is more.
A general tendency is thus to provide more number of embedded plates or plates
of bigger size to cover the eventualities. Since this is completely dependent on the
engineers personal judgement it has not been uncommon that at times the tonnage
has become as high as 30% extra then the estimated value and incurred unnecessary
wastage.
Similarly for any valves or other sundry equipment resting on the turbine top deck
their location and anchor bolt details etc needs to be finalised during drawing prepara-
tion stage. If the procurement department has not finalised with these equipment or the
vendor drawings are not available – the design engineer could be in a lot of difficulty.
With steel structure on the contrary most of the difficulties as discussed above is not
encountered. In spite of the fact that steel structure provides low damping, for large
turbine foundations steel as a construction material do have some distinct advantage
over RCC.
Firstly every thing need not be erected at the site; the top deck, which generally
consists of a rigid grillage, can be constructed at the shop under a careful controlled
condition and be carried to the site and erected over the columns.
As welds are susceptible to rapid fatigue failure under dynamic loads due to
reversible of stress the connections are usually bolted (bolted connections also provide
good damping and is more advantageous in such cases) and providing site connected
bolting is not a problem at all.
The major advantage in terms of construction is that the elaborate arrangement
one requires for RCC structures in terms of inspection and checking of laying of re-
bars, controlling the concrete quality and large amount of human resources one has
to deploy at the site is not required at all.
In fact the fabrication of the top deck at shop can start much ahead of the erection
of powerhouse it self and can be erected at site simultaneously.
This significantly saves construction cost as well as time too.
From design engineering point of view one need not worry about the location of
embedded plates and hangars, even with very late information welding locally steel to
steel is never a problem unlike anchoring plates on concrete top deck.
It can be logically perceived that steel foundation would be relatively high tuned one
compared to RCC foundations due to its lower mass. However they can be suitably
designed and adjusted to have the requisite frequency separation of 20%.
As we had stated earlier that turbine foundation usually does not become critical
during its normal operation but shows significant excitation during the start and stop
of the machine (mostly due to the soil participation) if the amplitude of vibration can
limited within the acceptable limit steel structures do have a very high potential as a
construction material for such type of structures.
73 Though technology exists where embedded plates can be anchored to concrete slabs after it is cast but
considering the critical nature of the turbine foundation such processes are usually not allowed for
Turbine top deck by the client and is not a good engineering practice too.

One of the major constraints encountered for turbines mounted on steel structures
is however the limitation in the available of ready made rolled sections.
For large turbines composite columns made out of industrially available rolled
section could become inadequate in terms of strength. This calls for usage of plate
girders in lieu of composite section. Due to inherent weakness of welds under dynamic
loading continuous butt welds are usually preferred instead of fillet welds. Continuous
butt welds specially at the flange and web junction calls for rigorous quality assurance
in terms of Radiographic test or dye test to ensure 100 % weld penetration and could
make the fabrication expensive.
However if the steel industry in India agrees to manufacture rolled sections of higher
sizes (beyond ISMB 600) steel structure can become a very strong competitor to RCC
foundations.
In Europe since rolled steel sections having much higher moment of inertia are
available, use of steel structure as an alternate to RCC has become a viable solution
there.
2.18.2.6 Design of RCC sections

The structural members are usually designed by using IS-456(2000) or the local code
of the country in which it is being constructed or as specified in the contract like
ACI318, BS8110, DIN etc.
In most of the case the geometric sizing is decided by the equipment supplier, based
on which the stress induced in the members itself are normally low and to our knowl-
edge there has been no such cases where Turbine foundation members have misbehaved
or failed due to strength failure.
Most of the cases where members have misbehaved can be attributed to
improper detailing or faulty construction for which cracks have been observed to
develop.
Based on above, proper detailing of the members are of primary importance.
Some good detailing practices are mentioned hereafter which could be followed
while detailing such foundations.
• The vertical reinforcing bars of the column shall have sufficient embedment in the
base slab to develop the required stresses.
• Reinforcement in beams and columns shall be provided in all four sides irrespective
of they are required or not.
• If design requirements do not guide the percentage of steel, the re-bars shall be
placed symmetrically on all four sides.
• The minimum Steel provided in different parts of the members are mentioned
hereunder
Sl. No. Structural member type Steel quantity
1 Base slab 40 kg/m3

2 Columns 70 kg/m3
3 Top deck (beam and slab) 90 kg/m3

• Shear stirrups to be provided to account for the total shear in the foundation
element.
• Splicing in columns if any shall always be done at the mid-height.
• The diameter of bar in beams and columns should be so chosen that the maximum
spacing of the bars are not more than 150 mm.
• Try to use lower diameter bar as far as practicable. For with lower diameter bars,
number of bars is more and distribution of stress and transfer of load between
concrete and steel is more uniform.
• Unless specified by the contract the cover to reinforcement is usually taken as
follows:
Base Slab 100 mm on top, bottom and sides.
Columns and Pedestals 50 mm on sides
Beams 40 mm on all sides
• Minimum development length for all bars irrespective of requirement shall not be
less than 50 times the diameter of the bar.
• Beam column junction should be provided with additional steel to ensure that
cracks do not develop due to continuous reversal of stresses due to the application
of cyclic loads.
Example 2.18.1
Shown in Figure 2.18.5 is the layout plan of a Boiler feed pump framed
foundation with location of equipment loads as shown.
The dynamic loads under various operating conditions are as shown hereafter.
4590 Pump Side 4590 C/L Coupling Motor Side
1315
Y 16.65 +3.5m(TOC)
350
3.0m(TOC) 195kN
22.1kN 35 5 508 54
775
+3.5m(T.O.C.) 16.65 1200
1580
X + 4.0m(T.O.C.) 100kN
1765 4192 1585 2238
= =
Figure 2.18.5 Plan view of top deck with location of equipment load.

Table for dynamic loads on boiler feed pump top deck
Load condition Remarks Px Py Pz Mx My Mz
Short circuit moment Total force at top deck 0.0 226 268 0 0 0
Operating load (1) End frame (pump side) 58 44 0 0 0
Middle frame 80 0 0 0
End frame (motor side) 100 8.3 0 0 0
Operating load (2) Load per long beam (pump side) 37 0.0 130 0 0 0
Operating load (2) Load per long beam (at coupling) 0 0 25 0 0 0
• Operating Frequency of the machine = 5100 rpm

• Center line axis of shaft = 1.1 m above the top deck
• Bearing capacity of soil 150 kN/m2
• Shear wave velocity of the soil = 115 m/sec
• Poisson’s Ratio of soil = 0.30
• Live load of top deck slab during operation = 5 kN/m2
• Unit Weight of soil = 19 kN/m3
• All columns = 600 mm × 600 mm
• All beams = 600 mm × 900 mm
• Grade of concrete M25
Analyze the frame by
• Rausch’s method
• By Barkan’s method
• By Major’s method
• By 2D soil-structure interaction model.
Compare the results of the analysis based on the above methods with time
history
Do detailed design of the frame.
Solution:
We start the problem sequentially.
Here the top deck consists of a flat slab 900 mm thick supported on columns
(600 mm × 600 mm) and practically does not have a framing system.
Here for analysis and design we perceive a frame having edge beams in both
transverse and longitudinal direction having depth of 600 mm × 900 mm as
shown in Figure 2.18.5 and 6.
The load from the slab is transferred to the idealized frame (as shown by the
dotted lines, Figure 2.18.7) and the frame is analyzed for vibration in vertical
and horizontal mode.

3200
900
3600
(+/-)0.0
1500
Figure 2.18.6 Elevation of the frame in transverse direction.
4590 4590
A
1 2
2600 22.5 195 16.7 16.7

508
54
690 2571 1329 1417 965 1308 900
1 2 3
Figure 2.18.7
Calculation of UDL load transferred to frame
Thickness of slab = 900 mm

Self weight of slab = 0.9 × 25 = 22.5 kN/m2

Live load = 5.0 kN/m2 : Thus, DL + LL = 27.5 kN/m2

wlx 27.5 × 2.6
Equivalent UDL transferred to frame 1 and 3 = = =
3 3
23.83 kN/m
Equivalent UDL on frame 2 = 23.83 × 2 = 47.66 kN/m
From Figure 2.18.7,
2
wlx lx 27.5 × 2.6
Equivalent UDL on frame A and B = 3− = ×
6 ly 6

2.6 2
3− = 31.92 kN/m
4.59
Load on longitudinal beam from the area of hydraulic coupling having local
projection of 1.4 m, of width 1585 mm
0.5 × 2.6 × 25
w=
2
= 16.21 kN/m
Thus the total superimposed UDL coming on the frame is as shown in

Figure 2.18.8
48 KN/m
24 KN/m
32 KN/m 48 KN/m 48 KN/m
24 KN/m
32 KN/m
Figure 2.18.8 Frame with uniformly distributed load from top deck slab.

Calculation of Concentrated load
For slab panel 1 along center line axis the distribution of load is as shown in
Figure 2.18.9.
R=217.5
Ly R=217.5
22.5 195
Lx
690 2571 1329
2995
Figure 2.18.9 Load distribution slab panel-1.
Here we first out the point through which the resultant of this two concen-
trated force acts
22.5 × 690 + 195 × 3261
yc = = 2995 mm from frame 1
22.5 + 195
Now the slab being restrained at all sides subjected to a load of 217.5 at
distance of 2995 mm from frame 1 it is evident that displacement at point O
shall be same for long and short span.
Thus considering the middle strip as a beam fixed at both ends in long direction
Py a 3 b 3
δl =
3EIL3y
Here Py = the net concentrated load acting in long direction; a = 2995 mm;
b = 1595 mm; Ly = 4590 mm, and Lx = 2600 mm
Px L3x
Displacement in short span is given by, δs =
192EI
Here Px = The load transferred to short span.
Since here due to displacement compatibility, δl = δs , we have
Px L3x P y a3 b3 a3 b3
= ➔ Px = 64Py 3
192EI 3EIL3y Ly
Lx

Since by law of static V = 0 we have, Px + Py = 217.5
5 3
ab
➔ Py = 218 64 +1
(Ly /Lx )

Now substituting the values a, b, Ly and Lx , we have → Py = 19.27 kN and

Px = 198.73 kN.
Since in short direction the load is symmetrical load on long beams along
198.73
Row A and B = ≈ 99.4 kN
2
In long direction, the position of the load is as shown in Figure 2.18.10.
19.3
R1 2995 1595 R2
Figure 2.18.10
19.3 × 1595
Thus R1 = = 6.7 kN and R2 = 12.6 kN.
4590
Concentrated load transferred to edge beams from the slab panel 1 is shown
in Figure 2.18.11.
99.4
6.7 12.6
99.4
Figure 2.18.11 Concentrated load transferred to edge beams from the slab panel-1.
For slab panel 2 the loading arrangement is as shown Figure 2.18.12.
Ly = 4500
16.7 16.7 Lx = 2600

508
54
1417 965 1308 900
Figure 2.18.12 Loading on slab panel-2.

Ly = 4500
Lx = 2600
87.4 194
3006 1584
Figure 2.18.13 Resultant load on slab panel-2.
Total concentrated load acting on the slab = 16.7 × 2 + 54 = 87.4 kN

The c.g. of the loads in y direction (Figure 2.18.13) is given by
16.7 × 1308 + 16.7 × 2273

yc = = 684 mm and
87.4
33.4 × 508
xc = = 194 mm
87.4
Here
Py a3y b3y Px a3x b3x
δl = and δs = , when based on displacement compatibility
3EIL3y 3EIL3x
3
Py a3y b3y Px a3x b3x a x bx 3 Ly
δl = δs when we have = ➔ P y = Px ×
3EIL3y 3EIL3x ay by Lx
Here ax = 1494 mm; bx = 1106 mm; ay = 3006 mm; by = 1584 mm;
Ly = 4590 mm and Lx = 2600 mm.
Substituting the values in the above equation we have, Py = 0.2299Px

For V = 0 we have, Px + Py = 87.4 kN
Thus, we have Px = 71.06 kN and Py = 16.34 kN. Now, proceeding in the
similar manner as explained for slab panel 1 we find out the load transferred on
the frame beams shown in Figure 2.18.14.

Concentrated load transferred to edge beams from the slab panel 2 in

Figure 2.18.14.
30.22
5.64 10.7
140.83
1494
3006
Figure 2.18.14 Concentrated load transferred to edge beams from slab panel-2.
The net distribution of concentrated force on the frame is as shown in

Figure 2.18.15.
30.2
10.7
99.4
18.2
141
6.7
1584
99.4
Figure 2.18.15 Frame with concentrated load from top deck slab.

1 Rausch’s method
Calculation of geometric properties of the frame
Area of transverse beam Ab = (600 × 900) = 0.6 × 0.9 = 0.54 m2
Area of column Ac = (600 × 600) = 0.6 × 0.6 = 0.36 m2
1
Ib = × 0.6 × 0.93 = 0.03645 m4 ;
12
1
Ic = × 0.6 × 0.63 = 0.0108 m4 ; and
12
Ib h 0.03645 × 4.05
ψ= = = 5.257
Ic L 0.0108 × 2.6
Calculation of load transverse frame 1

Self weight of beam = 0.6×0.9×25 = 13.5 kN/m; UDL from slab = 24 kN/m
Total, UDL(q) = 24 + 13.5 = 37.5 kN/m
Self weight of long beam = 13.5 kN/m; UDL on long beam = 32 kN/m
Total UDL on long beam = 32 + 13.5 = 45.5 kN/m
4.59
Load from long beam = 45.5 × + 0.5 × 99.4 = 154.15 kN
2
Load from column = 0.6 × 0.6 × 1.8 × 25 = 16.2 kN
Total load transferred to column (N) = 16.2 + 154.15 = 170.4 kN
Load on beam from machine (P) = 6.7 kN; Modulus of concrete Ec = 3×
108 kN/m2
➔ Ec Ib = 3 × 108 × 0.03645 = 1.0935 × 107 kN m2 ;

Ec Ab = 3 × 108 × 0.54 = 1.62 × 108 kN
Ec Ac = 3 × 108 × 0.36 = 1.08 × 108 kN
Calculation of displacements transverse frame 1

PL3 2ψ + 1 6.7 × (2.6)3 2 × 5.257 + 1
δ1 = =
96EIb ψ +2 96 × 1.0935 × 107 5.257 + 2
= 1.769 × 10−7 m

QL3 5ψ + 2 37.5 × 2.6 × (2.6)3 5 × 5.257 + 2
δ2 = =
384EIb ψ +2 384 × 1.0935 × 107 5.257 + 2
= 1.5893 × 10−6 m

3 L Q 0.6 × 2.6 98
δ3 = P+ = 6.7 + = 5.364 × 10−7 m
5 EAb 2 1.62 × 108 2


h P+Q 4.05 6.7 + 98
δ4 = N+ = 170.4 +
EAc 2 1.08 × 108 2

4
= 8.353 × 10−6 m ➔ δi = 1.06557 × 10−5 m
i=1
30 30
Thus, fv = √ = √ = 9190 rpm
δv 1.06557 × 10−5
Total UDL(q) = 48 + 13.5 = 61.5 kN/m
Self weight of long beam = 13.5 kN/m; UDL on long beam = 32 kN/m
Load from machine (P) = 18.2 kN
Total UDL on long beam = 32 + 13.5 = 45.5 kN/m
Load from long beam = 45.5 × 4.59 = 208.85 kN; Load from column =
0.6 × 0.6 × 1.8 × 25 = 16.2 kN
Shown in Figure 2.18.16.
99.4 30.2 for Row A

141 for Row B
16 kN/m
1584
1585
4590
2730
1 2 3
Figure 2.18.16 Load distribution on longitudinal girder.
Load on beam from machine on row A.

99.4 1584 2730
Load on row A = + 30.2 × + 16 × 1.585 × = 75.2 kN
2 4590 4590
99.4 1584 2730

Load on row B = + 141 × + 16 × 1.585 × = 113.4 kN
2 4590 4590
Load on column along row A = 208.85 + 75.2 + 16.2 = 300 kN
Load on column along row B = 208.85 + 113.4 + 16.2 = 338.45 kN
Average load = 319.225 kN
Thus loading on frame 2 is as shown in Figure 2.18.17.

319 (kN) 18.2 (kN) 319 (kN)
62 kN/m (q)
Figure 2.18.17 Load on transverse frame-2.
Calculation of displacements for transverse frame 2

PL3 2ψ + 1 18.2 × (2.6)3 2 × 5.257 + 1
δ1 = =
96EIb ψ +2 96 × 1.0935 × 107 5.257 + 2
= 4.832 × 10−7 m

QL3 5ψ + 2 62 × 2.6 × (2.6)3 5 × 5.257 + 2
δ2 = =
384EIb ψ +2 384 × 1.0935 × 107 5.257 + 2
= 2.614 × 10−6 m

3 L Q 0.6 × 2.6 161.2
δ3 = P+ = 18.2 +
5 EAb 2 1.62 × 108 2
= 9.514 × 10−7 m

h P+Q 4.05 18.2 + 161.2
δ4 = N+ = 319 +
EAc 2 1.08 × 108 2
= 1.53263 × 10−5 m

4
30 30
δi = 1.9375 × 10−5 m ➔ fv = √ = √ = 6815 rpm.
i=1
δ v 1.9375 × 10−5

Total UDL(q) = 24 + 13.5 = 37.5 kN/m
Self weight of long beam = 13.5 kN/m; UDL on long beam = 32 kN/m;

Load from machine (P) = 10.7 kN

Total UDL on long beam = 32 + 13.5 = 45.5 kN/m; Load from long beam =
4.59
45.5 × = 113 kN
2
Load from column = 0.6 × 0.6 × 1.8 × 25 = 16.2 kN
As shown in Figure 2.18.18, Load due to concentrated load on long beam on
Row A and B
3006 1860
Load on row A = 30.2 × + 16 × 1.585 × = 30.6 kN
4590 4590
3006 1860
Load on row B = 141 × + 16 × 1.585 × = 102.62 kN
4590 4590
Average load = 66.61 kN
Total load on column (N) = 66.61 + 113 ∼ = 180 kN
30.1 for Row A

141 for Row B
16 kN/m
1584
2730
2
3
Figure 2.18.18 Load on longitudinal girder.
Calculation of displacements for transverse frame 3

PL3 2ψ + 1 10.7 × (2.6)3 2 × 5.257 + 1
δ1 = =
96EI b ψ +2 96 × 1.0935 × 107 5.257 + 2
= 2.825 × 10−7 m

QL3 5ψ + 2 37.5 × 2.6 × (2.6)3 5 × 5.257 + 2
δ2 = =
384EI b ψ +2 384 × 1.0935 × 107 5.257 + 2
= 1.5893 × 10−6 m

3 L Q 0.6 × 2.6 98
δ3 = P+ = 10.7 + = 5.75 × 10−7 m
5 EAb 2 1.62 × 108 2

h P+Q 4.05 10.7 + 98
δ4 = N+ = 180 +
EAc 2 1.08 × 108 2
= 8.788 × 10−6 m


4
30 30
δi = 1.1234 × 10−5 m; Thus, fv = √ = √ = 8950 rpm.
i=1
δv 1.1234 × 10−5
9190 + 6815 + 8950

Thus average vertical frequency of the frame = =
3
8318 > 5100 rpm
Frequency in horizontal direction

Weight of top deck = 9.78 × 3.2 × 0.9 × 25 = 704 kN
Weight from machine = 22.1 + 195 + 16.65 × 2 + 54 + 100 = 404
Total weight = 704 + 404 = 1108 kN

12EIc 6ψ + 1 12 × 3 × 108 × 0.0108 6 × 5.257 + 1
Khi = = ×
h3 3ψ + 2 (4.05)3 3 × 5.257 + 2
= 1071751 kN/m

Kh1 + Kh2 + · · · · · · + KhN 3 × 1071751
fh = 30 = 30
W 1108
= 1616 < 5100 rpm.
The method does not have any provision of amplitude check and only check
for resonance with the operating frequency of the machine.
2 Barkan’s method
Calculation of stiffness for transverse frame 1 in vertical direction
2EAc 2 × 3 × 108 × 0.6 × 0.6
k1 = is the stiffness of the columns = =
h 4.05
5.333 × 107 kN/m
For transverse beam
L3 (1 + 2ψ) 3L
δv = +
96EI b (2 + ψ) 8GAb
(2.6)3 (1 + 2 × 5.257) 3 × 2.6
= × +
96 × 1.0935 × 107 7.257 8 × 1.5 × 108 × 0.6 × 0.9
= 3.86014 × 10−8 m
1
k2 = → k2 = 2.6 × 107 kN/m
δv
The stiffness matrix thus becomes

k + k2 −k2 7.93 −2.6
[K] = 1 = × 107 kN/m
−k2 k2 −2.6 2.6

m1 = mL + 0.255mb + 0.35mc
(24 + 13.5) × 2.6
mb = Mass of cross girder = = 9.938 kN · sec2 /m
9.81
mL = Mass transferred from long girder
104.45 × 2 + 99.4
= = 31.43 kN · sec2 /m
9.81
0.6 × 0.6 × 3.6 × 25 × 2
mc = Mass of column = = 6.605 KN/sec2 /m
9.81
m1 = 0.255 × 9.938 + 0.35 × 6.605 + 31.43 = 36.27 kN · sec2 /m
m2 = m0 + 0.45mb
m0 = Load from machine on transverse girder;
6.7
m2 = + 0.45 × 9.938 = 5.15kN · sec2 /m
9.81
36.27 0
Thus, [M] =
0 5.15
Thus based on eigen value solution

7.93 × 107 − 36.27λ −2.6 × 107
= =0
−2.6 × 107 2.6 × 107 − 5.15
The above on solution gives74
λ1 = 1.2369 × 106 ➔ ω1 = 1112 rad/sec(10168 rpm) > 5100 rpm
λ2 = 5.998 × 106 ➔ ω2 = 2449 rad/sec(23386 rpm) > 5100 rpm
The normalized eigen vector is given by

0.14837 −0.07413
[ϕ]n =
0.19654 0.39428
Based on loading table for dynamic load, Pv = 44 + 130 = 174 KN.

Thus equation of motion becomes

36.27 0 z̈1 7.93 −2.6 z 0.0
+ × 107 1 = sin 534t
0 5.15 z̈2 −2.6 2.6 z2 174
74 We will not solve this equation directly. Please refer to Chapter 5 (Vol. 1) on structural dynamics
where we have solved in detail such eigen value problem.

Calculation of amplitude for transverse frame 1 in vertical direction

0.14837 0.19654 0 34.19
Thus [ϕ]Tn {P} = = Considering 5%
−0.07413 0.39428 174 68.60
damping for concrete on orthogonal transformation of the above equation of
motion we have
ξ̈1 + 111ξ̇1 + 1.2369 × 106 ξ1 = 34.19 sin 534t;
ξ̈2 + 245ξ̇2 + 5.998 × 106 ξ2 = 68.60 sin 534t
ωm 534
Here, r = = 0.48 for mode 1 and r = = 0.21,
ωn 2449
34.19 sin 534t
➔ ξ1 = 1.2369×106
= 3.584×10−5 sin 534t m
(1−(0.48) )2 −(2×0.05×0.48)2
2
68.60 sin 534t

andξ2 = 5.998×106
= 11.92×10−6 sin 534t m
(1−(0.21) )2 −(2×0.05×0.21)2
2

z1 0.14837 −0.07413 35.48
We have then, {Z} = [ϕ]{ξ } = =
z2 0.19654 0.39428 11.92

4.38
× 10−6 sin 534t = sin 534t × 10−6 m
11.657
2EAc 2 × 3 × 108 × 0.6 × 0.6

k1 = is the stiffness of the columns = =
h 4.05
5.333 × 107 kN/m
For transverse beam
L3 (1 + 2ψ) 3L
δv = +
96EI b (2 + ψ) 8GAb
(2.6)3 (1 + 2 × 5.257) 3 × 2.6

= × +
96 × 1.0935 × 10 7 7.257 8 × 1.5 × 108 × 0.6 × 0.9
= 3.86014 × 10−8 m
1
➔ k2 = = 2.6 × 107 kN/m
δv

The stiffness matrix thus becomes

k + k2 −k2 7.93 −2.6
[K] = 1 = × 107 kN/m
−k2 k2 −2.6 2.6
m1 = mL + 0.255mb + 0.35mc
(48 + 13.5) × 2.6
9.81
(32 + 13.5) × 4.95 75.2 113.4
= + + = 42.20 kN sec2 /m.
9.81 9.81 9.81
0.6 × 0.6 × 3.6 × 25 × 2
mc = Mass of column = = 6.605 kNsec2 /m
9.81
m1 = 0.255 × 16.29 + 0.35 × 6.605 + 42.20 = 48.65 kN × sec2 /m
m2 = m0 + 0.45mb
m0 = Load from machine on transverse girder
18.2
m2 = + 0.45 × 16.29 = 9.185 kN · sec2 /m.
9.81

48.65 0
Thus [M] =
0 9.185

7.93 × 107 − 48.65λ −2.6 × 107
Thus the eigen value solution = = 0
−2.6 × 107 2.6 × 107 − 9.185λ
λ1 = 861700 ➔ ω1 = 928 rad/sec(8861 rpm) > 5100 rpm

6
λ2 = 3.599 × 10 ➔ ω2 = 1897 rad/sec(18115 rpm) > 5100 rpm

0.1215 −0.0754
The normalized eigen vector is given by, [ϕ]n =
0.1748 0.27985
Based on loading table for dynamic load
Pv = 25 + 130 = 155 kN.
75 We will not solve this equation directly. Please refer to Chapter 5 (Vol. 1) where we have solved in
detail such eigen value problem.


48.65 0 z̈1 7.93 −2.6 7 z1 0.0
+ × 10 = sin 534t
0 9.185 z̈2 −2.6 2.6 z2 155

0.1215 0.1748 0 27.094
[ϕ]T
n {P} = =
−0.0759 0.27985 155 43.3768
Considering 5% damping for concrete on orthogonal transformation of the

above equation of motion we have
ξ̈1 + 92.8ξ̇1 + 861700ξ1 = 27.094 sin 534t;

ξ̈2 + 189.7ξ̇2 + 3.599 × 106 ξ2 = 43.3768 sin 534t
ωm 5100
Here r = = 0.575 for mode 1 and r = = 0.28
ωn 18115
27.094 sin 534t

➔ ξ1 = 861700
= 4.68 × 10−5 sin 534t m
(1 − (0.575)2 )2 − (2 × 0.05 × 0.575)2
43.3768 sin 534t
3.599×106
and ξ2 = = 13.07 × 10−6 sin 534t m
(1 − (0.28) )2 − (2 × 0.05 × 0.28)2
2
Since {Z} = [ϕ] {ξ },

z1 0.1215 −0.0754 46.8
= sin 534t × 10−6
z2 0.1748 0.27985 13.07

4.70
= sin 534t × 10−6 m.
11.838
Referring to the previous calculation the stiffness matrix is

k + k2 −k2 7.93 −2.6
[K] = 1 = × 107 kN/m
−k2 k2 −2.6 2.6

m1 = mL + 0.255mb + 0.35mc
(24 + 13.5) × 2.6

9.81
(32 + 13.5) × 4.95 30.6 102.62
= + + = 25.06 kN · sec2 /m.
9.81 × 2 9.81 9.81
0.6 × 0.6 × 3.6 × 25 × 2

mc = Mass of column = = 6.605 kN · sec2 /m
9.81
m1 = 0.255 × 9.93 + 0.35 × 6.605 + 25.06 = 29.90 kN · sec2 /m

m2 = m0 + 0.45mb
m0 = Load from machine on transverse girder
10.7
m2 = + 0.45 × 9.93 = 5.56 kN · sec2 /m
9.81

30 0
Thus [M] =
0 5.56
Thus based on eigen value solution

7.93 × 107 − 30λ −2.6 × 107
= =0
−2.6 × 107 2.6 × 107 − 5.56λ
λ1 = 1404600 ➔ ω1 = 1185 rad/sec(11316 rpm) > 5100 rpm
λ2 = 5915000 ➔ ω2 = 2432 rad/sec(23224 rpm) > 5100 rpm
The normalized eigen vector is given by

0.1555 −0.0956
[ϕ]n =
0.2194 0.3612
Based on loading table for dynamic load
Pv = 8.3 + 25 = 33.3 kN.
76 We will not solve this equation directly. Please refer to Chapter 5 (Vol. 1) where we have solved in
detail such eigen value problem.


30 0 z̈1 7.93 −2.6 7
z1 0.0
+ × 10 = sin 534t
0 5.56 z̈2 −2.6 2.6 z2 33.3

0.1555 0.2194 0 7.30
Thus [ϕ]T
n {P} = =
−0.0956 0.3612 33.3 12.03
ξ̈1 + 118.5ξ̇1 + 1404600ξ1 = 7.30 sin 534t;

ξ̈2 + 243.2ξ̇2 + 5915000ξ2 = 12.03 sin 534t.
ωm
Here r = = 0.45 for mode 1 and r = 0.22
ωn
7.30 sin 534t

ξ1 = 1404600
= 6.5 × 10−6 sin 534t and
(1 − (0.45) ) − (2 × 0.05 × 0.45)2
2 2
12.03 sin 534t

ξ2 = 5.915×106
= 2.136 × 10−6 sin 534t
(1 − (0.22) )2 − (2 × 0.05 × 0.22)2
2
Since {Z} = [ϕ]{ξ }

z1 0.1555 −0.0956 6.5
Hence = sin 534t × 10−6
z2 0.2194 0.3612 2.136

0.822
= sin 534t × 10−6 m
2.219
Calculation of horizontal frequency
Table for calculation of mass and stiffness
Frame m0 mb mc ml mi δhi K hi
1 0.683 9.938 6.605 31.43 44.030 9.333E-07 1071467

2 1.855 16.29 6.605 42.2 62.326 9.333E-07 1071467
3 1.0907 9.93 6.605 25.06 38.060 9.333E-07 1071467
144.416 3214401

h3 (2 + 3ψ)
Here mi = m0i + mbi + 0.3mci + mLi and δhi =
12EIc (1 + 6ψ)
Table for calculation of second, mass moment and stiffness inertia
Frame mi Khi di midi Khi dI xgi (Xg − di ) Xhi mi xgi2 Khi Xhi2
1 44.030 1071467 0 0 0 4.40 4.59 852.5194 22573770

2 62.326 1071467 4.59 286.08 4918032.8 −0.190 0 2.243947 0
3 38.060 1071467 9.18 349.39 9836065.6 −4.78 −4.59 869.5175 22573770
635.47 14754098 1724.281 45147541
635.476 14754098
Xg = = 4.40 m; Xh = = 4.59 m
144.416 3214401
Thus e = 4.59 − 4.40 = 0.190 m

Dynamic loads
Phi = 58 + 80 + 100 = 238 kN and

N
Mh = Phi Xgi = 238 × 4.40 = 1047.2
i=1

M 0 ẍ Kh Kh e x Ph cos ωm t
+ =
0 Jϕ ϕ̈ Kh e Kh e2 + γ ϕ Mh cos ωm t

144.416 0.0 ẍ 3.215 × 106 0.6095 × 106 x
+
0.0 1724.3 ϕ̈ 0.6095 × 106 4.53 × 106 ϕ

238 cos 534t
=
1047.2 cos 534t
Thus for natural frequency we have

3.215 × 106 − 144.416λ 0.6095 × 106
=0
0.6095 × 106 4.53 × 106 − 1724.3λ
The above on expansion and solution gives
λ1 = 21919 ➔ ω1 = 148 rad/sec(1413 rpm) < 5100 rpm

λ2 = 26614 ➔ ω2 = 163 rad/sec(1557 rpm) < 5100 rpm

The normalized eigen vector is given by,

0.08011 0.022485
[ϕ]n =
−6.5109 × 10−3 0.023186

T 0.08011 −6.5109 × 10−3 238 cos 534t
[ϕ]n {P} =
0.022485 0.023186 1047.2 cos 534t

12.25 cos 534t
=
248.15 cos 534t

ξ̈1 + 14.8ξ̇1 + 21919ξ1 = 12.25 cos 534t;

ξ̈2 + 16.3ξ̇2 + 26614ξ2 = 248.15 cos 534t.
ωm
Here r = = 3.609 for mode 1 and r = 3.27
ωn
12.25 cos 534t

i.e. ξ1 = 21919
= 4.6707 × 10−5 cos 534t
(1− (3.6)2 )2 −(2 × 0.05 × 0.3.6)2
248.1552 cos 534t
and ξ2 = 26614
= 9.61416 × 10−4 cos 534t
(1−(3.27) ) −(2×0.05×3.27)2
2 2
Since {X} = [ϕ]{ξ }, hence

x 0.08011 −0.022485 4.6707
➔ = cos 534t × 10−5
ϕ −6.5109 × 10−3 0.023186 96.1416

2.5359
= cos 534t × 10−5 m
2.1987
Thus, displacement of frame 3

This is generically given by xnet = x + X ϕ where
X = is the farthest point form the center of gravity point G
x3 = 2.5359 × 10−5 + 4.78 × 2.1987 × 10−5 = 1.30456 × 10−4
Displacement of frame 2
x2 = 2.5359 × 10−5 + 0.190 × 2.1987 × 10−5 = 2.953 × 10−5
Displacement of frame 3
x1 = 2.5359 × 10−5 − 4.40 × 2.1987 × 10−5 = −7.13828 × 10−5

3 Major’s combined method of analysis

In this case the natural frequency in vertical direction is same as shown in
Rausch’s method earlier except for the case of soil when Major considers the
total frame as stick model and combines with soil displacement.
Thus for transverse frame 1, we have

4
δi = 1.06557 × 10−5 m
i=1
4
For calculation of i=1 δi refer to previous calculation by Rausch’s method
30 30
and fv = √ = √ = 9190 r.p.m
δv 1.06557 × 10−5
While Major calculated the dynamic force based on Rotor weights, here
dynamic force coming on the frame has directly been given, thus
Load on transverse beam (Cb ) = 44 kN
Load on column from Long beams(Cc ) = 130 kN

C L3 2ψ + 1 3 L h Cb + C c
Thus δ1 = b + C +
96EIb ψ +2 5 EAb b EAc 2
44 × (2.6)3 11.514 0.6 × 2.6 × 44

= × +
96 × 1.0935 × 10 7 7.257 1.62 × 108

4.05 44 + 130
+
1.08 × 108 2
= 4.855 × 10−6 m
As ωn > ωm hence corrected value of operating frequency is ωn =

0.8 × 9190 = 7352 r.p.m
1 5100
M.F = , here r = = 0.69 and = 0.4
(1 − r2 )2 + ( ∇π )2 (r)2 7352
Substituting the above values, we have

M.F. = 1.900
Thus δv1 = 4.855 × 10−6 × 1.9 = 9.2245 × 10−6 m

For transverse frame 2, we have

4
δi = 1.9375 × 10−5 m
i=1
4
30 30
and fv = √ = √ = 6815 rpm
δv 1.9375 × 10−5
Load on transverse beam (Cb ) = 0.0 kN
Load on column from Long beams(Cc ) = 130 + 25 = 155 kN

C L3 2ψ + 1 3 L h Cb + C c
Thus δ1 = b + Cb +
96EIb ψ +2 5 EAb EAc 2

4.05 155
= 0.0 + 0.0 + = 2.9063 × 10−6 m
1.08 × 108 2
As ωn > ωm hence corrected value of operating frequency is ωn = 0.8×6815 =

4948 rpm which is less than the operating speed of 5100 r.p.m.
At transient resonant condition as per Major M.F = 7.85, here r =
5100
= 0.69
7352
Substituting the above values we have
Thus, δv2 = 2.9063 × 10−6 × 7.85 = 2.281 × 10−5 m
For transverse frame 3 we have

4
δi = 1.1234 × 10−5 m
i=1
4
30 30
and fv = √ = √ = 8950 rpm
δv 1.1234 × 10−5
Load on transverse beam (Cb ) = 8.30 kN

Load on column from Long beams(Cc ) = 25 kN

Cb L3 2ψ + 1 3 L h Cb + C c
Thus δ1 = + C +
96EIb ψ +2 5 EAb b EAc 2
8.3 × (2.6)3 11.514 0.6 × 2.6 × 8.3

= × +
96 × 1.0935 × 10 7 7.257 1.62 × 108

4.05 8.3 + 25
+ = 9.2478 × 10−7 m
1.08 × 108 2
As ωn > ωm hence corrected value of operating frequency is ωn = 0.8×8950 =

7160 rpm which is greater than the operating speed of 5100 rpm.
Here
1 5100
M.F = , r= = 0.69 and = 0.4
[(1 − r2 )2 + ( ∇π )2 (r)2 ] 7352
which gives M.F. = 1.995

Substituting the above values we have
δv3 = 9.2478 × 10−7 × 1.995 = 1.8446 × 10−6 m
Calculation of vertical frequency including the soil effect
In this case we consider the total top deck including the column as a stick
model 4 −5
4
Thus, for frame 1, i=1 δi = 1.06557 × 10 ; For frame 2,
−5
i=1 δi = 1.9375

× 10 m;
For frame 3, 4i=1 δi = 1.1234 × 10−5 m; Thus, δav = 1.37549 × 10−5 m.
Shear wave velocity = 115 m/sec; unit weight of soil = 19 kN/m3 ; mass density
of soil = 1.936 kN sec2 /m4 .
Dynamic shear modulus, G = 1.936 × 115 × 115 = 25614 kN/m2 .

9.78 × 3.2
Base Area = 9.78 m × 3.2 m; equivalent radius r0 = = 3.156 m
π
4Gr0 4 × 25614 × 3.156
Thus vertical spring stiffness of soil = = =
(1 − ν) 0.7
461930 kN/m
Weight of top deck = 9.78 × 3.2 × 0.9 × 25 = 704.16 kN

Weight of column = 6 × 0.36 × 3.625 = 194.4 kN
Weight of machine = 22.1 + 195 + 2 × 16.65 + 54 + 100 = 404 kN
Weight of base mat = 9.78 × 3.2 × 1.5 × 25 = 1173.6 kN

Total vertical Load = 704.16 + 194.4 + 404 + 1173.6 = 2476.16 kN
Pv 2476.16
δs = = = 5.36046 × 10−3 m
Kv 461930
Total displacement δv = δ1 + δ2 + δ3 + δ4 + δs = 1.37549 × 10−5 + 5.36046 ×

10−3 = 5.37422 × 10−3 m
30 30
Knowing fv = √ cpm; we have fs = √ = 409 rpm
δv 5.37422 × 10−3
Calculation of horizontal frequency
For horizontal frequency we know that

1 1
2 2
2 2 N
2 3 Khi Ih
3
(fn )h = 30 α0 ± α0 s − i=1
2 c.p.m
N J
i=1 Wi ϕ
1
Here, Khi = Lateral stiffness of the ith frame i and Khi = where δhi =
δhi
h3 (2 + 3ψ)
12EIc (1 + 6ψ)
Wi = total weight of the ith frame plus weight of the machine plus weight of
the transverse beam and the longitudinal beams, Jφ = Mass moment of inertia ∼ =
N
i=1 Wi Xgi ; Xg = distance of weight W from the resultant center of mass point
2

G; Ih = N i=1 Khi Xhi ; Xh = distance of each frame from the centre of rigidity
2
H,
N
1 2 N i=1 K hi K hi I h
and α0 = e + i=1
N
+
2 Jϕ i=1 Wi
Jϕ

1 3 × 1.0717 × 106 3 × 1.0717 × 106
Here, α0 = 0.1902 +
2 1724.3 × 9.81 144.416 × 9.81

2 × 2.257 × 107
+ = 2472.42
1724.3 × 9.81
Here all the data within the parenthesis were calculated while doing the
calculation based on Barkan’s method.
1
2 N
2 Khi Ih
3
Thus (fn )1 = 30 α0 − α0 − i=1
2
N J
i=1 Wi ϕ

1
2
2 3 × 1.0717 × 106 2 × 2.257 × 107
3
or, (fn )1 = 30 2472.4 − 6112853 − ×
144.416 × 9.81 1724.3 × 9.81
√
= 30 2472.4 − 56797
√
➔ (fn )1 = 30 2472.4 − 238 = 1418 rpm(148 rad/sec);
√
(fn )2 = 30 2472.4 + 238 = 1562 rpm(164 rpm/sec).
Calculation of Horizontal amplitude as per Figure 2.18.19
4590 4590
58 kN 80 kN 100 kN
Figure 2.18.19 Top deck slab with transverse load.
The resultant of the Horizontal dynamic load acts at
80 × 4.59 + 58 × 4.59 × 2
x̄ = = 3.78 m
100 + 58 + 80
Thus eccentricity between center of rigidity and x̄ is 0.81 m is
K CK X
Ci = C N hi + e N hi hi hence for frame 1 we have
2
i=1 Khi i=1 Khi Xhi
1.0716×106 238×1.0716×106 ×4.59

C1 = 238 + 0.81 = 100.33 kN
3×1.0716× 106 2×2.257×107

Ci 100.33
Knowing δhi = we have, δh1 = = 9.36264 × 10−5
Khi 1.0717 × 106
As the foundation is under tuned thus maximum amplitude at transient
condition is given by
δh1 = 9.36264 × 10−5 × 7.85 = 7.34967 × 10−4 m.
For frame 2, we have
1.0716 × 106 238 × 1.0716 × 106 × 0.0

C2 = 238 + 0.81 = 79.33 kN
3 × 1.0716 × 106 2 × 2.257 × 107
Ci 79.33
Khi 1.0717 × 106

δh2 = 7.4277 × 10−5 × 7.85 = 5.8307 × 10−4 .
For frame 3 we have
1.0716 × 106 238 × 1.0716 × 106 × −4.59

C1 = 238 + 0.81 = 58.33 kN
3 × 1.0716 × 10 6 2 × 2.257 × 107
Ci 58.33
Khi 1.0717 × 106
δh3 = 5.44306 × 10−5 × 7.85 = 4.273 × 10−4 m
4 Based on 2D soil structure interaction model

Calculation in vertical direction
The mathematical model for this case is as shown in Figure 2.18.20. Mathe-
matical model of the turbine foundation with soil spring in vertical direction.

The equation of motion is given by

⎡ ⎤⎧ ⎫ ⎡ ⎤⎧ ⎫
m1 0 0 ⎨ ⎪z̈1 ⎪
⎬ c 1 + c2 −c2 0 ⎪ ⎨ż1 ⎪
⎬
⎢ ⎥ ⎢ ⎥
⎣0 m2 0 ⎦ z̈2 + ⎣ −c2 c2 + c 3 −c3 ⎦ ż2
⎪ ⎪ ⎪ ⎪
0 0 m3 ⎩z̈3 ⎭ 0 −c3 c3 ⎩ż3 ⎭
⎡ ⎤⎧ ⎫ ⎧ ⎫
k1 + k2 −k2 0 ⎨z1 ⎬ ⎨ 0 ⎬
+ ⎣ −k2 k2 + k 3 −k3 ⎦ z2 = 0 sin ωm t
⎩ ⎭ ⎩ ⎭
0 −k3 k3 z3 Pv
m3
k3
m2
k2
m1
k1
Figure 2.18.20 Mathematical model of the turbine foundation with soil spring in vertical
direction.
Here, mass of column
0.36 × 3.6 × 6 × 25
mc = = 19.816
9.81
Here m1 = Mass of the bottom raft
9.78 × 3.2 × 1.5 × 25

m1 = + 5.944 = 125.57 ≡ 126 kN · sec2 /m
9.81

m2 = 0.25 times the weight of the transverse girder + weight of machine from
longitudinal girder + self weight from longitudinal girder + 0.3 times the weight
of column.

24 + 13.5 48 + 13.5
m2 = × 2.6 × 2 + × 2.6 × 0.25
9.81 9.81
4.59 × 2 × (32 + 13.5) 16 × 1.585 × 2
+ +
9.81 9.81
99.4 × 2 + 30.2 + 141 0.3 × 6 × 0.36 × 3.6 × 25
+ +
9.81 9.81
= 100.451 ≡ 100 kN · sec2 /m
m3 = Concentrated mass on transverse girder + 0.45 times the self weight

For frame 1 we have
6.7 (24 + 13.5)

m31 = + × 2.6 × 0.45 = 5.15 kN · sec2 /m
9.81 9.81
For frame 2 we have
18.2 (48 + 13.5)

m32 = + × 2.6 × 0.45 = 9.19 kN · sec2 /m
9.81 9.81
For frame 3 we have
10.7 (24 + 13.5)

m33 = + × 2.6 × 0.45 = 5.56 kN · sec2 /m
9.81 9.81
Thus for the complete frame we have, m3 = 5.15 + 9.19 + 5.56 = 19.9 ≡ 20
The mass matrix is thus given by
⎡ ⎤
126 0 0
[M] = ⎣ 0 100 0 ⎦
0 0 20
Calculation of stiffness matrix
In vertical direction the displacement of the transverse girder is given by
L3 (1 + 2ψ) 3L
δv = +
96EIb (2 + ψ) 8GAb

2.63 × (1 + 2 × 5.257) 3 × 2.6

δv = +
96 × 1.0935 × 10 × (2 + 5.257) 8 × 1.5 × 108 × 0.54
7
= 3.86014 × 10−8
3
Thus for three frames we have, k3 = = 7.772 × 107
3.86014 × 10−8
2EAc 2 × 3 × 108 × 0.36
For columns we have, k2 = = × 3 = 18.0 × 107
h 3.6
For the soil the equivalent spring stiffness is given by
4Gr0
k1 = = 461930
1−ν
⎡ ⎤
k1 + k 2 −k2 0
[K] = ⎣ −k2 k2 + k3 −k3 ⎦
0 −k3 k3
⎡ ⎤
18.0461930 −18.0 0
=⎣ −18 25.772 −7.772⎦ × 107
0 −7.772 7.772
Thus for eigen value solution we have77

⎡ ⎤
180461930 − 126λ −180000000 0
⎣ −180000000 257720000 − 100λ −77720000 ⎦ = 0
0 −77720000 77720000 − 20λ
This gives the eigen values and the corresponding three natural frequencies as
⎡ ⎤
0.002 × 106 0 0
[λ] = ⎣ 0 2.5641 × 106 0 ⎦
0 0 5.3294 × 10 6
The corresponding eigen vector are given by

⎡ ⎤
0.5768 −0.3675 0.1266
[ϕ] = ⎣0.5775 0.2983 −0.3454⎦ ;
0.5778 0.8771 0.9299
⎡ ⎤
44.721 0 0
[ω] = ⎣ 0 1601 0 ⎦ rad/sec
0 0 2308
77 We have solved the eigen problem in Math-Cad directly.

Based on orthogonal transformation [ϕ]T [M][ϕ]

⎡ ⎤
81.9436
[ϕ]T [M][ϕ] = ⎣ 42.1484 ⎦
31.2429
Thus the scale factors are given by

√ √
Mr1 = 81.9436 = 9.052; Mr2 = 42.1484 = 6.492;
√
Mr3 = 31.2429 = 5.5895.
Thus the normalised eigen vector is given by

⎡ ⎤
0.06372 −0.05661 0.02265
[ϕ]n = ⎣0.06379 0.04595 −0.06179⎦
0.06383 0.13510 0.16636
Calculation of damping matrix

√
Critical damping, Cc = 2 km
Let the damping ratio for RCC structure be 5%.
Thus critical damping Cc is given by

c3 = 0.05 × (2 7.772 × 107 × 20) = 3942.6;

c2 = 0.05 × (2 18.00 × 107 × 100) = 13416
The soil spring is calculated based on Richart’s formula as shown hereafter78
(1 − ν)Wf 0.7 × 1173.6 0.425

Bz = = = 0.3438; Dz = = 0.7247
4ρs r30 4 × 19 × (3.156)3 Bz
√
c1 = 0.7247 × (2 461930 × 126) = 11057.6
Thus the damping matrix is given by

⎡ ⎤ ⎡ ⎤
c1 + c2 −c2 0 24474 −13416 0
[C] = ⎣ −c2 c2 + c 3 −c3 ⎦ = ⎣−13416 17359 −3943⎦
0 −c3 c3 0 −3943 3943
78 For further details refer to the section of Design and Analysis of Block Foundation.

Thus on orthogonal transformation for each individual mode we have

⎡ ⎤
44.8982
[ϕ]T [C][ϕ] = ⎣ 207.9 ⎦
306.57
→ 2D1 ω1 = 44.8902 ⇒ D1 = 0.502
Calculation for load vectors

The total vertical dynamic force is given by, Pv = 44 + 8.3 + 130 × 2 + 25 × 2 =
362.3 kN
Performing the operation
⎡ ⎤⎧ ⎫
0.06372 0.06379 0.06383 ⎨ 0 ⎬
[ϕ]T ⎣
n {P} = −0.05661 0.04595 0.13510⎦ 0 sin 534t
⎩ ⎭
0.02265 −0.06179 0.16636 362.3
⎧ ⎫
⎨23.1256⎬
= 48.9467 sin 534t
⎩ ⎭
60.2772
Thus the three uncoupled equation of motion is given by
ξ̈1 + 44.89ξ̇1 + 2000ξ1 = 23.1256 sin 534t
ξ̈2 + 207.89ξ̇2 + 2564100ξ2 = 48.9467 sin 534t
ξ̈3 + 207.89ξ̇3 + 2564100ξ3 = 60.2772 sin 534t
ξ1 = 0.502, ξ2 = 0.0649 and ξ3 = 0.0664
Based on the above the displacement vector is given by

3
pi sin ωm t
ξi =
i=1 (1 − r2i )2 + (2Di ri )2
Once we know the displacement vectors in un-coupled state the displacement

in the global structural co-ordinate is given by
{Z} = [ϕ]{ξ }
The results are shown hereafter in tabular form

Case 1
Operating frequency of the machine = 534 rad/sec
Eigen Eigen Eigen

Natural Corrected vector 1st vector 2nd vector 3rd Damping Disp
Amplitude frequency frequency mode mode mode P ratio r M.F. (uncoupled) Z
z1 44.72 53.664 0.06372 −0.05661 0.2265 23.13 0.502 9.950 0.01014 1.17 × 10−04 8.96 × 10−06

z2 1601 1280.8 0.06379 0.04595 0.06179 48.95 0.065 0.416 1.20780 2.304 × 10−05 9.30 × 10−06
z3 2308 1846.4 0.06383 0.1351 0.16636 60.28 0.066 0.289 1.09033 1.235 × 10−05 1.26 × 10−05
Case 2
Transient frequency during start and stop of machine = 44.72 rad/sec
Eigen Eigen Eigen

Natural vector 1st vector 2nd vector 3rd Damping
Z frequency mode mode mode P ratio r M.F. Disp (uncoupled) Z
Z1 44.72 0.06372 −0.05661 0.2265 23.1256 0.502 1 0.996016 1.1517 × 10−02 7.3537 × 10−04
Z2 1601 0.06379 0.04595 0.06179 48.9467 0.065 0.027933 1.000774 1.9111 × 10−05 7.3627 × 10−04
Z3 2308 0.06383 0.1351 0.16636 60.2772 0.066 0.019376 1.000372 1.1320 × 10−05 7.3774 ×10−04
The results are compared hereafter by bar chart shown in Figure 2.18.21.
Vertical amplitude columns

0.8
0.7
0.6
Amplitude (mm)
0.5
0.4 Vertical amplitude

columns
0.3
0.2
0.1
0
Operating Ist Transient
Operating Case
Vertical amplitude Transverse girder

0.8
0.7
0.6
Amplitude (mm)
0.5
Vertical amplitude
0.4
Transverse girder
0.3
0.2
0.1
0
Operating Case
Vertical amplitude Bottom Raft

0.8
0.7
Amplitude (mm)
0.6
0.5
0.4 Vertical amplitude

Bottom Raft
0.3
0.2
0.1
0
Operating Case
Figure 2.18.21 Comparison of transient and operating response.

Analysis in coupled horizontal and rocking mode
The mathematical model for the turbine foundation for this mode is perceived
as Figure 2.18.22.
m0, Jφ y
Kh
h
Kθ
Kx
θ u
Figure 2.18.22 2D mathematical model for coupled translation and rocking including soil springs.
The un-damped equation of motion for free vibration in coupled horizontal

and rocking motion is given by
⎡ ⎤⎧ ⎫
m0 m0 e m0 m0 h ⎪
⎪ ÿ ⎪
⎢ m0 e Jϕ + m0 e2 ⎥ ⎨ ϕ̈ ⎪
⎬
⎢ m 0 e m 0 eh ⎥
⎣ m0 m0 e m0 + m f m0 h ⎦ ⎪ Ü ⎪
⎪
⎩ ⎪ ⎭
m0 h m0 eh m0 h Jθ + m0 h 2 θ̈
⎡ ⎤⎧ ⎫
Kh 0 0 0 ⎪ ⎪y⎪ ⎪
⎢ 0 Kh e2 + Iϕ 0 0 ⎥ ⎨ϕ ⎬
+⎢ ⎣0
⎥ =0
0 Ky 0 ⎦ ⎪ ⎪ U⎪
⎩ ⎪ ⎭
0 0 0 Kθ θ
Calculation of mass matrix

Here
m0 = mass of the top deck + weight of the machine + 0.3 times the weight
of the column
e = distance between the centre of gravity and centre of rigidity

h = height between centre of the top deck mass to the centre of the bottom
raft mass
mf = mass of the bottom raft.
Jφ = mass moment of inertia of top deck in plan for the lumped masses @

Jφ = N 2
i=1 mi Xgi
Jθ = mass moment of inertia of bottom raft in transverse plane
Referring to calculation done in Barkan’s method earlier we have

N
m0 = 144.416 kN · sec2 /m; Jφ = 2
mi Xgi = 1724.3;
i=1
e = 4.59 − 4.40 = 0.19 m
44.03 × 0.45 + 62.326 × 0.45 + 38.06 × 0.45

z̄ = = 0.45 (refer to calcu-
144.416
lations based on Barkan’s method for individual mass data)
Thus, h = 3.6 + (0.9 − 0.45) + 1.5/2 = 4.8 m
9.78 × 3.2 × 1.5 × 25

mf = = 120 kN · sec2 /m;
9.81
m 2 120
Jθ = (ly + lz2 ) = (3.22 + 1.52 ) = 125
12 12
Thus
⎡ ⎤
m0 m0 e m0 m0 h
⎢ m0 e Jϕ + m0 e2 m0 e m0 eh ⎥
[M] = ⎢
⎣ m0
⎥
m0 e m0 + m f m0 h ⎦
m0 h m0 eh m0 h J θ + m 0 h2
⎡ ⎤
144.416 27.44 144.416 693
⎢ 27.44 1730 27.44 132 ⎥
=⎢
⎣144.416
⎥ kN · sec2 /m
27.44 264 693 ⎦
693 132 693 3452
Calculation of stiffness matrix

The stiffness matrix of the frame including the soil spring is given by
⎡ ⎤
Kh 0 0 0
⎢0 Kh e2 + Iϕ 0 0⎥
[K] = ⎢
⎣0
⎥
0 Ky 0⎦
0 0 0 Kθ
Here shear wave velocity of the soil = 115 m/sec

Unit weight of soil γs = 19 kN/m3 ; Poisson’s ratio of soil νs = 0.3

19
Thus dynamic shear modulus of soil G = (115)2 = 25614 kN/m2
9.81

9.78 × 3.2 4 9.78 × (3.2)
3
ry = = 3.156 m; rθ = = 2.414 m
π 3π
32 (1 − νs ) Gry 32 × 0.7 × 25614 × 3.186

Thus, Ky = = = 3.93645 ×
7 − 8νs 7 − 2.4
105 kN/m
8Gr3θ 8 × 25614 × (2.414)3

Kθ = = = 13.73 × 105 kN/m
3(1 − νs ) 2.1
Referring to the table for calculation of stiffness in Barkan’s method, done

earlier

3
Khi = 3 × 1.072 × 106 = 3.216 × 106 kN/m;
i=1

N
2
Iφ = Ki Xhi = 2 × 2.257 × 107 = 4.514 × 107
i=1
We had calculated earlier

Thus Kh e2 + Iφ = 3.216 × 107 × 0.192 + 4.514 × 107 = 4.5256 × 107 kN/m
Thus the stiffness matrix can be represented by
⎡ ⎤
3.216 × 106 0 0 0
⎢ 0 4.5256 × 107 0 0 ⎥
[K] = ⎢
⎣
⎥
⎦
0 0 3.94 × 105 0
0 0 0 1.373 × 106
Calculation for eigen value

⎡
3.216 × 106 − 144.416λ −27.44 |
⎢ −27.44 4.5256 × 10 7 − 1730λ |
⎢
⎣ −144.416 −27.44 |
−693 −132 |
⎤
| −144.416 −693
| −27.44 −132 ⎥
⎥=0
| 3.94 × 10 − 264
5 −693 ⎦
| −693 1.373 × 106 − 3452

Solving the above we have79

⎡ ⎤
644590
⎢ 340 ⎥
[λ] = ⎢
⎣
⎥
⎦
3650
26260
which gives the natural frequency as

⎡ ⎤
803
⎢ 18.43 ⎥
[ω] = ⎢
⎣
⎥ rad/ sec80
⎦
60.41
157
The corresponding eigen vectors are given by

⎡ ⎤
0.9812 0.0675 −0.0128 −0.0043
⎢−0.0005 0.0009 −0.0002 −0.9992⎥
[ϕ] = ⎢⎣−0.0411 0.6156 0.9761
⎥
0.0051 ⎦
−0.1888 0.7851 −0.2170 0.0386
Now performing
⎡ the operation [ϕ]T [M][ϕ]⎤we have,
4.9
⎢ 2984.4 ⎥
[ϕ]T [M][ϕ] = ⎢
⎣
⎥
⎦
120.8
1722.3
Thus the scaled factors are given by
√ √
Mr1 = 4.9 = 2.213; Mr2 = 2984.4 = 54.63;
√ √
Mr3 = 120.8 = 10.99, and Mr4 = 1722.3 = 41.5.
Thus the normalized eigen vector is given by

⎡
0.4438 1.2392 × 10−3 |
⎢−2.2593764 × 10−4 1.6474 × 10−5 |
[ϕ]n = ⎢
⎣ −0.018798 0.0112685 |
−0.085314 0.0142961 |
79 We have solved the eigen value problem directly in Math CAD.

80 Reader to check that
matrix reduces to M =
in case we ignore the effect of soil, the mass
144.416 27.44 3.216 0
and the stiffness matrix K = × 106 and the above on
27.44 1730 0 45.256
eigen value solution gives natural frequencies as 148 rad/sec and 163 rad/sec which is exactly
same as what we have obtained based on Barkan or Major’s method.

⎤
| −1.1646952 × 10−3 −1.0361446 × 10−4
| −1.8198362 × 10−5 −0.0240771 ⎥
⎥
| 0.08882 1.2289157 × 10 ⎦
−4
| −0.0197452 9.30212048 × 10−4
Calculation of damping matrix

Let the damping ratio for the RCC frame be 5% then we have
√
Frame No. Mass Kh Xh Cc = 2 km Ch (5% of Cc ) Ch · xh2
1 44.03 1.0717 × 106 4.59 13738 687 14474

2 62.326 1.0717 × 106 0.0 16346 817 0.0
3 38.06 1.0717 × 106 −4.59 12773 639 13463
144.416 2143.05 27937
Calculation for soil damping
(7 − 8ν) Wf (7 − 2.4) × 1177.2

By = = = 0.404 and
32 (1 − ν) γs r3x 32 × 19 × 0.7 × (3.156)3
0.288 0.288
Dy = √ =√ = 0.452
Bx 0.404

Cy = 2Dy Ky m = 2 × 0.452 3.94 × 105 × 120 = 6216 kN · sec /m
0.375(1 − υ)Jφ g 0.375 × 0.7 × 125 × 9.81
Bφ = = = 0.2066
γs r5
φ 19 × (2.414)5
0.15 0.15
Dφ = = = 0.274
(1 + Bφ ) Bφ 1.2066 × 0.454
√
Thus Cφ = 2Dφ Kφ Jφ = 2 × 0.274 × 1.373 × 106 × 125 = 7179
KN · sec/m.
The equation being dynamically coupled, the damping matrix is given by
⎡ ⎤
Ch 0 0 0
⎢0 Ch e2 + Ch Xh2 0 0⎥
[C] = ⎢
⎣0
⎥
0 Cy 0⎦
0 0 0 Cθ
Substituting the values calculated above we have

⎡ ⎤
2143 0 0 0
⎢ 0 28014 0 0 ⎥
[C] = ⎢⎣ 0
⎥
0 6216 0 ⎦
0 0 0 7179

Now performing the operation [ϕ]T [C][ϕ] and for each mode separately we
have
⎡ ⎤
476.532 0 0 0
⎢ 0 2.26 0 0 ⎥
[2Di ωi ] = ⎢
⎣
⎥
⎦
0 0 51.84 0
0 0 0 16.2462
Thus dividing each term of the above matrix by 2ωi , we have

⎡ ⎤
0.296 0 0 0
⎢ 0 0.0605 0 0 ⎥
[Di ] = ⎢
⎣ 0
⎥
0 0.429 0 ⎦
0 0 0 0.052
Calculation of load vector

⎧ ⎫
⎪
⎪ 238 ⎪ ⎪
⎨ ⎬
1092.42
Here the load vector is given by {Ph } = , here Mφ = N i=1 Phi Xhi
⎪
⎪ 0.0 ⎪
⎪
⎩ ⎭
1511.3
N
and Mθ = i=1 Phi h and h = 3.6 + 0.9 + 1.1 + 0.75 = 6.35 m, this is from the
center line of shaft to the c.g of the bottom raft. ⎧ ⎫
⎪−23.5575⎪
⎪ ⎪
⎨ ⎬
21.9186
Performing the operation [ϕ] {P} we have, [ϕ] {P} =
T T
⎪ −30.138 ⎪
⎪ ⎪
⎩ ⎭
−24.9211
Thus the four uncoupled equation of motion is given by
ξ̈1 + 476.532ξ̇1 + 644590ξ1 = −23.5575 sin 534t

ξ̈2 + 2.26ξ̇2 + 340ξ2 = 21.9186 sin 534t
ξ̈3 + 51.849ξ̇3 + 3650ξ3 = −30.138 sin 534t
ξ̈4 + 16.2462ξ̇4 + 24670ξ4 = −24.9211 sin 534t
Once we know the displacement vectors in un-coupled state the displacement

in the global structural co-ordinate is given by

4
pi sin ωm t
{Y} = [ϕ]{ξ } → ξi =
i=1 (1 − r2i )2 + (2Di ri )2
The results are tabulated as follows: Here we have calculated the response for
four cases
1 One during normal operation
2 For three transient case during start and stop of the machine

Displacement under normal operation
Operating frequency = 534 rad/sec
Eigen Eigen Eigen Eigen Uncoupled

Corrected Vectors 1st Vectors 2nd Vectors 3rd Vectors 4th force Damping Frequency M.F. Uncoupled Net
Displacement Frequency frequency mode mode mode mode vector Ratio Ratio - amplitude displacement
Y 802 642 0.443 1.24 ×10−03 −1.16 ×10−03 −1.04 ×10−04 −23.6 0.30 0.83 1.72 −6.32 ×10−05 −2.77 ×10−05
φ 18 22 −2.26 ×10−04 1.65 ×10−05 −1.82 ×10−05 −2.41×10−02 21.9 0.06 24.15 0.00 1.11 ×10−04 −3.48 ×10−06
U 60 72 −0.0188 0.11268 0.08882 1.23 ×10−04 −30.13 0.43 7.37 0.02 0.000154 −1.17 ×10−08
θ 157 188 −0.0853 0.01429 0.01975 9.30 ×10−04 −24.9 0.05 2.83 0.14 −0.00014 9.88 ×10−06
Displacement under 1st transient
Transient frequency = 18.43 rad/sec

Vectors 1st Vectors 2nd Vectors 3rd Vectors 4th Damping Ratio Uncoupled Net
Displacement Frequency mode mode mode mode vector force Frequency M.F. amplitude displacement
Y 802.8 0.44338 1.24 −1.16×10−03 −1.04×10−04 −23.55 0.296 0.02 1.00 −3.672×10−05 6.55×10−04
φ 18.43 −2.26×10−04 1.65×10−05 −1.82×10−05 −2.41×10−05 21.91 0.061 1.00 8.26 0.5333 3.36×10−05
U 60.41 −0.018798 0.11268 0.08882 1.23×10−04 −30.13 0.429 0.31 1.06 −0.00875 5.93×10−02
θ 157 −0.085314 0.01429 0.019745 9.30×10−04 −24.9 0.052 0.12 1.01 −0.001025 7.80×10−03

Displacement under 2nd transient
Transient frequency = 60.41 rad/sec

Vectors 1st Vectors 2nd Vectors 3rd Vectors 4th force Damping Frequency M.F. Uncoupled Net
Displacement Frequency mode mode mode mode vector Ratio Ratio - amplitude displacement
Y 802.8 0.44338 1.24×10−03 −1.16×10−03 −1.04×10−04 −23.55 0.296 0.08 1.00 −3.687×10−05 −3.18×10−05
φ 18.43 −2.26×10−04 1.65×10−05 −1.82 ×10−05 −2.41×10−02 21.91 0.0605 3.28 0.10 0.006617 2.88×10−05
ϒ 60.41 −0.018798 0.112685 0.08882 1.23×10−04 −30.13 0.429 1.00 1.17 −0.009625 −1.09×10−03
θ 157 −0.085314 0.014296 0.019745 9.30×10−04 −24.92 0.052 0.38 1.17 −0.00118543 2.87 ×10−04
Displacement under 3rd transient
Transient frequency = 157 rad/sec

Vectors 1st Vectors 2nd Vectors 3rd Vectors 4th force Damping Frequency M.F. Uncoupled Net
Displacement Frequency mode mode mode mode vector Ratio Ratio - amplitude displacement
Y 802.8 0.44338 1.24 ×10−03 −1.16 ×10−03 −1.04 ×10−04 −23.55 0.296 0.20 1.03 −3.789 ×10 − 05 −1.31 ×10−05
φ 18.43 −2.26×10−04 1.65 ×10−05 −1.82×10−05 −2.41 ×10−02 21.91 0.0605 8.52 0.01 0.00090156 2.34 ×10−04
U 60.41 −0.018798 0.112685 0.08882 1.23×10−04 −30.13 0.429 2.60 0.16 −0.001334 1.77×10−05
θ 157 −0.085314 0.014296 0.019745 9.30 ×10−04 −24.92 0.052 1.00 9.62 −0.009722 3.35×10−05

Net amplitude of the frames

The net amplitude of the frame is given by the expression
yinet = yi + Ui + Xhi ϕ + hθ
where h = 4.8 m and Xhi = +4.59 m, 0.0, −4.59 m respectively for

Frame 1, 2 and 3.
The results are as shown hereafter
Operation condition Frame 1 Frame 2 Frame 3
Normal operating case 3.57 × 10−05 1.97 × 10−05 3.78 × 10−05

1st Transient (18.43 rad/sec) 9.76 × 10−02 9.74 × 10−02 9.73 × 10−03
2nd Transient (60 rad/sec) 1.4 × 10−03 1.27 × 10−03 1.14 × 10−03
3rd Transient (157 rad/sec) 1.20 × 10−03 1.30 × 10−04 −9.45 × 10−04
The results are shown based on bar chart as given in Figure 2.18.23.
Amplitude of the frames

1.20E-01
Displacement of top deck
1.00E-01
8.00E-02
Normal operating force
6.00E-02
1st transient
4.00E-02 2nd transient
2.00E-02 3rd transient
0.00E+00
-2.00E-02 Frame1 Frame2 Frame3

Frame Numbers
Figure 2.18.23 Comparison of displacement operating vs transient.
The results of natural frequencies are compared based on various methods.

Frequency in vertical direction
Mode number Rausch Barkan Major 2D soil structure interaction
1st Mode – – 409 427

2nd Mode 8318 10265 8318 15288
3rd Mode – 21575 – 22040
Frequency in horizontal direction
Mode number Rausch Barkan Major 2D soil structure interaction
1st Mode – – – 176

2nd Mode 1616 1413 1418 577
3rd Mode – 1557 1562 1547
4th Mode – – – 7667

Comparison of results with time history response

Here we compare the results obtained above with time history response which we
have stated earlier as the most appropriate method of analysis for such coupled
soil-structure interaction analysis
We show below the plots (Figures 2.18.24 and 25) for the amplitude
of vibration and displacement of the frames under operating frequency of
534 radians/sec.
Amplitude based on Newmark - method

0.00004
0.00003
0.00002
Amplitude
0.00001 Y
0
-0.00001
0
19
0. 8
0. 7
5
0. 5
0. 5
75
25
75
25
75
25
75
3
5
09
28
47
0.
0.
0.
04
14
23
33
42
52
61
-0.00002
0.
0.
0.
0.
0.
0.
-0.00003
-0.00004
Time steps in seconds
Figure 2.18.24 Amplitudes under operating condition having frequency @ 534 rad/sec.
Amplitude of frames at top deck

2.00E-04
1.50E-04
Amplitude (meter)
1.00E-04
Y1
5.00E-05
Y2
0.00E+00 Y3
0.03
0.07
0.13
0.16
0.23
0.26
0.29
0.33
0.36
0.39
0.42
0.46
0.49
0.52
0.55
0.58
0.62
0.1
0.2
0
-5.00E-05
-1.00E-04
-1.50E-04
Time steps (sec)
Figure 2.18.25 Amplitude of displacement of the top deck under operating frequency of
534 rad/sec.
Under transient load like Major we assumed the operating frequency in reso-
nance with the natural frequency instantaneously and considering the function,
sin ωm t = 1.
For time history response we consider the operating frequency equal to first,
second and third transient frequency respectively and find out the transient
response (peak amplitude). The results are plotted graphically in Figures 2.18.26
through 31.

Amplitude based on Newmark- method

0.01
0.008
0.006
0.004 Y
Amplitude
0.002
0
0
0.45
1.35
2.25
3.15
4.05
4.95
5.85
0.9
1.8
2.7
3.6
4.5
5.4
-0.002 U
-0.004
-0.006
-0.008
-0.01
Figure 2.18.26 Amplitudes at the first transient frequency @ 18.43 rad/sec.

5.00E-02
4.00E-02
3.00E-02
Amplitude (meter)
2.00E-02
Y1
1.00E-02
Y2
0.00E+00 Y3
0.3
0.6
0.9
1.2
1.5
1.8
2.1
2.4
2.7
3.3
3.6
3.9
4.2
4.5
4.8
5.1
5.4
5.7
0
6
-1.00E-02
-2.00E-02
-3.00E-02
-4.00E-02
-5.00E-02
Time steps (sec)
Figure 2.18.27 Amplitude of top deck at the first transient frequency @ 18.43 rad/sec.

0.0015
0.001
Y
Amplitude
0.0005
0
U
0
3
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
2.2
2.4
2.6
2.8
-0.0005
-0.001
-0.0015
Figure 2.18.28 Amplitudes at the second transient frequency @ 60 rad/sec.


2.50E-03
2.00E-03
1.50E-03
Amplitude (meter)
1.00E-03
Y1
5.00E-04
Y2
0.00E+00
Y3
3
2.1
0
2.4
0.3
1.8
0.6
1.2
0.9
1.5
2.7
0.15
0.45
0.75
1.05
1.35
1.65
1.95
2.25
2.55
2.85
-5.00E-04
-1.00E-03
-1.50E-03
-2.00E-03
-2.50E-03
Time steps (sec)
Figure 2.18.29 Amplitudes of top deck at the second transient frequency @ 60 rad/sec.

0.0003
0.0002
0.0001 Y
Amplitude
0
U
0
0.5
0.05
0.09
0.14
0.18
0.23
0.27
0.32
0.36
0.41
0.45
0.54
0.58
-0.0001
-0.0002
-0.0003
Figure 2.18.30 Amplitudes at third transient frequency @ 157 rad/sec.

2.00E-03
1.50E-03
Amplitude (meter)
1.00E-03
5.00E-04
Y1
0.00E+00 Y2
Y3
03
0. 5
08
13
0. 5
18
0. 5
28
33
0. 5
38
43
0. 5
48
0. 3
57
23
55
0
6
2
4
0
5
5
0.
0.
-5.00E-04
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
-1.00E-03
-1.50E-03
-2.00E-03
Time steps (sec)
Figure 2.18.31 Amplitude of top deck at third transient frequency @ 157 rad/sec.

Discussions of the results

The basic amplitudes as obtained by the two methods are shown hereafter
Modal Time Critical

Sl. No. Displacements response history frequency Remarks
1 Y 2.77 × 10−05 2.617 × 10−05 534 This is under operating case

2 φ 2.34 × 10−04 2.09 × 10−04 157 Operating frequency in
resonance with this mode
3 ϒ 1.09 × 10−04 8.33 × 10−04 60.41 Operating frequency
in resonance with this mode
4 θ 7.80 × 10−03 7.38 × 10−03 18.43 Operating frequency in
resonance with this mode
The displacement of the top deck is as shown hereafter
Modal Time Critical

Sl. No. Displacements response history frequency Remarks
1 Y 3.57 × 10−05 1.6 × 10−04 534 This is under operating case

2 Y 1.20 × 10−03 1.31 × 10−03 157 Amplitude at third transient
3 Y 1.4 × 10−03 2.21 × 10−03 60 Amplitude at second transient
4 Y 9.76 × 10−02 4.12 × 10−02 18.43 Amplitude at first transient
Discussion on the results

It is pretty obvious from the results that time history gives a clearer picture so far as
the amplitude is concerned.
While by modal technique, taking sin ωm t = 1, we calculate the maximum amplitude
at various resonance points but actually at the time when the operating frequency
passes this particular frequency sin ωm t = 1 but something lesser than that. Since we
do not know this data as a conservative value this is usually taken as unity. Thus it is
evident from the result that while by time history we get a value of deflection of top
deck as 41 mm this gives about 98 mm by modal technique. While in all other cases
though the order of displacement does remain same they do show discrepancy in top
deck amplitude.
Specially under operating condition while modal response gives amplitude of
deflection of top deck as 3.57E-05 m time history gives an amplitude of 1.6E-04
(about 5 times). This can be attributed to the fact that while by modal response we
only find the steady state part, by time history the transient part of the vibration is
also considered in the response and this possibly gives an initial higher response at
operating frequency.
Thus for resonance check we can follow the modal response technique to find out the
eigen values, however for amplitude check carrying out time history response specially
to check the transient response is much more sensible.

Based on the above calculation following points can be concluded

• Soil has significant influence on the natural frequency of the system.
• Major’s hypothesis of transient response is correct, and there too soil could
play a significant role (Though Major’s method of calculating the response is
conservative).
• Increasing the thickness of bottom raft does not necessarily help. It could reduce
the peak operating frequency response but will not have much effect on the
transient response of the system during start and stop of the machine.
• Transient response can be significant, and all the significant transients to be
checked for the frame.
Though this may not structurally effect the frame (for soil transients would
mostly be rigid body displacements) its effect on the stress level of the connected
piping and load induced at nozzles could be significant.
• For critical structures amplitude should preferably be checked based on time his-
tory response to ensure that amplitudes are within acceptable limit. This should be
done specially under operating frequency where modal technique usually ignores
the transient part and only gives the maximum value for the steady state part.
SUGGESTED READING
1 Barkan, D.D. 1962, Dynamics of Bases and Foundations, McGraw-Hill Book Co. NY.
2 Srinivasalu, P. & Vadiyanathan, C.V. 1977, Handbook of Machine Foundations, Tata
Mcgraw-Hill, New Delhi.
3 Major, A. 1980, Dynamics in Civil Engineering – Analysis and Design, Vols. I–IV,
Akademia Kiado, Budaapest and Collets Holding London.
4 Arya, S.C., O’Neill, M.W. & Pincus, G. 1979, Design of Structures and Foundations for
Vibrating Machines, Gulf Publishing Co., Houston, Texas.
5 Kameswara Rao, N.S.V. 1998, Vibration Analysis and Foundation Dynamics, Wheeler
Publishing, New Delhi.
6 Verma, C.V.J. & Lal, P.K. Ed., Treatise on the design, analysis and testing of High capacity
Turbo Generator foundation, Central Board of Irrigation and Power Publication #262.
We also furnish some selected papers which we feel would

further ameliorate your insight to the problem
1 Almuti, A.M. 1976, ‘Large flexible Turbine foundation’, Methods of Structural Analysis,
ASCE, NY, pp. 707–719.
2 Aneja, I. 1975, ‘Dynamic Response of Systems – Turbine generators on Various Founda-
tions’, Proceedings of the American Power conference, Vol. 37, pp. 528–540.
3 Arya, A.S. & Drewer, R. 1997, ‘Mathematical modeling and computer simulation of ele-
vated foundations supporting vibrating Machinery’, Transaction of IMACS, Vol. XIX,
No. 4, Dec.
4 Design Criteria for Turbine Generator Pedestal 1970, Journal of Power Division, ASCE,
Vol. 96, Jan, pp. 1–22.
5 Kasten, H.L. & Kirkland, W.D. 1970, ‘Spring mounted Turbine Generator Spins Quitely,
Efficiently’, Electric Light and Power, E/D Edition, Nov., pp. 38–40.

6 Shen, G.T. & Stone, N.E. 1975, ‘Natural frequencies of turbine foundation’, Structural
Design of Nuclear Plant Facilities, Vol. II, ASCE, NY, pp. 302–330.
7 Srinivasulu, P. & Lakshmannan, N. 1978, ‘Dynamic response of turbo-generator pedestal’,
ASCE, Spring convention, Pittsburgh, Pensylvania, April, pp. 24–28.
8 Chowdhury, I. & Som, P.K. 1993, ‘Dynamic Pile structure interaction of Boiler Feed Pump
Frame Foundation’, Indian Geotechnical Conference, Vol. 1, pp. 411–414.
9 Task Committee on Turbine Foundations 1987, Design of large steam turbine-generator
foundations, ASCE, NY.
10 Rausch, E. 1959, ‘Maschinen Fundamente und andere dynamisch beanspruchte Baukon-
structionen’, VDI Verlag, Dusseldorf.
11 Wedpathak, A.V., Pandit, V.K. Guha, S.K. 1977, ‘Soil-Foundation interaction under sinu-
soidal and impact type dynamic loads’, Int. Symp. on Soil-Structure Interaction, University
of Roorkee, Roorkee.

SW1451 c002

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

SW1451 c002

Hochgeladen von

Copyright:

Verfügbare Formate

Chapter 2

Analysis and design of machine

• Basic concepts in structural dynamics.

2.1.1 Case history #1

© 2009 Taylor & Francis Group, London, UK

2.1.2 Case history #2

© 2009 Taylor & Francis Group, London, UK

2.2 DIFFERENT TYPES OF FOUNDATIONS

Foundations supporting machines can be classified into the following categories:

• Block foundations resting on soil or piles

2.2.1 Block foundations resting on soil/piles

Figure 2.2.1 Block foundation resting on soil.

© 2009 Taylor & Francis Group, London, UK

Figure 2.2.2 Block foundation on piles.

4 Coal mill foundations

2.2.2 How does a block foundation supporting rotating

© 2009 Taylor & Francis Group, London, UK

2.2.2.1 Resonance check

2.2.2.2 Amplitude check

• Rapid deterioration of the machine due to heavy wear and tear.

© 2009 Taylor & Francis Group, London, UK

Figure 2.2.3 Mathematical model of foundation in 2D.

Figure 2.2.4 Degrees of freedom in space.

2.2.3 Foundation for centrifugal or rotary type

2.2.3.1 Tschebotarioff’s (1953) method

© 2009 Taylor & Francis Group, London, UK

2.2.3.2 Alpan’s (1961) method

where, fn = natural frequency of the foundation in cycles per minute; W = weight

2.2.3.3 Newcomb’s (1951) method

Table 2.2.1 Value of constant, α.

Sl. No. Soil type Value of α

© 2009 Taylor & Francis Group, London, UK

2.2.4 Analytical methods

2.2.4.1 Hsieh’s (1962) method

where, m = mass of the foundation plus the machine; x, θ = displacement vectors in

F1 is usually represented in the form, α1 − α2 a20 .

Now substituting this value of F1 in the above differential equation we have,

For rocking mode,

Representing the above equations as

mx ẍ + cx ẋ + kx x = P0 sin ωm t and I θ̈ + cθ θ̇ + kθ θ = M0 sin ωm t (2.2.8)

© 2009 Taylor & Francis Group, London, UK

mx = (m + α2 ρr30 ) and I = (ϕ + α2 ρr5θ ) (2.2.11)

Table 2.2.2 Values of F1 and F2 .

Mode Poisson’s ratio F1 F2

Vertical (0 < a0 < 1.5) 0.0 4.0 − 0.5a02 3.3 + 0.4a0

© 2009 Taylor & Francis Group, London, UK

2.2.4.2 Barkan or IS-2974 method

2.2.4.2.1 Vertical direction

where, kz = equivalent spring in vertical direction; cz = coefficient of elastic uniform

© 2009 Taylor & Francis Group, London, UK

The natural frequency of the foundation is given by

the amplitude of vertical vibration is given by

where, r = ωm /ωn ; ωm = operating frequency of the machine.

2.2.4.2.2 For coupled horizontal and rocking mode

and the resistive moment is expressed

2.2.4.2.3 For horizontal force, H

where, A = area of base contact; Zc , x, x0 etc. are shown in Figure 2.2.5.

mẍ + H = Px sin ωm t or mẍ + cτ Af (x − Zc φ) = Px sin ωm t (2.2.18)

where m is the mass of the machine foundation.

3 Refer Chapter 5 (Vol. 1) for definition of D’Alembert’s equation.

© 2009 Taylor & Francis Group, London, UK

Jxφ φ̈ − cτ Af Zc x + φ(cφ IA − WZc + cτ AZc2 ) = Mx sin ωm t (2.2.19)

J0 (ωφ2 + ωx2 )ω2 ωφ2 ωx2 J0