Beruflich Dokumente
Kultur Dokumente
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
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.
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 .
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.
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.
• Resonance check
• Amplitude check
• 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.
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
1
1
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.
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.
α 1/4
fn = √ (Af ) (2.2.2)
W
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.
1 Peat 3900
2 Plastic clay 69000
3 Sand 82000
4 Sand stone 111000
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)
(ϕ + α2 ρr5θ )θ̈ + r4θ F2 ρGẋ + α1 Gr3θ θ = M0 sin ωm t (2.2.7)
√
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
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.
kz = cz · Af (2.2.12)
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.
Pz sin ωm t
kz
δz = (2.2.14)
1 − r2
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.
H = cτ Af x0 = cτ Af (x − Zc φ) (2.2.17)
Similarly for the moment equation about the minor axis of the foundation we have
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.
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
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)
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.
• 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
Based on the above equation we will see later how we develop further realistic model
of the coupled horizontal and rocking mode5 .
Table 2.2.3 Values of soil springs as per Richart and Lysmer (1970) model.
Table 2.2.4 Values of soil damping as per Richart and Lysmer (1970) model.
Y
L
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.
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
m 0 ẍ Cx −Cx Zc ẋ
+
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
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.
(2.2.34)
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
,
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
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
Solution:
The complete equation of motion for the foundation under coupled rocking and
sliding mode is given by
m 0 ẍ Cx −Cx Zc ẋ
+
0 Jxφ φ̈ −Cx Zc Cφx + Cx Zc2 φ̇
Kx −Kx Zc x P0
+ = sin ωm t
−Kx Zc Kφx + Kx Zc2 φ M0
3000 −4500 x P0
+ = sin ωm t
−4500 11750 φ M0
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
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
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.
50 0 3000 −4500
[C] = 0.48568 + 0.0727
0 100 −4500 11750
242 −327.15
=
−327.15 902.795
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.
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
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
m mZc ẍ K 0 x
+ x =0 (2.2.43)
mZc mZc2 + J φ̈ 0 Kφ φ
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.
Solution:
The complete equation of motion for the foundation under coupled rocking and
sliding mode based on Lagrange’s formulation is given by
P0
= sin ωm t
M0
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.
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
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
i.e. D2 = = 0.4248.
2 × 12.58
Thus
The table below gives comparative results based static and dynamic coupling
formulations for the examples solved above.
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.
Solution:
The operating frequency of the machine is 750 r.p.m. = (750 × 2 × π )/60 =
78.53 rad/sec.
Figure 2.2.7 gives a comparison of the amplitude value for the static and
dynamic coupling case.
It will be observed that
1.50E-03
-1.50E-03
Time steps
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
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 .
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
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
1
i.e. M.F. = (2.3.2)
(1 − r2 )2 + (2Dz r)2
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
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.
Solution:
750
The operating frequency of the machine is 750 r.p.m. = ×2×π =
60
78.53 rad/sec
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
As we have seen earlier that the frequency separation is greater than 3.5, hence
on orthogonal transformation without damping we have
On comparing the amplitudes for the damped and undamped case we have
the following results.
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:
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.
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.
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
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,
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.
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
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.
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.
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.
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.
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
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
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
Table 2.5.3 Vertical damping coefficients for floating piles as per Novak (1983).
f18,2
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.
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
Table 2.5.4 Values of S̄1 and S̄2 for various Poisson’s ratio.
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.
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.
Table 2.5.6 Values of Su1 and Su2 for various Poisson’s ratio.
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
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.
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.
Machine block
Soil line
h
Zc
Xr
2
f δ2 Zc Zc
cψ = δr40 Gs γs /g Sψ2 + + −δ S̄u2 (2.5.15)
3 r0 r0
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
2000 2000
3000 3000
500
(Typ)
3 6 9
2000
2 5 8
2000
1 4 7
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
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
= 7934 kN-sec/m.
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
18
Gs = × (60)2 = 6605 kN/m2
9.81
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
9
403091 × 9
Kx = = 2418546 kN/m;
1.5
i=1
9
3334 × 9
Cx = = 20004 kN-sec/m.
1.50
i=1
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.
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
2
δ2 Zc Zc
kψf = Gs r0 hS̄ψ1 + Gs r20 h + −δ S̄u1
3 r0 r0
g
N
kψ = [kψ1 + kz1 Xr2 + kx1 Zc2 − 2Zc kxψ1 ] + kψf
1
g
kψ = 9 × [5103158 + 4.924 × 106 × 4 + 403091 × 2.25
g
kψ = 2.342 × 108 kN/m
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
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.
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.
∂ 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.
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
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)
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)
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
Kv = Gb r0 Cb + GDf S1 (2.5.28)
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.
Poisson’s ratio Cb S1
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
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
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.
+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
+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
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.
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
⎡ ⎤⎡ ⎤
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
⎡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.
1
γp AL
m1 = F1 (ξ )2 dξ (2.5.48)
g
0
1
γp AL
m1 = (cos βξ )2 dξ (2.5.49)
g
0
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.
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)
Table 2.5.10 Values of damping coefficients based on Novak and Beredugo (1972).
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)
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 (2.5.57) 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.
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.
1
C1 = r0 ρGS̄2 L (2.5.60)
2
γp AL
m1 = . (2.5.61)
2g
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).
Rotating Machine
Pile Cap
Partially embedded piles
G.L.
L
L1
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
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)
+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
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
+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)
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
Df
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
K2 = GfSfDf
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
Once the stiffness, mass and damping matrices are established, the natural frequency
of the system may be obtained from the standard expression
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 .
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
⎡ ⎤
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.
η pL η pL η pL η pL
η pL η pL η pL η pL
√
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.
Figure 2.5.13 Comparison of bearing pile stiffness for slenderness ratio = 20.
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.
5.0000E+06
4.0000E+06 Kpile(bearing)
3.0000E+06 Novak(bearing)
2.0000E+06
1.0000E+06 Gazetas
0.0000E+00
0
00
00
25
50
0
00
10
50
10
Ep/Gs
Figure 2.5.15 Comparison of bearing pile stiffness for slenderness ratio = 40.
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.
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
Figure 2.5.16 Comparison of friction pile stiffness for slenderness ratio = 40.
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
Figure 2.5.17 Comparison of bearing pile stiffness for slenderness ratio = 80.
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
Figure 2.5.18 Comparison of friction pile stiffness for slenderness ratio = 80.
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
Figure 2.5.19 Comparison of bearing pile stiffness for slenderness ratio = 100.
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
Figure 2.5.20 Comparison of friction pile stiffness for slenderness ratio = 100.
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.
• 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.
• 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
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
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
Again, at z = 0, dx
dz
= θ0 one can have
θ0
C1 = x0 + (2.5.97)
p
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
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
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
L L
K = Ep Ip φi (z)2 dz + GSx1 φi (z)2 dz (2.5.110)
0 0
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
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
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
5X −2β ) − 3Y − 34 η
Ep Ip 4 (1 − e 8
K= 3 3
(2.5.118)
L (η − 1)
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)
Note: The value in parenthesis after Sx1 indicates Ep /Gs value of the soil.
Table 2.5.15 Suggested value of Sx1 for Poisson’s ratio of soil = 0.40.
L/r0
Poisson’s (Slenderness
ratio ratio) Sx1 (250) Sx1 (500) Sx1 (1000) Sx1 (2500) Sx1 (5000) Sx1 (10000)
Note: The value in parenthesis after Sx1 indicates Ep /Gs value of the soil.
Table 2.5.16 Suggested value of Sx1 for Poisson’s ratio of soil = 0.50.
L/r0
Poisson’s (Slenderness
ratio ratio) Sx1 (250) Sx1 (500) Sx1 (1000) Sx1 (2500) Sx1 (5000) Sx1 (10000)
Note: The value in parenthesis after Sx1 indicates Ep /Gs value of the soil.
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
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.
Cz = r0 ρb Gb C̄b + r0 ρG S̄2 Df (2.5.123)
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)
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
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.
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 (2.5.137) 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)
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)
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
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
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
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
P0
K sin ωm t
v(t) = (2.5.149)
(1 − r2 )2 + (2Dr)2
P0
K
v(t) = (2.5.150)
(1 − r )2 + (2Dr)2
2
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
Ep Ip P0
2β 2 − βz βz βz
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
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)
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)
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
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
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)
1
C2 V1 (pL) + C3 V2 (pL) = √ (2.5.172)
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
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
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)
0 0
(2.5.183)
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.
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.
Based on the above argument the error (ε) in the analysis is then given by
ε = ωc − ωf
GSx1 g I1
4 + 1 − ωf2 = 0 (2.5.188)
γp A p I2
where I2 is the corrected upgraded value and Sx2 is as obtained from Table 2.5.17.
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
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
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.
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
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.
Step-2: Showing solver on the verge of optimizing by changing Ep/G value by setting the error
to zero.
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.
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
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
dx
Again considering at z = 0, = θ0 , we have
dz
θ0
C1 = x0 + (2.5.195)
q
−qz θ0
x=e x0 cos qz + x0 + sin qz (2.5.196)
q
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
−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
−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.
Shape function
1
0.5 F(z)
9
15
45
75
0
0.
0.
0.
0.
0.
0.
-0.5
z/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
Gs Df D2f
Kθ = Gb r30 Cθ1 + Sθ1 + Sx1 (2.5.204)
G b r0 3r20
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
L L
Kij = Ep Ip ϕi (z)ϕj (z)dz + Gr20 Sθ1 ϕi (z)ϕj (z)dz (2.5.207)
0 0
L L
2
Kθ = Ep Ip ϕ (z) dz + Gr20 Sθ1 ϕ(z)2 dz (2.5.208)
0 0
⎡ ⎤
−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.
L/r0
Poisson’s (slenderness
ratio ratio) Sθ1 (250) Sθ1 (500) Sθ1 (1000) Sθ1 (2500) Sθ1 (5000) Sθ1 (10000)
Note: The value in Parenthesis after Sθ1 depicts the value of Ep /Gs value of the soil.
L/r0
Poisson’s (slenderness
ratio ratio) Sθ1 (250) Sθ1 (500) Sθ1 (1000) Sθ1 (2500) Sθ1 (5000) Sθ1 (10000)
Note: The value in Parenthesis after Sθ1 depicts the value of Ep /Gs value of the soil.
L/r0
Poisson’s (slenderness
ratio ratio) Sθ 1 (250) Sθ 1 (500) Sθ 1 (1000) Sθ1 (2500) Sθ1 (5000) Sθ1 (10000)
Note: The value in Parenthesis after Sθ1 depicts the value of Ep /Gs value of the soil.
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.
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
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
G s Df
D2f
Cθ = r40 ρG Cθ2 + Sθ2 + 2 Sx2 (2.5.214)
G r0 3r0
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 (2.5.218) 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)
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.
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
L L1
Kθ = Ep Ip φi (z) dz + GSθ 1 [φi (z)]2 dz
2
(2.5.224)
0 0
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
z − 2βz X Y 2βz 2βz
+ Gr20 Sθ 1 e L + cos + η sin dz (2.5.229)
L 2 2 L L
0
√
r30 ρGSθ2 L −2β 3Y η
Cx = X[1 − e (1 + β)] + + (2.5.231)
4β 2 4 2
L
Ep Ip β 2 − 2βz 2βz 2βz
Kθ = e L X − 2η cos + Y sin dz
L2 L L
0
L 2
z − 2βz X Y 2βz 2βz
+ Gr20 Sθ 1 e L + cos + η sin dz (2.5.232)
L 2 2 L L
0
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
given by
(M0 /Kθ )
θ (t) = (2.5.236)
(1 − r2 )2 + (2Dr)2
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)
where,
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
1
C2 V1 (pL) + C3 V2 (pL) = √ (2.5.250)
pL 2
⎧ ⎫
1 V2 (pL) −V3 (pL) ⎨
1 ⎬
C2 1
= (2.5.253)
C3 −V1 (pL) V2 (pL) ⎩ √ ⎭
pL 2
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)
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
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
(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.
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
2
K 4GSθ1 L ψ I1 + I2
Considering ωc2 = , we have − ωf2 = 0 (2.5.269)
Mx Mp I1 + 4λ2 I2
Cθ = r30 ρGSθ2 LI2 (2.5.270)
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
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.
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
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
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
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.
On study of the drawing see if the following check list is satisfied as a minimum
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.
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).
Eccentricity in double
Sl. No. Operating speed amplitude(inch)
1 750 0.014–0.032
2 1500 0.008
3 3000 0.002
2
Fdyn = meωm
where, m = mass of the rotating shaft; e = eccentricity developed in the shaft, and
ωm = operating speed of the machine.
12,000
e(mil) = α ≤ 1.0(mil)
r.p.m.
Peak to peak
displacement
Sl. No. Motor type Speed (RPM) amplitude (inch)
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.
• The rigid type block foundation should be so proportioned that it should have
following mass ratio with respect to the machine
• 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:
• 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.
1 1
Ixx = LB3 ; Iyy = BL3 , and Izz = Ixx + Iyy
12 12
Ixx = yi2 ; Iyy = x2i , and Izz = x2i + yi2 e
i i i
lx
ly lz
Jx = m/12(ly2 + lz2 )
Jy = m/12(lx2 + lz2 )
Jz = m/12(lx2 + ly2 )
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.
2.6.6 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
3600
Example 2.6.1
Design the gas turbine foundation shown in Fig. 2.6.1.
Design data
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
16 −185 Anchor Bolt for generator
17 −185 Anchor Bolt for generator
18 −185 Anchor Bolt for generator
Total −4760
Calculate the natural frequency and amplitude based on Figure 2.6.1 and
using
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
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 .
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
(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
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
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.)
Considering φ11 = 1.00 and solving the above homogenous equation we have,
φ12 = 0.186379442
φ11 1.00
Thus, =
φ12 0.186379442
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
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
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
[C] = α[M] + β [K] or [φ]T [C] [φ] = α [φ]T [M] [φ] + β [φ]T [K] [φ]
14222 −46071
→ [C] = .
−46071 318233
T 21.707808 4.045889184 88 p
Here [φ] {P} = sin 236t →
14.357356 −6.11756837 270 m
3.0
= sin 236t
−0.388
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
236 22.97
where r = √ = 6.63 and ζ = √ = 0.322
1265 2 1265
Y 21.707808 14.357356 5.4204
= × 10−8
4.045889184 −6.11756837 −0.7106
107.462
= × 10−8 sin 236t
26.27
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.
0.00001
0.000005
Amplitude
-0.00001
-0.000015
Time steps
Figure 2.6.2
Self weight of the pedestal = 0.8 × 5.0 × 1.6 × 25 = 160 kN; Weight from
machine = 3 × 200 + 3 × 350 = 1650 kN
ωm 236
And, r= = = 0.07.
ωn 3149
P0 /k 51 sin 236t
y= = = 2.80 × 10−8 m < 0.2 mm.
(1 − r )
2 1.83 × 109 (1 − 0.072 )
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
Thus considering the embedment factor the stiffness and damping value gets
modified to:
Kze 1230836
ωz = = = 29 rad/sec (277 r.p.m);
m 1476.4
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
1348214 − 1476.4λ −3397499
→ =0
−3397499 28075055 − 18588.4λ
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
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
Comparison of amplitude
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.
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
Figure 2.6.4
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.
21 Hammer foundation having eccentric anvil is though uncommon but surely not rare. We will deal with
this particular case separately later.
Hammer/Tup
RCC Fdn.
Anvil
Elastic Pad
F.G.L.
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 .
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
• 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,
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
Substituting the numerical data mentioned in the problem, one can have
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
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
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
√
where, ωd = ωn (1 − D2 ) and D = c/cc and cc = 2 km.
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
x = C2 e−Dωn t sin ωd t
v0
x= e−Dωn t sin ωd t
ωn (1 − D2 )
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
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
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.
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
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
m3
x3
k3 c3
m2
x2
k2 c2
x1
m1
k1 c1
N
N
{{ξ̈i } + ωi2 {ξi }} = 0 where {X} = [φ]{ξi } (2.7.6)
i=1 i=1
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
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
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
√ √
1/√2 1/ √2
[ϕ]n =
1/ 2 −1/ 2
The above when plotted at a time step of 0.05 sec gives plots as shown in
Figure 2.7.8.
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.
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
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
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
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
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
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
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.
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
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.
Table 2.8.1
1 30 3–4
2 20 2
3 Upto 10 1
20(1 + k)
Amin = v · Wh
σs
Ei
H= and V = 2gH
Wh
(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
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 .
−(ωa2 − ωn2
2 )(ω2 − ω2 )
a n1 sin ωn1 t sin ωn2 t
x1 = V −
ωa2 (ωn1
2 − ω2 )f
n2 n2
ωn1 ωn2
Wa + Wfr V
where δst = and δdyn =
Ka ωna
Ka δtot
σp =
Aa
Wa + Wf + Wfr + Kz ψ
σs =
Af
V̂ 1+k
in which, ψ = , where V̂ = V.
2π × fz 1+
Wf
Wa
• 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.
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.
• 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.
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.
2000
1400
1290
2150
(1 + k) 1.65
V= v = 156.25
5.94 = 1.112 m/sec
1+ Wa 1+ 20
Wh
Stiffness properties
For Soil
25
Equivalent Radius r0 = = 2.82 m
π
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
0.425
D= = 0.4832
Bz
m2
x2
k2
c2
m1
x1
k1 c1
k1 + k2 −k2 x1
+ =0
−k2 k2 x2
208 0 ẍ1 850.6 −625 x1
➔ + × 105 = 0
0 16 ẍ2 −625 625 x2
0.0667 0.01898
Thus the complete normalized eigen vector is [ϕ]N = .
0.0684 −0.240
850.6 −625 64952 1474
× × 105 =
−625 625 1474 6316
Calculation of amplitude
N
[ϕ]T [M]{V0 }
{X} = [ϕ]e−Di ωnit {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
The above when plotted at a time step of 0.0005 seconds shows displacement
plots as depicted in Figure 2.8.4.
0.4
0.3 Displacement amplitude
0.2 of foundation (mm)
0.1 Displacement amplitude
0 of anvil (mm)
-0.1 1 14 27 40 53 66 79 92 105 118 131 144
-0.2
-0.3 Time steps
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.
0.15
Amplitude(mm)
Amplitude of foundation
0.1 based on Newmark Method
0.05 Displacement amplitude
of foundation(mm)
0
1 15 29 43 57 71 85 99 113 127 141
-0.05
Time steps
Sl. Time
No. step x1 (disp) x1 (vel) x1 (acc) x2 (disp) x2 (vel) x2 (acc)
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
• 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.
ma 156
α= = = 0.076
mf + mfr 1958 + 85
−(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
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
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
Hammer(Wh)
e Anvil(Wa)
Foundation(Wf)
m2
z2
k2 c2
Cx
Cθ kθ m1 z1
kx
c1
k1
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)
Here,
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.
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
z2(t)
m
k c
z1(t)
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.
Zr (t) z2 (t)
⇒1 and 1
z1 (t) z1 (t)
Again, let
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
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
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.
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
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)
footing A0 sin t
t
soil
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.
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)
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.
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)
Phase angle,
90
A2/A1
0
1 0 0.5 1 1.5 2 2.5 3
Frequency ratio, r
0 6
1 2 4
r
√2
0.5 1
0
0 2 4 6
r
φ1 z1
( φ2-φ1)
cωz1 F
kz1 z2
cωz1
φ2
kz2
Reference
mω2z2 ωt
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.
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.
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
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
Example 2.11.1
Solution:
Solution:
Solution:
a F = 10 Hz → ω = 2π × 10 = 62.83 rad/sec
F0 /k
A= 2 sin (ωt − φ) = A0 sin (ωt − φ) (2.12.2)
(1 − r2 )2 + Ceq ωk + (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
A0
r
r=1
Figure 2.12.2
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
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
mω2x
FTr
F
cωx
φ
x kx
Figure 2.13.2 Vector diagram for the system shown in Figure 2.13.1.
F0
k F0
x= sin (ωt − φ) = κ sin (ωt − φ) (2.13.2)
(1 − r2 )2 + (2Dr)2 k
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
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)
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.
(a) (b)
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).
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.
2.14.1 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
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
Condenser
(spring mounted)
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.
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.
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
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
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).
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:
m1/2 m1/2
m2, k2
k1/2 k1/2
m2 z2
k2
m1
z1
k1 Fixed at base
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
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 .
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)
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).
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
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
The coupled natural frequency of the system can be obtained from the equations
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)
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).
• 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.
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
(P + Q + 2N) + Wf
δs = (2.14.19)
L f B f cu
fv = 30/ δv cpm. (2.14.20)
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
δst
➔ x̄max = ∇ 2 . (2.14.25)
(1 − r2 )2 + π (r)2
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
Major has suggested that the logarithmic decrement (∇) be taken for concrete as
0.4 when the maximum amplitude becomes, x̄max = 7.85 δst .
δst
x̄max = ∇ 2 (2.14.27)
(1 − r2 )2 + π (r)2
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)
1 0.2 ≥3000
2 0.16 1500
3 0.1 750
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
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.
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.
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.
1 Frequency in vertical direction
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
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.
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)
dt ∂ q̇i ∂qi ∂qi
i=1
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
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
∂U ∂U ∂U ∂U
= Kx U; = Kθ θ ; = Kh x and = Kh e2 φ + Iφ φ
∂u ∂θ ∂x ∂φ
n
d ∂T ∂T ∂
d(T + U) = − + dqi = 0
dt ∂ q̇i ∂qi ∂qi
i=1
⎡ ⎤⎧ ⎫
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
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.
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,
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
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).
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
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.
Node (Typ.)
Actual Frame
Idealized Model
Beam Node
Rigid Link
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
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.
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.
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
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
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 .
• 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.
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.
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.
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.
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
• 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.
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:
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 .
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.
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.
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:
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.
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.
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.
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
Mechanical Springs
Frame supporting
Top Deck
Bottom Raft
P0
Kz sin ωm t
δz = (2.18.4)
(1 − r2 )2 + (2Dz r)2
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
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
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
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.
Freq. ratio Damping ratio 0.0 0.05 0.10 0.15 0.20 0.25 0.3
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.
• 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.
Top deck
Spring Elements
Supporting frame
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.
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.
• 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
• 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.
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
= =
Figure 2.18.5 Plan view of top deck with location of equipment load.
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
• 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
4590 4590
A
1 2
1 2 3
Figure 2.18.7
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
48 KN/m
24 KN/m
24 KN/m
32 KN/m
Figure 2.18.8 Frame with uniformly distributed load from top deck slab.
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
2995
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 )
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.
Ly = 4500
Ly = 4500
Lx = 2600
87.4 194
3006 1584
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.
30.22
5.64 10.7
140.83
1494
3006
Figure 2.18.14 Concentrated load transferred to edge beams from slab panel-2.
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
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
Calculation of load transverse frame 2
Self weight of beam = 0.6×0.9×25 = 13.5 kN/m; UDL from slab = 48 kN/m
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.
1584
1585
4590
2730
1 2 3
62 kN/m (q)
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
1584
2730
2
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
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
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
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.
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
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
L3 (1 + 2ψ) 3L
δv = +
96EI b (2 + ψ) 8GAb
= 3.86014 × 10−8 m
1
➔ k2 = = 2.6 × 107 kN/m
δv
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
mb = Mass of cross girder = = 16.29 kN · sec2 /m
9.81
mL = Mass transferred from long girder
(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
0.1215 −0.0754
The normalized eigen vector is given by, [ϕ]n =
0.1748 0.27985
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
ωm 5100
Here r = = 0.575 for mode 1 and r = = 0.28
ωn 18115
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
k + k2 −k2 7.93 −2.6
[K] = 1 = × 107 kN/m
−k2 k2 −2.6 2.6
m1 = mL + 0.255mb + 0.35mc
10.7
m2 = + 0.45 × 9.93 = 5.56 kN · sec2 /m
9.81
30 0
Thus [M] =
0 5.56
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
ωm
Here r = = 0.45 for mode 1 and r = 0.22
ωn
Frame m0 mb mc ml mi δhi K hi
h3 (2 + 3ψ)
Here mi = m0i + mbi + 0.3mci + mLi and δhi =
12EIc (1 + 6ψ)
Frame mi Khi di midi Khi dI xgi (Xg − di ) Xhi mi xgi2 Khi Xhi2
635.476 14754098
Xg = = 4.40 m; Xh = = 4.59 m
144.416 3214401
ωm
Here r = = 3.609 for mode 1 and r = 3.27
ωn
Displacement of frame 2
Displacement of frame 3
x1 = 2.5359 × 10−5 − 4.40 × 2.1987 × 10−5 = −7.13828 × 10−5
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
= 4.855 × 10−6 m
1 5100
M.F = , here r = = 0.69 and = 0.4
(1 − r2 )2 + ( ∇π )2 (r)2 7352
4
δi = 1.9375 × 10−5 m
i=1
4
For calculation of i=1 δi refer to previous calculation by Rausch’s method
30 30
and fv = √ = √ = 6815 rpm
δv 1.9375 × 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 ) = 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
4
δi = 1.1234 × 10−5 m
i=1
4
For calculation of i=1 δi refer to previous calculation by Rausch’s method
30 30
and fv = √ = √ = 8950 rpm
δv 1.1234 × 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 ) = 8.30 kN
Cb L3 2ψ + 1 3 L h Cb + C c
Thus δ1 = + C +
96EIb ψ +2 5 EAb b EAc 2
1 5100
M.F = , r= = 0.69 and = 0.4
[(1 − r2 )2 + ( ∇π )2 (r)2 ] 7352
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
Pv 2476.16
δs = = = 5.36046 × 10−3 m
Kv 461930
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
√
➔ (fn )1 = 30 2472.4 − 238 = 1418 rpm(148 rad/sec);
√
(fn )2 = 30 2472.4 + 238 = 1562 rpm(164 rpm/sec).
4590 4590
58 kN 80 kN 100 kN
80 × 4.59 + 58 × 4.59 × 2
x̄ = = 3.78 m
100 + 58 + 80
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
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
Ci 79.33
Knowing δhi = we have, δh2 = = 7.4277 × 10−5
Khi 1.0717 × 106
Ci 58.33
Knowing δhi = we have, δh3 = = 5.44306 × 10−5
Khi 1.0717 × 106
As the foundation is under tuned thus maximum amplitude at transient
condition is given by
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.
⎡ ⎤⎧ ⎫ ⎧ ⎫
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.
0.36 × 3.6 × 6 × 25
mc = = 19.816
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
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
L3 (1 + 2ψ) 3L
δv = +
96EIb (2 + ψ) 8GAb
= 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
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
78 For further details refer to the section of Design and Analysis of Block Foundation.
3
pi sin ωm t
ξi =
i=1 (1 − r2i )2 + (2Di ri )2
{Z} = [ϕ]{ξ }
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
Case 2
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
Analysis and design of machine foundations 373
The results are compared hereafter by bar chart shown in Figure 2.18.21.
0.7
0.6
Amplitude (mm)
0.5
0.2
0.1
0
Operating Ist Transient
Operating Case
0.5
Vertical amplitude
0.4
Transverse girder
0.3
0.2
0.1
0
Operating Ist Transient
Operating Case
0.7
Amplitude (mm)
0.6
0.5
0.2
0.1
0
Operating Ist Transient
Operating Case
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.
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
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
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
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 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
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
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
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
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
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
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
yinet = yi + Ui + Xhi ϕ + hθ
The results are shown based on bar chart as given in Figure 2.18.23.
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
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.
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.
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
Time steps in seconds
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.
0.001
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
Time steps in seconds
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.
0.0002
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
Time steps in seconds
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.
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.
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.