SW1451 c001 PDF

Chapter 1
Dynamic soil structure interaction
1.1 INTRODUCTION
This chapter deals with some of the basic concepts of dynamic soil-structure interaction
analysis. At the advent of this chapter we expect you to have some background on
• Static soil structure interaction

• Theory of Vibration/structural dynamics
• Basic theory of soil dynamics
∗∗
Based on the above , we build herein the basic concepts of dynamic soil structure
interaction, which is slowly and surely gaining its importance in analytical procedure
for important structures.
1.1.1 The marriage of soil and structure

As was stated earlier in Chapter 4 (Vol. 1) even twenty years ago struc-
tures and foundations were dealt in complete isolation where the structural and
geo-technical/foundation engineers hardly interacted1 .
While the structural engineer was only bothered about the structural configuration
of the system in hand he hardly cared to know anything more about soil other than the
allowable bearing capacity and its generic nature, provided of course the foundation
design is within his scope of work. On the other hand the geotechnical engineer only
remained focused on the inherent soil characteristics like (c, φ, Nc , Nq , Nγ , eo , Cc, G
etc.) and recommending the type of foundation (like isolated footing, raft, pile etc.)
or at best sizing and designing the same.
The crux of this scenario was that nobody got the overall picture, while in reality
under static or dynamic loading the foundation and the structure do behave in tandem.
∗∗ For theoretical background on these topics please consider Volume 1.

1 Even today there are companies which has divisions like structural and civil engineering!! Where the
responsibility of the structural division is to design the superstructure considering it as fixed base frame,
furnish the results (Axial load, Moments and Shear) and the column layout drawing to the civil division
who releases the foundation drawing based on this input data.
© 2009 Taylor & Francis Group, London, UK

2 Dynamics of Structure and Foundation: 2. Applications
In chapter 4 (Vol. 1), in the problem Example 1.3.1, we have shown how the
soil stiffness can affect the bending moment and shear forces of a bridge girder and
ignoring the same how we can arrive at a result which can be in significant variation
to the reality.
Drawing a similar analogy one can infer that ignoring the soil stiffness in the overall
response (and treating it as a fixed base problem) the dynamic response of structure
(the natural frequencies, amplitude etc.) can be in significant variation to the reality
in certain cases.
This aspect came to the attention of engineers while designing the reactor building
of nuclear power plant for earthquake. Considering its huge mass and stiffness, the
fundamental time period for the fixed base structure came around 0.15 sec while
considering the soil effect the time period increased to 0.5 second giving a completely
different response than the fixed base case.
With the above understanding – that underlying soil significantly affects the response
of a structure, research was focused on this topic way back in 1970, and under the
pioneering effort of academicians and engineers, the two diverging domain of technol-
ogy was brought under a nuptial bond of “Dynamic soil structure interaction”, where
soil and structure where married off to a unified integrated domain. To our knowl-
edge the first significant structure where the dynamic effect of soil was considered in
the analysis in Industry in India was the 500 MW turbine foundations for Singrauli
where the underlying soil was modeled as a frequency independent linear spring and
the whole system was analyzed in SAP IV (Ghosh et al. 1984).
1.1.2 What does the interaction mean?

We have seen earlier that considering the soil as a deformable elastic medium the
stiffness of soil gets coupled to the stiffness of the structure and changes it elastic
property. Based on this the characteristic response of the system also gets modified.
This we can consider as the local effect of soil.
On the other hand consider a case of a structure resting on a deep layer of soft
soil underlain by rock. It will be observed that its response is completely differ-
ent than the same system when it is located on soft soil which is of much shallow
depth or resting directly on rock2 . Moreover the nature of foundation, (isolated
pad, raft, pile), if the foundation is resting or embedded in soil, layering of soil,
type of structure etc. has profound influence on the over all dynamic response of the
system.
We had shown for static soil-structure interaction (Chapter 4 (Vol. 1)) case that the
soil can be modeled as equivalent springs or as finite elements and are coupled with the
superstructure.
Thus for a simple beam resting on an elastic support can be modeled as shown
in Figure 1.1.1 and an equivalent mathematical model for the same is shown in
Figure 1.1.2.
Based on matrix analysis of structure the element stiffness for this element may be
written as
2 The reason for these effects we will discuss subsequently.

Dynamic soil structure interaction 3
Node iii Node j
Soil Spring Ki Soil Spring Kj
Figure 1.1.1 Equivalent beam element connected to soil springs.
2 4
1 1 3 2
Figure 1.1.2 Mathematical model of the equivalent beam element.
⎡ ⎤
12 6L 0 −12 6L 0
⎢ ⎥
⎢ 6L 4L2 0 −6L 2L2 0 ⎥
⎢ ⎥
⎢ ⎥
⎢ IxL2 −IxL 2 ⎥
⎢ ⎥
EIz ⎢ 0 0
2Iz(1 + ν)
0 0
2Iz(1 + ν)0 ⎥
[Kbeam ] = 3 ⎢ ⎥ (1.1.1)
L ⎢ ⎢−12 6L 0 12 6L 0 ⎥
⎥
⎢ ⎥
⎢ 6L 2L2 0 6L 4L2 0 ⎥
⎢ ⎥
⎢ ⎥
⎣ −IxL2 IxL 2 ⎦
0 0 0 0
2Iz(1 + ν) 2Iz(1 + ν)
and the displacement vector is given by
{δ} = <δ1 θ1 θ2 δ2 θ3 θ4>T (1.1.2)
When the soil springs are added to the nodes, the overall stiffness becomes
⎡ ⎤
L3 Kii
⎢ 12 + 6L 0 −12 6L 0 ⎥
⎢ EIz ⎥
⎢ ⎥
⎢ 6L 4L2 0 −6L 2L2 0 ⎥
⎢ ⎥
⎢ ⎥
⎢ ⎥
⎢ IxL2 −IxL2 ⎥
⎢ 0 0 0 0 ⎥
EIz ⎢ 2Iz(1 + ν) 2Iz(1 + ν) ⎥

[Kbeam ]= 3 ⎢ ⎥
L ⎢ ⎢ L3 Kjj
⎥
⎥
⎢ −12 6L 0 12 + 6L 0 ⎥
⎢ EIz ⎥
⎢ ⎥
⎢ ⎥
⎢ 6L 2L2 0 6L 4L2 0 ⎥
⎢ ⎥
⎢ ⎥
⎣ −IxL2 IxL 2 ⎦
0 0 0 0
2Iz(1 + ν) 2Iz(1 + ν)
(1.1.3)


where, [Kbeam ] = combined stiffness matrix for the beam and the spring; Kii = Kjj =
spring values of soil at node i and node j of the beam respectively.
The above is a very convenient way of representing the elastic interaction behavior
of the underlying soil and can be very easily adapted in a commercially available finite
element or structural analysis package.
1.1.3 It is an expensive analysis do we need to do it?

This is a common query comes to the mind of an engineer before starting of an analysis.
Based on this fact an engineer do become apprehensive if his/her analysis would suffer
from a cost over run or whether he/she will be able to finish the design within the
allocated time frame.
If he is convinced that soil structure interaction do takes place and the structure is
a crucial one3 our recommendation would be ‘its worth the effort rather than to be
sorry later’. The additional engineering cost incurred is trivial compared to the risk
and cost involved in case of a damage under an earthquake or a machine induced load.
Now the first question is for what soil condition does dynamic soil structure
interaction takes place?
Veletsos and Meek (1974) suggest that chances of dynamic soil structure interaction
can be significant for the expression
Vs
≤ 20 (1.1.4)
fh
where Vs = shear wave velocity of the soil; f = fundamental frequency of the fixed
base structure; h = height of the structure.
Let us now examine what does Equation (1.1.4) signifies?
Knowing the time period T = 1/f , the above expression can be rewritten as
Vs T
≤ 20 (1.1.5)
h
For a normal framed building considering the fixed base time period as (0.1n), where
n is the number of stories and thus, we have
Vs n
≤ 200 (1.1.6)
h
For a normal building the average ratio of h/n (height : storey ratio) is about 3 to
3.3 meter. Thus considering h/n = 3, we have
Vs ≤ 600 m/sec. (1.1.7)
3 Like Power House, Turbine foundations, Nuclear reactor Building, Main process piper rack, distillation
columns, bridges, high rise building catering to large number of people etc.

From which we conclude that for ordinary framed structure, when shear wave
velocity is less or equal to 600 meter/sec we can expect dynamic soil structure
interaction between the frame and the soil.
Incidentally, Vs = 600 m/sec is the shear wave velocity which is associated with rock.
Thus it can be concluded that for all other type of soil, framed structures will behave
differently than a fixed base problem-unless and until it rests on rock. For Cantilever
structures like tall vessels, chimneys etc of uniform cross section fundamental time
period T is given by

mh4
T = 1.779 (1.1.8)
EI
where, m = mass per unit length of the system; h = height of the structure; EI =
flexural stiffness of the system.
Substituting the above value in Equation (1.1.5) we have

mh4
Vs T Vs 1.779 EI 11.24 EI
≤ 20; or ≤ 20; or, Vs ≤ (1.1.9)
h h h m
Considering, I = Ar2 and m = ρ · A, where A = area of cross section; r = radius of

gyration; ρ = Mass density of the material, we have

11.24r E
Vs ≤ (1.1.10)
h ρ
Shear Wave Velocity for Soil-Structure interaction for

Chimneys
1400.00
1200.00
1000.00
velocity(m/sec)
Shear Wave velocity

Shear Wave
800.00 steel chimney

600.00 Shear Wave velocity
400.00 concrete chimney
200.00
0.00
0
0
10
12
15
17
20
22
25
27
30
Slenderness Ratio
Figure 1.1.3 Chart to assess soil-structure interaction for steel and concrete chimney.

For steel structure the above can be taken as, Vs ≤ 57580/λ where λ = h/r, the
slenderness ratio of the structure.
For concrete structure we have
123970
Vs ≤ (1.1.11)
λ
Based on the above expressions one can very easily infer if soil structure interaction
is significant or not.
The chart in Figure 1.1.3 shows limiting shear wave velocity below which soil-
structure interaction could be significant for a steel and concrete chimney.
1.1.4 Different soil models and their coupling

to superstructure
The various types of soil model that are used for comprehensive dynamic analysis are
as follows:
1 Equivalent soil springs connected to foundations modeled as beams, plates, shell

etc.,
2 Finite element models (mostly used in 2D problems),
3 Mixed Finite element and Boundary element a concept which is slowly gaining
popularity.
Of all the options, spring elements connected to superstructure still remain the most
popular model in design practices due to its simplicity and economy in terms of analysis
especially when the superstructure is modeled in 3-dimensions.
It is only in exceptional or very important cases that the Finite elements and Bound-
ary elements are put in to use and that too is mostly restricted to 2 dimensional cases.
1.2 MATHEMATICAL MODELING OF SOIL & STRUCTURE
We present hereafter some techniques that are commonly adopted for coupling the
soil to a structural system.
1.2.1 Lagrangian formulation for 2D frames or stick-models

This formulation is one of the most powerful tool to couple the stiffness of soil to the
superstructure-specially when one is using a stick model or a 2D model.
For the frame shown hereafter we formulate the coupled stiffness and mass matrix
for the soil structure system which can be effectively used for dynamic analysis.
In the system shown in Figure 1.2.1, mf , Jθ = mass and mass moment of inertia
of the foundation; m1 , J1 = mass and mass moment of inertia of the 1st story; m2 ,
J2 = mass and mass moment of inertia of the top story; Kx , Kθ = translational and
rotational stiffness of the soil; and k1 , k2 = stiffness of the columns.

y2
m2 J2
k2
y1 h2
m1, J 1
k1 h1
Kx K mf , J
Figure 1.2.1 2D Mathematical model for soil structure interaction.
The equation for kinetic energy of the system may be written as
1 1 1 1
T= m u̇2 + Jθ θ̇ 2 + m1 (u̇ + h1 θ̇ + ẏ1 )2 + J1 θ̇ 2
2 f 2 2 2
(1.2.1)
1 1
+ m2 (u̇ + (h1 + h2 )θ̇ + ẏ2 )2 + J2 θ̇ 2
2 2
1 1 1 1
U= Kx u2 + Kθ θ 2 + k1 y12 + k2 (y2 − y1 )2 (1.2.2)
2 2 2 2

d ∂T ∂U
Considering the expression4 , dt ∂ q̇i + ∂qi = 0, we have the free vibration
equation as
⎡ ⎤⎧ ⎫
mf + m1 + m2 m1 h 1 + m 2 H m1 m2 ⎪ ü ⎪
⎪
⎪ ⎪ ⎪
⎢ J + m1 h21 + m2 H 2 ⎥ ⎪
m2 H ⎥ θ̈ ⎪
⎨ ⎬
⎢ m1 h 1 m1 h1
⎢ ⎥
⎢ 0 ⎥
⎣ m1 m1 h1 m1 ⎦⎪⎪
⎪
ÿ1 ⎪
⎪
⎪
⎪
⎩ ⎪
m2 m2 H 0 m2 ÿ2 ⎭
⎡ ⎤⎧ ⎫
Kx 0 0 0 ⎪ u⎪
⎪
⎪ ⎪
⎪
⎢0 0 ⎥⎨θ ⎪
⎥ ⎪ ⎬
⎢ Kθ 0
+⎢
⎢0
⎥
⎥ =0 (1.2.3)
⎣ 0 k1 + k2 −k2 ⎦ ⎪
⎪y1 ⎪
⎪
⎪
⎪ ⎪
⎪
0 0 −k2 k2 ⎩y2 ⎭
4 Refer Chapter 2 (Vol. 2) for further application of this formulation where we have derived a 2D soil-
structure interaction model for a Turbine framed foundation.

Figure 1.2.2 Typical finite element mesh with soil springs, for a flexible raft.
where

J = Jθ + J1 + J2 sum of all mass moment of inertia;
H = h1 + h2 = the total height of the structure.
Above formulation can very well be used in cases the foundation is significantly rigid
and can be modeled as rigid lumped mass having negligible internal deformation5 .
However for cases where the foundation is more flexible one usually resorts to finite
element modeling of the base raft which is connected to the soil springs as shown in
Figure 1.2.2.
For the problem as shown above irrespective of the raft being modeled as a beam
or a plate the soil stiffness is directly added to the diagonal element Kii of the global
stiffness matrix to arrive at the over all stiffness matrix of the system.
Before we proceed further we explain the above assembly by a conceptual problem
hereafter.
Example 1.2.1
For the beam as shown in Figure 1.2.3, compute the global stiffness matrix when
supported on a spring at its mid span. Take EI as the flexural stiffness of the
beam. The spring support has stiffness @ K kN/m.
Solution:
For a beam having two degrees of freedom per node as shown in Figure 1.2.4,
the element stiffness matrix is expressed as follows.
5 A classic example is a turbine frame foundation resting on a bottom raft whose thickness is usually
greater than 2.0 meter.

L L
Figure 1.2.3 Spring supported beam.
1 4
3
Figure 1.2.4 Two degrees of freedom of a beam element.
The element matrix for such case is given by
1 2 3 4
⎡ ⎤
12EI 6EI −12EI 6EI
⎢ L3 L2 L3 L2 ⎥
⎢ ⎥
⎢ 6EI 4EI −6EI 2EI ⎥
⎢ ⎥
⎢ ⎥
⎢ L2 L L2 L ⎥
Kij = ⎢ ⎥
⎢ −12EI −6EI 12EI −6EI ⎥
⎢ ⎥
⎢ L3 L2 L3 L2 ⎥
⎢ ⎥
⎣ 6EI 2EI −6EI 4EI ⎦
L2 L L2 L
Assembling the element matrix for the two beams we have
⎡ 12EI 6EI −12EI 6EI ⎤

0 0 0 0
⎢ L3 L2 L3 L2 ⎥
⎢ ⎥
⎢ 6EI −6EI ⎥
⎢ 4EI 2EI ⎥
⎢ 0 0 0 0 ⎥
⎢ L2 L L2 L ⎥
⎢ ⎥
⎢ −12EI −6EI −6EI −12EI ⎥
⎢ 12EI 12EI 6EI 6EI ⎥
⎢ + + 2 0 0 ⎥
⎢ L3 L2 L3 L3 L2 L L3 L2 ⎥
⎢ ⎥
⎢ 6EI −6EI −6EI ⎥
⎢ 2EI 6EI 4EI 4EI 2EI ⎥
⎢ + 2 + 0 0 ⎥
[K]g = ⎢ L2 L L2 L L L L2 L ⎥
⎢ ⎥
⎢ −12EI −6EI ⎥
⎢ 12EI −6EI ⎥
⎢ 0 0 0 0 ⎥
⎢ L3 L2 L3 L2 ⎥
⎢ ⎥
⎢ ⎥
⎢ 6EI 2EI −6EI 4EI ⎥
⎢ 0 0 0 0 ⎥
⎢ L2 L L2 L ⎥
⎢ ⎥
⎢ 0 ⎥
⎢ 0 0 0 0 0 0 0 ⎥
⎣ ⎦
0 0 0 0 0 0 0 0

As Left hand support is fixed hence we have to eliminate row and column 1
and 2.
Similarly, as right hand support is hinged we have to eliminate row and column
5 from the above when we have
⎡ 24EI 6EI ⎤
0
⎢ L3 L2 ⎥
⎢ ⎥
⎢ 8EI 2EI ⎥
[K]g = ⎢
⎢ 0
⎥ with appropriate boundary conditions.
⎢ L L ⎥⎥
⎣ ⎦
6EI 2EI 4EI
L2 L L
To use the spring support, the spring is now directly added to the diagonal
element of the global matrix.
Thus the combined stiffness matrix is given by
⎡ 24EI 6EI ⎤
+ Ks 0
⎢ L3 L2 ⎥
⎢ ⎥
⎢ 8EI 2EI ⎥
[K ]g = ⎢
⎢ 0 ⎥
⎢ L L ⎥⎥
⎣ ⎦
6EI 2EI 4EI
L2 L L
The above is the normal practice adapted in global assemblage of soil spring
in a finite element assembly.
We further elaborate the phenomenon with a suitable practical numerical
example.
Example 1.2.2
Shown in Figure 1.2.5 is a bridge girder across a river is resting at points A and B
on rock abutments at ends, and resting on a pier at center of the girder (point C)
A 5.0 m C 5.0 m B
Water Level
Figure 1.2.5 Bridge girder across abutments.

1 1 2 2 3 3 4 4 5
A C B
Figure 1.2.6 Idealisation of the bridge girder ignoring soil effect.
which is resting on the soil bed of the river. The flexural stiffness of the girder is
EI = 100,000 kN · m2 . Area of girder is 5.0 m2 . The dynamic shear modulus
of soil is G = 2500 kN/m2 . The bridge pier foundation has plan dimension of
6 m × 6 m. Determine the natural frequencies of vibration of the girder consid-
ering with and without soil effect. Unit weight of concrete = 25 kN/m3. Mass
moment of inertia per meter run = 30 kN · sec2 · m.
Solution:
The bridge girder can be mathematically represented by a continuous beam as
shown in Figure 1.2.6. Here node 2 and 4 are at the center of beam.
Thus, for beam element 1, 2, 3, and 4, we have element stiffness matrix as
⎡ ⎤
12 6L −12 6L
⎢
EI ⎢ 6L 4L2 −6L 2L2 ⎥
⎥
[Kij ] = 3 ⎢ ⎥
L ⎣−12 −6L 12 −6L⎦
6L 2L2 −6L 4L2
The unconstrained combined stiffness matrix as
[Kij ]
⎡ ⎤
12 6L −12 6L 0 0 0 0 0 0
⎢ ⎥
⎢ 6L 4L2 −6L 2L2 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ⎥
⎢−12 −6L 24 0 −12 6L 0 0 0 0 ⎥
⎢ ⎥
⎢ 6L 2L 2
0 8L 2
−6L 2L 2
0 0 0 0 ⎥
⎢ ⎥
⎢ ⎥
EI ⎢
⎢ 0 0 −12 −6L 24 0 −12 6L 0 0 ⎥⎥
= 3⎢ ⎥
L ⎢ 0 0 6L 2L2 0 8L2 −6L 2L2 0 0 ⎥
⎢ ⎥
⎢ ⎥
⎢ 0 0 0 0 −12 −6L 24 0 −12 6L ⎥
⎢ ⎥
⎢ ⎥
⎢ 0 0 0 0 6L 2L 2
0 8L 2
−6L 2L2 ⎥
⎢ ⎥
⎢ 0 0 0 0 0 0 −12 −6L 12 −6L ⎥
⎣ ⎦
0 0 0 0 0 0 6L 2L 2
−6L 4L2
Substituting the values we have

[K] =
76800 96000 −76800 96000 0 0 0 0 0 0

96000 160000 −96000 80000 0 0 0 0 0 0
−76800 −96000 153600 0 −76800 96000 0 0 0 0
96000 80000 0 320000 −96000 80000 0 0 0 0
0 0 −76800 −96000 153600 0 −76800 −96000 0 0
0 0 96000 80000 0 320000 −96000 80000 0 0
0 0 0 0 −76800 −96000 153600 0 −76800 96000
0 0 0 0 96000 80000 0 320000 −96000 80000
0 0 0 0 0 0 −76800 −96000 76800 96000
0 0 0 0 0 0 96000 80000 −96000 160000
Now imposing the boundary condition that vertical displacement are zero at
1, 3, 5,6 we have
[K] =
160000 −96000 80000 0 0 0 0

−96000 153600 0 96000 0 0 0
80000 0 320000 80000 0 0 0
0 96000 80000 320000 −96000 80000 0
0 0 0 −96000 153600 0 96000
0 0 0 80000 0 320000 80000
0 0 0 0 96000 80000 160000
Lumped mass at each node is given by → Mii = 25 × 5 × 2.5/9.81 = 31.85

kN · sec2 /m.
Mass moment of inertia at each node is given by → Jii = 30 × 1.25 = 37.5.
Thus combined mass matrix is given by
⎡ ⎤
37.5 0 0 0 0 0 0 0
⎢ 0 31.85 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ 0 0 ⎥
⎢ 0 65 0 0 0 0 ⎥
⎢ ⎥
⎢ 0 0 0 31.85 0 0 0 0 ⎥
[M] = ⎢
⎢ 0
⎥
⎢ 0 0 0 37.5 0 0 0 ⎥⎥
⎢ 0 0 ⎥
⎢ 0 0 0 0 31.85 0 ⎥
⎢ ⎥
⎣ 0 0 0 0 0 0 65 0 ⎦
0 0 0 0 0 0 0 37.5
6 We assume that since the bridge is supported on hard rock at ends, displacement at node 1 and 5
are zero.

1 2 3 4 5
A C B
Kz
Figure 1.2.7 Idealisation of the bridge girder considering soil effect.
Considering the equation

[K] − [M] ω2 = 0 we have
MODE 1 2 3 4 5 6 7
Eigen value 692 1328 2684 4897 7448 7787 11722

Natural 26.30 36.44 51.80 69.97 86.59926 88.24996 108.26855
frequency
(rad/sec)
Considering the effect of soil we can construct the model as in Figure 1.2.7.

4Gr0 LxB
Here Kz = where r0 = , Here L = B = 6.0 m
1−ν π
Here r0 = 3.38 m and for G = 2500 kN/m2 and ν = 0.3 Kz = 48285.71

kN/m.
Now imposing the boundary condition that vertical amplitude at node 1 and
5 are zero (node 3 is not zero) we have
[K] =
160000 −96000 80000 0 0 0 0 0

−96000 153600 0 −76800 96000 0 0 0
80000 0 320000 −96000 80000 0 0 0
0 −76800 −96000 201959.1 0 −76800 −96000 0
0 96000 80000 0 320000 −96000 80000 0
0 0 0 −76800 −96000 153600 0 96000
0 0 0 96000 80000 0 320000 80000
0 0 0 0 0 96000 80000 160000
The Mass matrix remains same as derived earlier.

Performing the eigen value solution we have
Modes 1 2 3 4 5 6 7 8
Eigen-values 75 692 2684 3045 7067 7448 9489 11722

Natural 8.660 26.30 51.80 56.18 84.06 86.30 97.41 108.27
frequency
(rad/sec)

Having established the fact as to how soil affects the dynamic response let us
see further what different type of soil model is possible. For design office practices
spring values considered are usually based on Richart/Wolf’s model which are effec-
tively combined with structure as shown above to find out the overall response of a
system.
The example above, though it has been worked out based on beam the theory, it is
effective for any kind of structural elements like plates, shells, 8-nodded brick element
etc. Thus implementing the above in a general purpose Finite element package is
quite straight forward. For raft modeled as beam with underlain spring, the essence
of arriving at individual springs at each node is same as shown in the case of static
analysis based on influence zone7 .
The only difference being that the nodal influence area is to be converted into an
equivalent circular area to arrive at vertical spring values. The horizontal springs are
based on the full area and are divided equally at the end.
1.2.2 What happens if the raft is f lexible?

Methodology described in previous section is usually adapted when the raft is uncon-
ditionally rigid. However there could be cases where the raft could be perfectly flexible
or intermediate (i.e. somewhere between perfectly rigid and perfectly flexible) when
the calculation of spring values is different than what has been mentioned in the
preceding.
Before we get into this issue the obvious query would be what is the boundary
condition for raft rigidity in terms of dynamic loading?
Unfortunately there is none, and the condition pertaining to static load still applies8 .
Thus as explained in Chapter 4 (Vol. 1), if L is the c/c distance between the columns,
then for
• λL ≤ π4 the raft will behave as rigid raft

• For λL ≥ π the raft will behave as flexible raft
• For all values between π/4 ≤ λL ≤ π, the slab behave in between rigid/flexible

kB
in which λ = 4
4Ec I , k = modulus of sub-grade reaction, (in kN/m3 ); B = width
of raft in meter; Ec = modulus of elasticity of concrete (in kN/m2 ); I = moment of
inertia of the raft (in m4 ).
1.2.2.1 Calculation of spring constant for rigid raft

The rigidity of raft plays a significant role in the soil spring values connected to the
plate elements as mentioned above.
7 Refer Example 4.6.1 in Chapter 4 (Vol. 1) for further details.

8 This is not illogical for dynamic load can be conceived as a system under static equilibrium at a time t.
Thus condition of rigidity as explained in Chapter 4 (Vol. 1) should hold good.

When the raft is rigid the gross spring value is obtained based on the full raft
dimension and then are broken up into discrete values

Ap
kz = Kz (1.2.4)
AG
where, kz = value of discrete spring for the rigid finite element; Kz = value of gross
spring considering the overall dimension of the raft; Ap = area of the finite element
plate, and AG = gross area of the raft.
1.2.2.2 Calculation of spring constant for f lexible raft

When the raft is flexible an equivalent radius within which the load gets dispersed is
first obtained from the formula
1
Ec 1−ν 3
r0 = 0.8ts (1.2.5)
Gs 1 − νc2
The gross spring value is then obtained based on this equation. Finally the discrete
spring for the finite element is obtained as

Ap
kz = Kz (1.2.6)
π r20
where, r0 = equivalent radius within which the load gets dispersed; Ec = dynamic
modulus of the concrete raft; Gs = dynamic shear modulus of the soil; ν = Poisson’s
ratio of soil; νc = Poisson’s ratio of the raft, and ts = thickness of the raft.
A suitable problem cited hereafter elaborates the above more clearly.
Example 1.2.3
A raft of dimension 30 m × 15 m is resting on a soil having dynamic shear
modulus of 35000 kN/m2 and Poisson’s ratio of soil = 0.4. Determine the soil
springs for plate elements of size 2.0 m × 2.0 m for finite element analysis
considering,
• The raft as rigid

• Considering the raft as flexible.
The thickness of the raft is 1.8 m.

Solution:
Considering the raft as rigid:

30 × 15
r0 = = 11.968 meter;
π
4Gr0 4 × 35000 × 11.96
Kz = = = 2790666.67 kN/m
1−ν 0.6
For finite element of size 2 m × 2 m discrete spring value will be

Ap 2×2
kz = Kz ➔ kz = 2790666.67 = 24806 kN/m
AG 30 × 15
Thus spring values at four nodes are 6201 kN/m i.e 1/4th of the above
calculated value. When the raft is considered flexible, we have:
1/3
Ec 1−ν
r0 = 0.8ts
Gs 1 − νc2
Here Ec = 3 × 108 kN/m2 ; νc = 0.25(say),

1/3
3 × 108 1 − 0.4
then r0 = 0.8 × 1.8 = 25.39 m
35000 1 − 0.252
4Gr0 4 × 35000 × 25.39

Thus Kz = = = 5924333.333 kN/m
1−ν 0.6
Thus for finite element of size 2 m × 2 m the discrete spring value is

2×2
kz = 5924333.333 = 11701 kN/m
π × 25.392
Thus spring values at four nodes are 2925 kN/m

It will be observed that the spring values vary considerably for the two different
approach.
1.2.2.3 What sin thou make in treating foundation

& the structure separately?
Difficult to pass a sweeping judgment for depending on the situation, the sin could be
cardinal or even trivial.

Based on a number of analysis carried out it can be stated that treating them
in isolation can result in conservative design9 or dangerously un-conservative, thus
resulting in an unsafe structure which could be a danger to human life and property.
Having made the above statement a number of questions obviously come to
mind10 like
1 How conservative or how susceptible the system can be ignoring the soil effect?
2 Considering soil effect (specially for FEM analysis) makes the analysis more
laborious and time consuming – thus more costly – is it worth?
3 My boss is a traditionalist and under project time pressure – can I convince him
it is worth the effort.
4 Before doing the detailed analysis itself can I come up with a quantitative value
based on which I can assess how far this effect will be (for good or worse) and
thus convince my boss on the value addition to this effort?
5 What is the risk in terms of cost and safety if I do not do this analysis?
The questions are surely pertinent and not always very easy to answer. However with
a little bit of intelligent analysis it is not difficult to come up with a logical conclusion
on this issue.
We try to explain. . .
The obvious answer is ‘it essentially could modify the natural frequency/time period
of the system’11 .
What needs to be evaluated is – what is the effect of this modified time period
on the system compared to, if the soil is ignored (i.e. it is considered a fixed base
problem).
The two classes of problems under which dynamic soil structure interaction plays a
significant role are
• Systems subjected to vibration from machines like block foundations (machine

foundations for pumps, compressors, gas turbines etc), frame foundations (turbine
foundations, compressor foundations, boiler feed pump foundations)
• Structures subjected to earthquake.
For the machine foundation source of disturbance is the machine mounted on the
system the dynamic waves generated are transferred from the machine – via structure
to the surrounding soil-which is an infinite elastic half space.
While for earthquake the source of disturbance is the ground itself where elastic
waves generate within the soil mass due to the tectonic movement/rupture of the rock
mass (geologically known as faults).
It is obvious that soil will affect these two classes of problem in different ways.
For instance a machine supported on a frame- the frame is usually made signifi-
cantly stiff to ensure stress induced in it are not significant and are generally made
9 For big projects which could mean a cost over run.

10 Specially for freshman new to the topic who has got a lead engineer and a departmental HEAD to answer to.
11 We say the word “could” as because the extent of modification will depend upon the shear wave velocity
of the soil. We had shown previously the boundary limits within which it can have a significant effect.

over tuned for medium or low frequency machine when considered as a fixed based
problem.
But in reality considering the soil effect, the foundation may actually be under tuned
or even hover near the resonance zone when the underlying soil participates in the
vibration process. Thus the amplitude of vibration could significantly vary than the
calculated one.
Generically, considering the soil stiffness will make the system more flexible then
a fixed base problem and it can be intuitively deduced that though the stress might
remain within the acceptable level the amplitude of vibration will be more and could
well exceed the acceptable limit which might have secondary damaging effect to the
machine and its appurtenances.
For earthquake the effect is quite different. In this case the structure resting on the
site can be visualized as a body resting on an infinite elastic space (similar to a ship
floating in sea). Due to rupture in the fault as waves dissipate in all direction the soil
mass starts vibrating at its own fundamental frequency known as the free field time
period of the site.
In such case the earthquake acts as an electronic filter and tries to excite the super-
structure resting on it to its own fundamental frequency and suppressing or even
eliminating other modal frequencies12 . Thus if the fixed base frequency of the struc-
ture matches the fundamental frequency of the soil strata on which it is resting, they
are in resonance and catastrophe could well be a reality.
Before dwelling into the mathematical aspect of it we further substantiate the above
statement by some real life facts and observations.
Dowrick (2003) reports that in the Mexico earthquake in 1957 extensive damage
occurred to the buildings that were tall and were found to be resting on alluvium soil
of depth >1000 m. In 1967, the Caracas earthquake showed identical result where the
tall structures underwent extensive damage and those were resting on deep alluvium
soil overlying bedrock. In 1970 earthquake at Gediz in Turkey a part of a factory
was demolished in a town about 140 Km from the epicenter while no other build-
ings in the town underwent any damage! Subsequent investigation revealed that the
fundamental period of the building matched the free field time period of the site. The
Caracas earthquake as cited earlier also showed a distinctive pattern where medium
rise buildings (5–9 storeys) underwent extensive damage where depth to bedrock was
less than 100 m, while buildings over 14 stories were damaged where the depth to
bedrock was greater than 150 meters.
Let us see why such thing happened and how does it substantiate the free field time
period phenomenon as stated earlier.
The free field time period of a site is given by the equation
4H
Tn = (1.2.7)
(2n − 1)Vs
12 It can be visualized as a giant hand trying to shake a small body resting on it. Since the body is much
weaker to the giant it tries to follow the same phase of vibration as the soil medium.

120
Number of Stories
100
80
n for RCC frame
60
n for steel frame
40
20
0
0
15
45
75
05
35
3
5
0.
0.
0.
1.
1.
0.
0.
0.
1.
1.
Depth of soil/Shear wave Velocity
Figure 1.2.8 Limiting value of storeys for frames.
where, T = time period of the free field soil (i.e. without the structure); H = depth of
soil over bedrock13 ; n = number of mode; and Vs = shear wave velocity of the soil.
Thus based on the explanation above it can be argued that if the fixed base frequency
of structure is in the close proximity of the free field time period of the site the structure
may be subjected to significant excitation.
The above statement can be extended to a very interesting hypothesis.
If we equate the free field time period of the site to the fixed base time period of
the structure we can arrive at some limiting design parameters which can result in
significant dynamic amplification and which should be avoided at the very out set of
planning of the structure.
For instance as per IS-1893 RCC moment resisting frames with no infill brick work,
the fundamental time period is given by
T = 0.075h0.75 (1.2.8)
Thus equating it to fundamental free field time period of the site we have
4/3
4H 160H
0.075h0.75 = , which gives h = (1.2.9)
Vs 3Vs
Considering 1 floor is of height 3.3 m, we can further simplify the equation to

4/3
160H
n = 0.303 (1.2.10)
3Vs
13 Here bedrock is perceived as that level where the shear wave velocity of soil is greater or equal to
600 m/sec.

The curves shown in Figure 1.2.8 give limiting stories for RCC and steel frames for
which resonance can occur in a structure during an earthquake as per IS-189314 for
various values of H/Vs .
Let us now probe the problem a bit more based on a suitable numerical problem.
Example 1.2.4
A particular site has been found to consist of 100 m soil overlying bedrock
when the shear wave velocity of the soil is 222.22 m/sec. Find the limiting
number of stories of height 3.3 meter for an RCC frame for which resonance
can occur. What would be resonance story if the depth of the overlying soft soil is
only 30 m.
Solution:
Based on above data H/Vs = 100/222.2 = 0.45 when H = 100 m.
As per the chart as shown above the limiting story for which resonance can
occur is 18.
Thus for a 18 storied building resonance can very well occur and the strategy
would be to build the building at least (±)25% away i.e. either it should be 23
storied or more or 14 storied or less.
30
When the depth of soil is only 30 m, H/Vs = 222.2 = 0.135.
Based on the above chart the limiting story height is roughly 4-storey only.
Thus to avoid resonance the building should be either more than 5-storey or less
than 3-storey.
The above problem well explains the phenomenon as to what happened in the
Mexico and Turkey earthquakes and perhaps challenges the myth quite prevalent in
many design offices – that for one or two storied building earthquake is not important
and can well be ignored.
It is evident from the above problem that the response depends on the depth of soil
on which it is resting and depending on the free field time period the response can
either amplify or attenuate. It can well affect even a one storied building.
The chart in Figure 1.2.9 shows limiting story height of buildings with infill brick
panels and all other type of frames as per IS 1893 for different width of building
varying from 10 meter to 50 m15 .
The above theory is though explained in terms of building, can very well be adapted
for any class of structure for which it is possible to establish the fundamental time
period expression.
14 In this case time period for steel frame is considered as T = 0.085(h)0.75 as per IS-1893.
15 Time period of the fixed base structure considered as T = 0.09h/(d)0.5 as per, Indian Standards Institution
(1984, 2002). “Indian Standard Criteria for Earthquake Resistant Design of Structures”, IS: 1893
(Part 1), ISI, New Delhi, India.

40
35
30 n for d=10m
Number of story
n for d=15m
25 n for d=20m
20 n for d=25m
n for d=30m
15
n for d=35m
10 n for d=40m
5 n for d=45m
n for d=50m
0
0
6
08
15
23
38
45
53
68
75
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
Depth of soil/Shear wave velocity
Figure 1.2.9 Limiting story for building with infill brick panel.
Having assessed the resonance criteria and making sure at planning stage that the
two periods do not match one would still like to quantify the combined time period
of the overall soil structure system and assess whether there is any amplification or
attenuation of the earthquake force.
Before plunging into detailed analysis based on FEM or otherwise it would be useful
to have a rough estimate as to how much the underlying soil affects the overall response.
Veletsos and Meek (1974) has given a very useful expression based on which it is
possible to estimate the modified time period of a structure, and is given by

k̄ Kxh̄2
T̄ = T 1 + 1+ (1.2.11)
Kx Kθ
where T̄ = modified time period of the structure due to the soil stiffness, T = time
2
period of the fixed base structure, k̄ = stiffnessof the fixed base structure @ 4πgTW2 ,
Kx , Kθ = horizontal and rotational spring constant of the soil (IS-1893), h̄ = effective

height or inertial centroid of the system, and, W = total weight of the structure.
Based on the above expression one can immediately arrive at a rough estimate as
to how strong could be soil response at the very outset of a design. We elaborate the
above based on two suitable problems hereafter.
Example 1.2.5
An RCC Chimney 150 meter in height has a uniform cross section area of Ac =
8.5 m2 and moment of inertia I = 92.5m4 . Evaluate the base moment and

shear under earthquake considering the problem as fixed base as well as the
soil effect. The structure is located in zone IV as per IS 1893. The structure
is supported on raft of diameter 18 meter. The soil has a dynamic shear wave
velocity of 120 m/sec and unit weight of 19 kN/m3 . Consider 5% damping
for the analysis.16 The grade of concrete used is M30 having dynamic Econc =
3.12 × 108 kN/m2 .
Solution:
Height of the structure = 150 m; Area of shell = 8.5 m2
Weight of chimney = 150 × 8.5 × 25 = 31875 kN (unit weight of conc. =
25 kN/m3 )
! !
Radius of gyration of the chimney = I/A = 92.5/8.5 = 3.298 m
150
Thus slenderness ratio H/r = = 45.4.
3.298
As per IS 1893 CT = 82.8.
IS 1893 time period of a fixed base chimney is given by, T =

As per
CT WH
.
3.13 Ec Ac
where, W = weight of chimney in N; Ec = Dynamic Young’s modulus of conc.
@ 3 × 108 kN/m2

82.8 31875 × 150

Thus, T = = 1.13 sec
3.13 3.12 × 108 × 8.5
For 5% damping referring to chart in IS-1893 we have Sa/g = 0.10.

Sa
Thus the horizontal seismic coefficient is given by, αh = βIFo .
g
Here β (Soil foundation factor) = 1.0 for chimney resting on raft, I = 1.5
Importance factor, Fo = Zone factor @ 0.25 for zone IV.
This gives
αh = 1.0 × 1.5 × 0.25 × 0.10 = 0.0375
The Bending moment and shear force are given by,

x 1 x 4
2 5x 2 x 2
M = αh W H̄ 0.6 + 0.4 and V = Cv αh W −
H H 3H 3 H
16 In this case it is presumed that reader has some idea of how to use the code IS-1893 or is at least
familiar with it.

Here, Cv = a coefficient depends on the slenderness ratio and as per the present
problem is 1.47 as per IS 1893; H̄ = height of c.g. of the structure above base
@ 75 meter for the problem; x = distance from the top.
Substituting the appropriate values, we have
1
M = 0.0375 × 31875 × 75 [0.6(1.0) 2 + 0.4(1.0)4 ] = 89648 kN/m
V = 1.47 × 0.0375 × 31875 [(5/3) − (2/3)] = 1757 kN.
Considering the soil effect we have the dynamic shear modulus of soil, G = ρvs2
Or G = (19/9.81) × 120 × 120 = 27890 kN/m2 .
8GR
With radius of raft = 9.0 m, Kx = , ν = Poisson’s ratio of the soil
2−ν
considered as 0.35,
8 × 27890 × 9
Kx = = 1217018.2 kN/m
2 − 0.35
8GR3 8 × 27890 × 93
And Kθ = which gives, Kθ = = 83412554 kN/m
3(1 − ν) 3(1 − 0.35)
The fixed base stiffness of chimney is given by
4π 2 W 4 × π 2 × 31875
k̄ = = = 100458 kN/m
gT 2 9.81 × 1.132
Substituting the above numerical values in Veletsos’ equation we have

k̄ Kxh̄2
T̄ = T 1 + 1+
Kx Kθ

100458 1217018.2 × 752
➔ T̄ = 1.13 1 + 1+ = 3.2 sec
1217018.2 83412554
As per IS 1893 for T = 3.2 sec, Sa/g = 0.05, which gives
0.0375
αh = × 0.05 = 0.01875 (By proportion)
0.10
The base moment and shear are given by
89648
M= × 0.01875 = 44824 kN · m;
0.0375

1757
and V = × 0.01875 = 878.5 kN
0.0375
The results are compared hereafter17
Case Moment Shear Remarks
Without soil 89648 1757 Reduction in moment and shear by 34%

With soil 44824 878.5
The problem shows a clear attenuation of the response.
We show another example hereafter.
Example 1.2.6
Shown in Figure 1.2.10 is a horizontal vessel having empty weight of 340 kN
and operating weight of 850 kN is placed on two isolated footing of dimension
8.5 m × 3 m. The center to center distance between the two foundations is 5.5
meter. The center line of vessel is at height (H f ) of 4.5 meters from the bottom of
the foundation. Thickness of the foundation slab is 0.3 meter. The RCC pedestal
is of width 1.0 meter, length 6 meter having height of 3.45 meter. The shear wave
velocity of the soil is 200 m/sec having Poisson’s ratio of 0.3. Allowable bearing
capacity of the foundation is 150 kN/m2 . Calculate the design seismic moment
considering the effect of soil and without it, if the site is in zone III as per IS-1893.
Consider soil density @ 18 kN/m3 and unit weight of concrete as 25 kN/m3 ?
Solution:
Plan are of footing = 8.5 × 3 = 25.5 m2

Af 25.5
Equivalent circular radius = = = 2.849 m
π π

1 1
Moment of inertia of the foundation about X-axis BL3 = 3 × 8.53 =
12 12
153.5313 m4

1 1
Moment of inertia of the foundation about Y-axis LB3 = 8.5 × 33 =
12 12
19.125 m4
17 Without an elaborate analysis it could be an effective calculation to convince the boss that
you can save some money and the worth of a dynamic soil-structure interaction analysis.

yp
Hf
yp
Hp
Ds
Wp
Y
Lp Lf
X
Bf Bf
Ls
Figure 1.2.10 A horizontal vessel.

1 64Ixx 0.25
Equivalent circular radius about X axis = = 3.739183 m
2 π

1 64Iyy 0.25
Equivalent circular radius about Y axis = = 2.221 m
2 π
18
Mass density of soil (ρ) = = 1.835 kN/m3
9.81
Dynamic shear modulus = ρvs2 = 1.835 × 2002 = 73400 kN/m2

32Gr0 (1 − ν)
Lateral spring in X and Y direction = = 1018306 kN/m
7 − 8ν
8Gr3x
Rocking spring about X axis = = 14627886 kN/m
3(1 − ν)
8Gr3y
Rocking spring about Y axis = = 3063462 kN/m
3(1 − ν)
1
Moment of Inertia of the pedestal about X axis = × 1 × 63 = 18 m4
12
1
Moment of Inertia of the pedestal about X axis = × 6 × 13 = 0.5 m4
12
3EIx 3 × 3.2 × 108 × 18

Structural stiffness of pedestal about X axis = = =
L3 3.453
3.95 × 108 kN/m
3EIy 3 × 3.2 × 108 × 0.5

Structural stiffness of pedestal about Y axis = 3
= =
L 3.453
1.10 × 107 kN/m
340
Contributing mass for the vessel empty case = = 17.33 kN-sec2 /m
2 × 9.81
850
Contributing mass for the vessel operating case = =
2 × 9.81
43.323 kN-sec2 /m
Contributing uniformly distribute load for the pedestal = 25.5 × 25/9.81 =

64.98 kN/m
The mathematical model for the pedestal thus constitute of a beam element
(pedestal) having a mass lumped at its tip (mass contribution from the vessel) is
shown in Figure 1.2.11.
The time period of such fixed base model is given by (Paz 1991)

(M + 0.25mb )
T = 2π
K
mb M
Figure 1.2.11 Mathematical model for the pedestal.

and the modified time period considering soil effect is given by

2
k̄ Kxh̄
T̄ = T 1 + 1+
Kx Kθ
The time periods and the corresponding Sa/g values as per IS-1893 for 5%
damping are as show hereafter.
Time period Time period Time period Time period

(vessel empty) (vessel empty) (operating) (operating)
about X about Y about X about
Sl no Case direction direction direction Y direction
1 Without soil 0.0018 0.011 0.0024 0.015

2 With soil effect 0.0579 0.1057 0.0771 0.1408
Corresponding Sa/g value is given by
Sa/g Sa/g Sa/g Sa/g

about about about about
Sl no Case X direction Y direction X direction Y direction
1 Without soil 0.1000 0.120 0.1000 0.130

2 With soil effect 0.2 0.2 0.2 0.2
Base shear as per IS 1893 considering Importance factor as 1.0 for vessel empty
case and 1.25 for vessel in operation case we have
Shear Shear Shear Shear

1 Without soil 21.25 25.5 26.5625 34.53125
2 With soil effect 42.5 42.5 53.125 53.125
The moment at the foundation level is given by
Moment Moment
(vessel (vessel Moment Moment
empty) empty) (operating) (operating)
1 Without soil 100.9375 121.125 126.1719 164.0234

2 With soil effect 201.875 201.875 252.3438 252.3438
This case clearly shows an amplification of force considering the soil effect.

Having established a basis of how to evaluate the coupled soil-structure interaction

under dynamic loading we now extend the above theory to system with multi degree
of freedom where the theory can be very well be adapted as a powerful tool for a
detailed yet economic dynamic analysis.
1.3 A GENERALISED MODEL FOR DYNAMIC SOIL STRUCTURE

INTERACTION
In this section we present a generalised model for dynamic soil-structure interaction.

Though the model is developed based on 3D frames can also be adapted for a three
dimensional Finite Element analysis.
1.3.1 Dynamic response of a structure with multi degree

of freedom considering the underlying soil stiffness
We had shown earlier that for a single degree of freedom system the modified time
period of a structure considering the soil effect is given by

k̄ Kxh̄2
T̄ = T 1 + 1+ (1.3.1)
Kx Kθ
Both ATC (1982) and FEMA has adapted this formula for practical design office usage
(Veletsos & Meek 1974, Jennings & Bielek 1973). The nomenclatures of the formula
are as explained earlier. Now squaring both sides of the above equation we have

k̄ ¯2
kh
T̄ 2 = T 2 1+ + (1.3.2)
Kx Kθ
2π
Considering the expression T = ω we have

4π 2 4π 2 4π 2 m 4π 2 mh̄2 4π 2 4π 2 mω2 ω2 mh̄2
= 2 1+ 2 + or 2 = 2 1 + +
ω̄2 ω T Kx T 2 Kθ ω̄ ω Kx Kθ
(1.3.3)
Simplifying and expanding the above we have
1 1 m mh̄2
= + + which can be further modified to
ω̄2 ω2 Kx Kθ
1 1 1 1
= 2+ 2+ 2 (1.3.4)
ω̄2 ω ωx ωθ
which gives the modified natural frequency relation for a system with single degree of
freedom. This formulation has also been shown in, Kramer, S. (2004).

!
Now considering generically ω = k/m we have
m m m mh̄2 1 1 1 h̄2
= + + , or, = + + (1.3.5)
ke k Kx Kθ ke k Kx Kθ
where ke = equivalent stiffness of the soil structure system having single degree of
freedom.
We shall extend the above basis to multi degree of freedom hereafter (Chowdhury
and Dasgupta 2002).
1.3.2 Extension of the above theory to system

with multi degree of freedom
A 3-D frame shown in Figure 1.3.1, is considered for the presentation of the proposed
method. The frame structure has n degrees-of freedom and subjected to soil reactions
in the form of translational and rotational springs.
For a system having n degrees of freedom the above equation can be written in the
form
" #
[M]n×n [M]n×n [M]n×n [M]n×n h2 n×n
= + + (1.3.6)
[Ke ]n×n [K]n×n Kx Kθ
X
O
Kx
-- mass points Kθ
Figure 1.3.1 A 3-D Frame having multi-degree-of freedom with representative foundation spring.

Here, [Ke ] = equivalent stiffness matrix of the soil structure system of order n, [M] =
a diagonal mass matrix of order n having masses lumped at the element diagonals,
[h̄2 ] = radius vectors of the lumped masses to the center of the foundation springs
of order n, Kx , Kθ = translation and rotation spring stiffness of the total foundation
system represented by a unique value.
Taking out the common factor [M], we have
" 2#
[I] [I] [I] h
= + + (1.3.7)
[Ke ] [K] Kx Kθ
where, [I] = identity matrix of order n having its diagonal element as 1.

or [I][Ke ]−1 = [I][K]−1 + [I/Kx ] + [h2 /Kθ ]
⇒ [Fe] = [F] + [Fx ] + [Fθ ] (1.3.8)
where [F] = Flexibility matrix of the system with suffixes as mentioned earlier for
stiffness matrices.
Once the flexibility matrix of the equivalent soil structure system is known the
stiffness matrix may be obtained from the expression
[Ke ] = [Fe]−1 (1.3.9)
Now knowing the modified stiffness matrix the eigen solution may be done based
on the usual procedure of
[Ke ] [ϕ] = [λe ] [M] [ϕ]. (1.3.10)
1.3.3 Estimation of damping ratio for the soil

structure system
While calculating the damping ratio, the normal process is to guess a damping ratio
for the structure like 2–5%, and consider the same damping ratio for all the mode and
obtain the value of Sa/g value for the particular structure per mode corresponding to
the time period based on the curves given in IS-1893.
The basis of assuming this damping ratio is purely judgmental and is dependent
on either the experience of the engineer, recommendation of codes, or based on field
observations on the performance of similar structure under previous earthquakes.
When the effect of soil is neglected it is possible to obtain the material damping ratio
of the structure depending on what constitute the material like steel, RCC etc.
However when the whole system is resting on soil an analyst is usually faced with
the following stumbling blocks for which clear solution is still eluding us specially for
modal analysis in time domain.

The difficulties encountered can be summarised as follows
• The damping matrix of the coupled soil-structure system becomes non-

proportional for which the damping matrix does not de-couple based on
orthogonal transformation.
• As the damping ratio of the structure and the soil could be widely varying it
becomes difficult to assess a common damping ratio which would affect the soil
as well as structure.
• Even after elaborate FEM modelling of the soil, the damping ratio contribution
per mode still remains guess estimation at the best.
We present hereafter a method by which one can estimate approximately the contri-
bution of combined soil structure system under earthquake for various modes, without
resorting to an elaborate modelling of the soil itself.
We only estimate the contribution of the soil damping to the structural system whose
response we are interested in. The estimation is surely approximate but at least gives
a rational mathematical basis to arrive at some realistic damping value rather than
guessing a damping value at the outset and presuming that it remain same for each
mode, specially for coupled soil structure system where widely varying damping for the
foundation and structure makes it difficult for the analyst to arrive at unified rational
value applicable to the system.
1.3.4 Formulation of damping ratio for single degree

of freedom
Neglecting the higher order, the material damping ratio for a soil structure system
having single degree of freedom is given (Kramer 2004) by
ζ̄ ζ ζx ζθ
= 2+ 2+ 2 (1.3.11)
ω̄ 2 ω ωx ωθ
where, ζ̄ = damping ratio of the equivalent soil structure system; ζ = damping ratio
of the fixed base structure; ζx = horizontal damping ratio of the soil, where ζx = 0.288
√
B x
(7−8ν)mg
and Bx = 32(1−ν)ρ 3 , where m = total mass of the structure and foundation; g =
s rx
acceleration due to gravity; ν = Poisson’s ratio of the soil; ρs = mass density of the
soil; rx = Equivalent circular radius in horizontal mode; ζθ = damping ratio of the
soil in rocking mode ζφx = (1+B 0.15√
) B
and Bθ = 0.375(1−ν)J
ρ r5
θg
; and Jθ = mass moment
θ θ s θ
of inertia of the foundation and the structure.

Converting the damping ratio equation to stiffness-mass basis we have
mζ̄ mζ mζx mh2 ζθ ζ̄ ζ ζx h 2 ζθ

= + + or = + + ;
ke k Kx Kθ ke k Kx Kθ

ζ ζx ζθ
➔ ζ̄ = ke + + (1.3.12)
k Kx Kθ

For very high value of Kx and Kθ ke → k when ζ̄ → ζ .
1.3.5 Extension of the above theory to systems

with multi-degree freedom
On extending the above to multi degree of freedom of order n, we have
[ζ̄ ][M]n×n [ζ ][M]n×n [ζx ][M]n×n [ζθ ][M]n×n [h2 ]n×n

= + +
[Ke ]n×n [K]n×n Kx Kθ
➔ [ζ̄ ] = [Ke ]{[ζ ][F] + [ζx ][Fx ] + [ζθ ][Fθ ]} (1.3.13)
[ζ̄ ] = Damping ratio matrix of the combined soil structure system having n number
of modes.
It is to be noted that [ζ̄ ] is non-proportional and not a diagonal matrix, and based
on the matrix operation as shown above has off-diagonal terms.
A study on the parametric effect shows that [ζ̄ ] becomes nearly a diagonal matrix
(i.e. the off diagonal terms vanishes or approaches zero) when damping ratio of the
structure and the soil foundation system are nearly equal.
However, when the damping ratio are widely varying the off diagonal terms do not
vanish however there magnitudes are relatively smaller than the diagonal terms (ζii )
which has the most dominant effect on the system.
Thus if it is possible to arrive at a foundation layout where the damping ratio of the
structure and foundation are closely spaced considering the diagonal terms as modal
damping ratio per mode is quite correct.
Even when the off diagonal term exists due to widely varying values for practical
design engineering purpose considering the ζii term of damping ratio matrix is realistic
for it gives a reasonably rational basis of estimation of the damping ratio per mode
rather than guessing a value based on gut feeling.
We explain the above theory based on suitable example hereafter
Example 1.3.1
Shown in Figure 1.3.1 is a three storied steel frame subjected to dynamic forces.
The damping ratio for steel is found to vary between 2 to 5%. Determine
• The fixed base natural frequencies of the structure.

• The fixed base eigen-vectors.
• Modified natural frequency with foundation stiffness.
• Modified eigen.
• Take K x = 35000 kN/m and K θ = 50000 kN/m for the soil-foundation.
• Analyse the floor shears for earthquake based on IS-1893 Zone III for
◦ Fixed base.
◦ Considering the soil effect.

G H X3
3000
E F X2
3000
C D X1
3000
A B
Figure 1.3.2
Here,
1 K AC = KDB = 1.5 × 103 kN/m M GH = 200 kN sec2 /m

2 K CE = KDF = 1.0 × 103 kN/m M EF = 400 kN sec2 /m
3 K EG = KFH = 0.75 × 103 kN/m M CD = 400 kN sec2 /m
Solution:
The stiffness and mass matrix is given by
⎡ ⎤ ⎡ ⎤
5000 −2000 0 400
[K] = ⎣−2000 3500 −1500⎦ and [M] = ⎣ 400 ⎦
0 −1500 1500 200
• Based on Figure 1.3.2, we have found earlier that
√ √
ω
√1 = 1.6426 = 1.281 rad/sec; ω2 = 10.00 = 3.162 rad/sec; ω3 =
17.104 = 4.135 rad/sec.

Thus the time periods for the fixed base structure is given by18
T1 = 4.97 sec, T2 = 1.987 sec, T3 = 1.52 sec
• The mode shapes or the eigen-vectors are

⎡ ⎤
1.00 1.0 1.0
[φ] = ⎣2.1715 0.5 −0.9208⎦
2.7816 −1.50 0.719
• Normalised eigen vectors

⎡ ⎤
0.01615 0.03244 0.0344512
[ϕi ] = ⎣0.0350718 0.01622 −0.03172 ⎦
0.04493 −0.02433 0.02477
Calculation for the combined soil-structure system

Here stiffness matrix of the fixed base structure
⎡ ⎤
5000 −2000 0
[K] = ⎣−2000 3500 −1500⎦ which on inversion gives
0 −1500 1500
⎡ ⎤
0.000333 0.000333 0.000333
[F] = ⎣0.000333 0.000833 0.000833⎦
0.000333 0.000833 0.003145
⎡ ⎤
1/35000 0 0
[Fx ] = ⎣ 0 1/35000 0 ⎦
0 0 1/3500
⎡ ⎤
2.85714 0 0
=⎣ 0 2.85714 0 ⎦ × 10−5
0 0 2.85714
⎡ ⎤
9 0 0
[h2 ] = ⎣0 36 0⎦
0 0 81
18 You can check the value by any of the method as explained in Chapter 5 (Vol. 1) for eigen value
analysis.

Thus
⎡ ⎤
9/50000 0 0
[Fθ ] = ⎣ 0 36/50000 0 ⎦
0 0 81/50000
⎡ ⎤
0.00018 0 0
=⎣ 0 0.00072 0 ⎦
0 0 0.00162
As [Fe] = [F] + [Fx ] + [Fθ ] we have

⎡ ⎤
0.000542 0.000333 0.000333
[Fe] = ⎣0.000333 0.001581905 0.000833⎦
0.000333 0.0008333 0.001982
which is combined flexibility matrix of the soil structure system.

Inversion of the above flexibility matrix gives
⎡ ⎤
2195.19 −344.34 −224.4212
[Ke ] = ⎣−344.34 804.662 −306.223 ⎦
−224.42 −306.223 671.0682
The above gives the combined stiffness matrix for structural system consider-
ing the soil compliance19 .
Thus based on the above modified stiffness matrix and mass matrix as
⎡ ⎤
400
[M] = ⎣ 400 ⎦
200
√
√ ω1 = 1.2163 = 1.10286 rad/sec; ω2 =
√ We have, based on eigen solution,
3.9666 = 1.9916 rad/sec; ω3 = 5.8255 = 2.4136 rad/sec.
Thus the time periods for the combined soil-structure system is given by
T1 = 5.697 sec, T2 = 3.154 sec, T3 = 2.603 sec
19 Watch the numbers. . . . . it is symmetric and is completely different than when you add the
springs directly to the diagonal. This matrix has no rigid body mode and can be used directly
for static analysis too.
Moreover if we take Kx and Kθ very high the Ke converges to the fixed base matrix K.

Normalised modified eigen vectors considering soil stiffness is given by

⎡ ⎤
−0.013 0.0479 0.00589
[ϕi ] = ⎣−0.0409 −0.00772 −0.0276⎦
−0.0360 −0.0169 0.05835
Calculation of modal damping

Considering, ζ = 5% for the structure, ζx = 10% for the soil in translation
mode, ζθ = 15% for the soil in rocking mode
We have, [ζ̄ ] = [Ke ]{[ζ ][F] + [ζx ][Fx ] + [ζθ ][Fθ ]}
Substituting the values as mentioned and calculated above we have
⎡ ⎤
0.092 −0.028 −0.020
[ζ̄ ] = ⎣−0.007 0.10908 −0.028⎦
−0.002 −0.0126 0.1115
It will be seen that that the main diagonal terms are dominant and
can be considered as the modal damping ratio contribution for each
mode.
Suppose we had closely spaced damping data like ζ = 5% for the structure;
ζx = 6% for the soil in translation mode; ζθ = 5.5% for the soil in rocking
mode, the modal damping matrix reduces to
⎡ ⎤
0.0525 −0.0015 −0.001016
[ζ̄ ] = ⎣ −0.0004 0.05312 −0.00144 ⎦
−0.00014 −0.00066 0.05315
When the matrix become practically diagonal dominant with off diagonal
terms having very low values.
Thus for the present problem ζ may be considered as ζ1 = 9.2% for
first mode, ζ2 = 10.9% for second mode; and ζ3 = 11.1% for the third
mode.
Calculation of earthquake force fixed base structure
m φ1 mφ1 mφ12 φ2 mφ2 mφ22 φ3 mφ3 mφ32
400 0.01615 6.46 0.104329 0.03244 12.976 0.420941 0.03445 13.7804 0.47475407
400 0.03507 14.028 0.491962 0.01622 6.488 0.105235 −0.01372 −5.488 0.07529536
200 0.04493 8.986 0.403741 −0.02433 −4.866 0.118389 0.02477 4.954 0.12271058
29.474 1.000032 14.598 0.644565 13.2464 0.67276001

Modal mass participation factor
29.474
κ1 = = 29.47306 for the first mode,
1.000032
14.598
κ2 = = 22.64777 for the second mode,
0.644565
13.2464
κ3 = = 19.689 for the third mode.
0.67276
Assuming 5% damping for the structure we have,
Mode Time period (secs) Sa (m/sec 2 ) Remarks
1 4.9 0.4905 Sa value obtained from the chart

given in IS-1893 for 5% damping
2 1.98 0.6867 Do
3 1.52 0.7848 Do
For zone III:

K = 1.0, β = 1.0, I = 1.2F0 = 0.2 as per the code

Thus base shear is given by; V = 3i=1 K · I, β · F0 · κi · Sa mi φi
Substituting data on the above formula we have
Mode Base shear V Remarks
1 102 Fixed base case

2 5.45
3 4.91
Calculation for coupled soil-structure interaction.
m φ1 mφ1 mφ12 φ2 mφ2 mφ22 φ3 mφ3 mφ32
400 −0.013 −5.2 0.0676 0.0479 19.16 0.917764 0.006 2.4 0.0144
400 −0.041 −16.4 0.6724 −0.0077 −3.08 0.023716 −0.0276 −11.04 0.304704
200 −0.036 −7.2 0.2592 −0.0169 −3.38 0.057122 0.0583 11.66 0.679778
−28.8 0.9992 12.7 0.998602 3.02 0.998882
Modal mass participation factor

−28.8
κ1 = = −28.8231 for the first mode,
0.9992
12.7
κ2 = = 12.717 for the second mode, and
0.998602
3.02
κ3 = = 3.0233 for the third mode.
0.9988

Modal damping for each mode, as calculated earlier.
Mode Damping Time (sec) Sa (m/sec 2 ) Remarks
1 9.2% 5.7 0.343 Calculated from curve based on interpo-

lation corresponding to 9.2% damping
2 10.9% 3.2 0.294 Do- with 10.9% damping
3 11.15% 2.6 0.245 Do- with 11.15% damping
Calculation for Base shear
Base shear for the frame with coupled soil-structure interaction is given by
Mode Base shear V Remarks
1 68.4 Couple soil-foundation system

2 11.4
3 0.537
Calculation of storey forces
The storey forces for the two cases are calculated hereafter
Coupled soil structure system Fixed base
Base Base Base Base Base Base

shear shear shear shear shear shear
mh2
Storey m h mh2 3 mode 1 mode 2 mode 3 mode 1 mode 2 mode 3
i=1 mh2
1st 400 3 3600 0.10526 7.20 ×10+00 1.20 ×10+00 5.66 ×10−02 1.08 ×10+01 5.74 ×10+00 5.17 ×10+00
2nd 400 6 14400 0.42105 2.88 ×10+01 4.80 ×10+00 2.26 ×10−01 4.31 ×10+01 2.29 ×10+01 2.07 ×10+01
Top 200 9 16200 0.47368 3.24 ×10+01 5.40 ×10+00 2.55 ×10−01 4.84 ×10+01 2.58 ×10+01 2.33 ×10+01
Comparison of results
Time period
Structure type T1 T2 T3
Fixed base structure 4.9 1.987 1.52

Soil-structure interaction 5.697 3.154 2.603
The time periods are increasing with introduction of soil springs as predicted
at the outset.
Acceleration
Structure type Mode 1 Mode 2 Mode 3
Fixed base structure 0.4905 0.6867 0.7848


The acceleration decreases with soil-structure effect in this case
Damping
Fixed base structure 5% 5% 5%

Soil-structure interaction 9.2% 10.9% 11.15%
Damping constant for all mode for fixed base case varies with mode for
coupled analysis but is neither 5% min. nor 15% maximum but somewhere
in-between which is quite logical.
Base Shear (kN)
Fixed base structure 102 54.5 49.1

➔ A significant reduction in base shear, considering the soil effect, though

conceptually it can be predicted that amplitude of vibration will increase.
Shear Force per floor
Fixed Coupled Fixed Coupled Fixed Coupled

base with soil base with soil base with soil
Modes ➔ 1 1 2 2 3 3
Storey
1 10.8 7.2 5.74 1.2 5.17 0.0056
2 43.1 28.8 22.9 4.8 20.7 0.226
Top 48.4 32.4 25.8 5.4 23.3 0.255
➔ Significant variation in floor shears per mode.
Based on the above example it can be concluded that
• The major advantage with this technique is the calculation of the time period
without resorting to an elaborate modelling of the soil. Two representative spring
value for the foundation is capable of modifying the stiffness of the super-structure
having any conceivable degree of freedom.
• This cuts down significantly the modelling as well as the cost of computation.
• No rigid body motion exists.
• Stiffness matrix of the soil structure system is symmetric and real.
• The structure can be discretized to as many degrees of freedoms one choose to
select.

• Beam, plates, shell, bricks anything can be used to model the super structure
system thus do not generically violate the procedures followed for FEM analysis
of the superstructure.
• Since the matrix has no rigid body mode may be also be used directly for calculating
the static response too. No additional computational effort is required.
• Though approximate, furnishes a rational basis of estimating the modal damping
ratio per mode for the coupled soil structure-system.
• The results are logical and in general satisfies the trend as observed based on more
rigorous analysis based on complex damping and eigen value problem (where a
matrix of order n × n gets inflated to the order 2n × 2n thus adding to the cost of
computation).
1.3.6 Some fallacies in coupling of soil and structure

(Chowdhury 2008)
You will observe here that we had advocated two types of coupling of soil spring, one
vide Equation (1.1.3) where the soil spring is directly added to the diagonal stiffness
element of the structural matrix, meaning thereby that it is a parallel connection and
the other by Equation (1.3.7) which shows that the spring are in series.
The first method has developed from the theory of nodal compatibility and is a very
popular technique in practice for the root of its development is in the realms of matrix
analysis of structure and can very well be adapted in commercially available software.
While the second formulation is developed in the frequency domain analysis as
suggested by Veletsos for a harmonic oscillator having single degree of freedom coupled
to a translational and rocking spring.
The question that remains as to which one is more realistic and gives the true
interaction of the soil with structure especially when we model the soil as boundary
springs.
One of the major flaws in parallel spring model is, as the boundary elements are
discrete and not a continuum it only gives a local effect and also affects the structural
node only locally.
The intention here is not to challenge or shock the structural engineers who have
been doing this for ages. But putting on the hat of a theoretical physicist and probing
this formulation a bit more – it comes up with some very interesting result.
Let us imagine that the beam in Figure 1.1.1 is made of RCC of say dimension
450 × 900 supported on a compliant foundation where the soil is modeled as a spring.
Now we put a motor on the beam which gives some dynamic force Psin wm t-say. We
want to find out the dynamic response of the beam. The problem shows no ambiguity
for the beam along with the soil spring vibrates with natural frequencies that can be
obtained based on the lumped mass matrix at node i & j and the stiffness matrix derived
vide Equation (1.1.3) and then subsequent amplitude and stresses can be calculated
by the usual procedure.
Now let us presume 100 years down the road scientists have developed a material
whose Young’s Modulus is say 9 × 1020 kN/m2 and we build this beam (of same
dimension) with this material (not the spring which represents the soil) and pose the
question as to does the system vibrate? Looking at Figure 1.1.1 one can intuitively say
that yes it does vibrate, but the beam here being very stiff (lim k → ∞) undergoes rigid

body mode and the vibration is now guided by the stiffness of the spring only. Now
the question is – does Equation (1.1.3) reflects this phenomenon – amazingly not! For
putting this value of E = 9 × 1020 kN/m2 we find that [K]g becomes an infinitely stiff
matrix where the poor Kii and Kjj (whose order would be of 105 to 106 ) is completely
gobbled up by the stiffness values of the beam that are exponentially higher and would
start giving time periods that are zero.
Like patch test in FEM, it is a test we can use to check the sanctity of a stiffness
formulation. We call this an RB (short of Rigid Body) test and we see it fails this
test with parallel spring connection, especially when the structure has got significant
stiffness compared to soil.
Now if we put Equation (1.3.7) which is the series connection, to RB test, we find
that it passes the test with flying colors for as Limit of K → ∞ the first term in the right
hand side of Equation (1.3.7) approaches zero and we are left with the soil springs
values only based on which the body vibrates and satisfies RB test conditions posed
earlier.
In Equations (1.3.8) and (1.3.9) it is clearly seen that the soil flexibility gets directly
added to the diagonal and then on inversion affects all the terms of the [Ke ] and gives the
true interaction unlike parallel spring which affects only locally the interaction effects
and does not possibly gives a true picture when the stiffness of the superstructure
becomes quit high compared to that of the soil.
1.3.7 What makes the structural response attenuate

or amplify?
In Example 1.2.5 and 1.2.6 we had shown two opposite cases of dynamic soil structure
response. While in the case of the chimney the response is attenuated, in case of the
horizontal vessel the response is however significantly amplified. One would obviously
be curious and wonder why does it happen? The riddle is surely not difficult to answer.
Shown in Figure 1.3.3 is the generic nature of the acceleration curve used for design
of structures under earthquake. The nature of the curve is almost common/similar for
all the earthquake codes around the world.
Based on the curve (Figure 1.3.3) it is evident that when the structure is very stiff
or massive and its fixed base time period hovers around the vicinity of point A, the
dynamic soil structure effect can show significant amplification so long as the cou-
pled time period of the soil-structure system is within the zone C. Thus structures
like massive gravity dams, nuclear reactor buildings, Massive turbine foundations20 ,
large vessels supported on short pedestals (which are stiff) could show significant
amplification in response when the effect of soil is considered in the analysis.
While for any structure whose fixed base time period is somewhere between point
B and C, if exceeds the point C consideration of the soil effect can undergo a major
attenuation. Normal buildings, RCC, Steel Chimneys, elevated water tanks etc would
possibly fall in this category. Thus depending on the stiffness of the structure, its mass
distribution, dynamic property of the soil one can either save some money (if there is
20 For instance the structural configurations used for old LMW type Russian turbo-generators used
commonly in India for 210 MW plant.

B C
Sa
g
A
D
Time period (sec)
Figure 1.3.3 Generic response spectra curve for earthquake.
attenuation) or could result in more costly design (for amplified response) which may
vary from case to case.
1.4 THE ART OF MODELLING
Computer Modelling of soil & structure optimally to arrive at a meaningful solution

is an art by itself, and can well be a topic of a complete book. We present hereafter
some major techniques that has been found to effective & reasonable.
1.4.1 Some modelling techniques

Experience shows that in many cases young engineers eager on get-going mode would
start from the very outset with an elaborate model of the whole soil-structure system21 .
They spend significant amount of time on data input and checking of such massive
model and come up with a result whose qualitative difference with a much simpler
model is only marginal. Moreover trying to handle a big data-base, an inadvertent
modeling or input data error passing the scrutiny is not at all uncommon.
So at the very outset our suggestion would be, start with a simple model without
trying to over sophisticate the issue from the very out set22 .
Start with a test case or a simplified model to check the results. For instance a simple
model given in a book or the user manual of the software in use is a very good starting
to have some idea what types of element to choose, what order of refinement suffice
and what type of simplified idealization is acceptable.
21 Problem modeled with a minimum 1000 degrees of freedom!

22 Mentioning the fact that you have used eight nodded brick elements, or 9-noded plate elements based
on iso-parametric formulation may look impressive as technical jargons in a design basis report but may
not always be cost-effective solution.

Super structures above ground
Ground level
Railway Carriage
Underground Tunnel
Figure 1.4.1 An underground tunnel for movement of train in a metro city.
You will be amazed to find that in most of the cases, modeling the soil intelligently
as linear springs (whose values are judiciously chosen) can be good enough for many
major soil structure interaction analyses. Specially, when the structure is modeled in
3D, avoid using Finite elements to model soil and coupling it to the structure.
Firstly, the model becomes huge resulting in more engineering time plus gives results
which become difficult to decipher and does not necessarily always gives a more
accurate or better result compared to a relatively simplified model.
Start with a simple model (preferably a stick model) and add the soil spring to get
a first order feel of how much the soil affects its response23 .
Get a basic feel as to how much the results vary in terms of fixed base problem-
if found significant one should then and only then resort to a much more detailed
analysis.
If the variation is say within 15%, one can well ignore the soil effect and consider
the problem as a standard fixed base problem and proceed with the analysis.
Keep your eyes open but do not be biased on the issue. Optimize your engineering
effort to the best possible way.
There are certain types of problem where resorting to FEM however would become
almost essential. For instance for the problem considered in Figure 1.4.1, it would be
impossible to arrive at reasonable solution without an application of FEM.
Shown in Figure 1.4.1 is a sketch of an underground tunnel catering to movement
of high speed trains. The movement of train generates dynamic forces which travels
through the soil to the surface and could adversely affect the structures built on the
surface like buildings, water tanks etc and becomes an important study for engineers
undertaking such kind of projects.
23 A computer analysis is not mandatory at this stage, a simple hand calculation or an analysis in spread
sheet or MATHCAD would suffice.

It is but evident that for these cases of modeling, the soil as spring element will not
work and a comprehensive finite element modeling of the soil based on plane strain
element is required. Here also, while doing the modeling, our suggestion would be
start with a crude model (say 20 to 30 elements) to get a fill of the first order effects
and then progressively refine the model to get a more accurate result.
In static loading case in Chapter 4 (Vol. 1) we had explained the principles of
meshing of such plane strain problem. Under dynamic loading the principles meshing
are generally done based on the following
1 Find the time period of the exciting frequency (Ts ) of the soil medium as 4H/vs .
2 If vs is the shear wave velocity of the soil medium then for λ being the wavelength of
the propagating waves they are related by vs = f λ. Here f is the natural frequency
of the medium and f = 1/Ts .
3 Thus obtain λ = vs · Ts .
4 The mesh size should preferably be λ/10 to λ/4 for linear or bilinear/quadratic
elements chosen.
One of the major limitations in FEM for wave propagation problem is that the
boundary has to be taken to a significant distance away from the source to ensure
no waves are reflected back which would otherwise generate spurious modes. This
often makes the problem expensive in terms of data input, checking and run time.
Moreover, it is difficult to gauge at the outset as to where can the boundary be
terminated.
Infinite finite element as discussed in Chapter 4 (Vol. 1) is one alternative which has
been found to have a strong potential for catering to such problem.
Other than this, paraxial boundaries or providing viscous dampers at the boundary
of soil domain capable of absorbing the propagating waves are often used for this type
of problems24 .
Else boundary elements have also been used to model such infinite domains and are
coupled to the superstructure (modeled by FEM) and an effective solution has been
sought.
Unfortunately most of the commercially available software do not have the provi-
sion of adding matrix which can be assembled to the FEM matrix and an engineer has
to write his own special purpose software to cater to such problems.
Finally a word on the soil. . . . . .

Irrespective of whether we use springs or finite element to model the soil, the fun-
damental property on which the stiffness depends, are the value G (Dynamic shear
modulus) and ν, the Poisson’s ratio. We had discussed in detail as to how to arrive at
the appropriate design values of these two parameters in the next section.
In spite of all the techniques used it should be clearly mentioned that the parameters
are still marred by uncertainties and the results thus obtained should be mellowed with
some judgment which comes out only of experience and sustained practice.
24 Refer Chapter 5 (Vol. 1) for detailed discussion on this issue.

It is always preferable to do some parametric study by varying the design soil values
by (±)15 to 20% (depending on how reliable and exhaustive has been the geotechnical
investigation) and check how much these results affect the design values and preferably
a conservative and safe value should be chosen (based on this variance).
We mention in Table 1.4.1, some suggestive models for different classes of structures
where we start with a primary model (i.e. to get a basic feel of the response) and a
secondary model which is a further improvement to the primary model.
Table 1.4.1 Some suggestive models.
Sl. Structure
No. type Primary model Secondary model Remarks
1 Framed Stick model with soil 2D or 3D frame system

building considered as two uni- with masses lumped at
que spring (rotational nodes. Soil modeled as
and translational) consti- springs under each
tuting all the foundations individual foundation
2 High-rise 2D frame for the beam 3D frame for the beam The horizontal slab
building column system while the column system with need not be too
with shear shear wall modeled as shear wall modeled as refined and should
walls an equivalent cantile- plain stress elements. be good enough to
ver with soil springs The horizontal slabs generate requisite
under each column and modeled as plane stress stiffness in its own
the shear wall elements. Soil modeled plane
as springs below each
foundation
3 Chimneys 2D stick model with soil No further refinement is For local effect model
and eleva- idealized as springs usually warranted unless the shell or super-
ted water some local effect of soil structure as a stick
tanks is required to be studied and the soil a axis
on surrounding structure. symmetric plain
strain element
4 Frames 2D frames with soil A detailed 3D model Refer Chapter 2 (Vol. 2)
support- modeled as springs. constituting of beam on detailed modeling
ing rotary Bottom raft considered elements with master and technique for these
machines infinitely stiff hence only slave node option. The type of foundation.
lumped mass contribu- bottom raft discretised
tion is taken. Soil mode- into beam or plate elem-
led as springs. 3 to 4 ents with soil modeled
degrees of freedom as springs and connected
usually suffice. at each node of the raft
elements.
5 Dams and A simple stick model with A comprehensive 2D
embank- soil modeled as springs model with the dam
ments else time period may be broken up into plain
found from formula strain element and soil
suggested in code and modeled as springs or
modified by Veletsos’s further refined into 2D
formula. plain strain element
depending on the comp-
lexity of the soil or the
importance of the dam

1.4.2 To sum it up
Dynamic soil structure interaction is still in its early days and investigators are still
looking for answers to many problems which are encountered in practice.
For instance soil are modeled as linear springs based on elastic half space theory,
considering it as a linear isotropic medium, but in reality it is not so. Layered soil
phenomenon, pore pressure dissipation under dynamic loading, liquefaction potential
and its effect, infinite domain problem, non linear and inelastic behaviour, radiation
and geometric damping are some of the important factors on which research is still in
progress to arrive at a more realistic model amenable to design office practice.
What has been presented in this chapter is only an introductory concept and what
is in vogue in practice at the present.
Hopefully in days to come our understanding in some of the issues mentioned above
will be more profound and engineers and researchers would come up with results which
would be more realistic and reliable.
However a word of caution should be pertinent at this juncture.
As stated earlier as the uncertainty plaguing the problem is many, one should not
loose the final outcome of what we are trying to achieve i.e. a safe and sound structure
which can stand the vagaries of nature.
So one should not get lost in the maze of sophisticated mathematics and try to always
economize on the structure based on what the computer out put reflects25 .
For facilities important to society the results should always be mellowed with sound
engineering practice like good detailing, robust geometric configuration, and good
quality of time tested construction practice.
All these aspects are equally important for a structure to survive the wrath of Mother
Nature whose ways are still not very clearly known to us.
1.5 GEOTECHNICAL CONSIDERATIONS FOR DYNAMIC

SOIL STRUCTURE INTERACTION
In this section we deal with the geotechnical considerations which go into the process
of a successful dynamic soil-structure interaction analysis.
At the very outset we would request readers specially with a strong structural leaning
not to ignore this section. For our experience shows that nemesis of many mistakes
lies in misinterpretation of this particular topic. As such before launching yourself
into linear or non-linear finite element analysis of soil-structure system, the conceptual
aspect of the influencing soil parameters, its limitations and its effects should be clearly
understood.
As a pre-requisite, we expect that you have some background on. . .
• Some fundamental concepts of Soil Mechanics

• Basic concepts in Soil Dynamics
25 The output is nothing but a reflection of man’s limited knowledge of nature and only an approximate
quantification of an idealized mathematical model which could be in significant variance to reality in
spite of our best effort.

1.5.1 What parameters do I look for in the soil

report?
To start with we pose the above fundamental question. To readers having some back-
ground on this issue may find it intriguing for his obvious answer would be the dynamic
shear modulus (G) and Poisson’s ratio (ν).
The obvious query that subsequently comes to mind is, does it require a full section
to be devoted to this issue?
The answer would surely be an emphatic yes, for in our opinion the values
adapted are often misunderstood/abused in many a case, and often makes the analysis
questionable or unrealistic.
The reasons that could be attributed to it are as follows:
• Geotechnical test (lab or field) based on which data evaluated are not understood
properly. As the limitations of such data are not clearly made; often results in
incorrect interpretation.
• Data considered are often not relevant or correct in terms of real situation in the
field, specially for layered soil.
• Insufficient data and or lack of knowledge on the strain level to which the
foundation-structure system will be subjected to – specially during earthquake.
• Lack of dynamic test data and improperly co-related value from static soil
parameter which could be widely varying with the reality.
• Finally, often forgetting the bottom line that unlike man made material like con-
crete and steel, soil is far more heterogeneous and unpredictable; thus for a real
soil structure interaction it is unfair to have an analysis on an absolute scale. It
should preferably be done for a particular range of values and the best estimate is
to be made out of it – and this is where engineering judgment would count to a
large extent.
Having made the above statements, let us evaluate various aspects of dynamic
property of soil which are important for an integrated soil-structure interaction
analysis.
Before even looking at soil report the analyst should be clear with himself on
• The type of structure he is dealing with

• Type of foundation that is anticipated like shallow foundation (could be isolated
or combined footing), raft or piles etc.
• What analysis he is looking for like is it an analysis for machine induced load,
earthquake, blast force etc.
Understanding of the above criteria will not only help him in understanding the
data obtained from different tests but could also possibly make him realize their
interpretation in a more realistic perspective.
The engineering parameters we look for in the soil report for develop-
ing the soil model either for finite element or linear/non-linear spring dashpot
model are

G2
(shear stress) G1
1 2
Figure 1.5.1 Shear stress-strain curve of soil under cyclic loading.
1 Dynamic shear modulus (G) or shear wave velocity (vs )26

2 Poisson’s ratio (v)
3 Damping value of soil both radiative and material.
The values are usually obtained either from field test, laboratory test or from theo-
retical co-relation with other engineering soil parameters. Before we step further into
the topic it would possibly be worthwhile to understand how soil behaves under cyclic
loading and what its characteristics are.
It should be remembered that even under low strain, soil behavior is essentially
non-linear though at low strain it does show some kind of linearity.
Shown in Figure 1.5.1, is the shear stress-strain curve of soil under cyclic loading.
It is evident from the above figure that shear strain varies with stress, and goes on
increasing with number of cycles of loading.
Thus before an analysis is being carried out one has to have an idea about the average
strain range to which the soil will be subjected to under the induced dynamic loading.
The characteristic curve which shows the variation of shear modulus with respect
to shear strain is shown in Figure 1.5.1a.
The curve shown above is otherwise known as Seed and Idriss’s (1970) curve which
shows the variation of dynamic shear modulus of soil with shear strain.
26 Relationship being G = ρvs2 .

1.2
0.8
G/G0
0.6
0.4
0.2
0
0 0.01 0.1 1 10 100 1000
Strain Ratio
Figure 1.5.1a Variation of Shear Modulus with strain under cyclic loading. (Seed & Idriss 1970).
Soil subjected to stress by machine foundation are usually low strain and varies
between to 10−4 to 10−3 %.
However for an earthquake of even moderate magnitude this will be much higher-
having strain range varying to 10−2 to even 10−1 % for very severe earthquake.
Since it is difficult to gauge at the outset of an analysis how much strain the soil will
be subjected to, the correction factor to be used to modify the data as obtained in the
soil report becomes difficult to quantify.
On the contrary rendering no correction would result in assuming a more stiff soil
and the result obtained based on this could be significantly varying from the reality.
Fortunately or unfortunately most of the tests carried out in the field or in the
laboratory for determination of the dynamic shear modulus is based on low strain
range having values restricted to 10−4 %.
Thus it should be clearly understood that the dynamic shear modulus data furnished
in the soil report is only valid for LOW strain range and can be only used directly
for analysis where the strain induced in the soil is significantly low like in design of
machine foundations only. For earthquake analysis where the site is situated in an
area of moderate to severe earthquake zone, direct use of such soil dynamic data may
not be valid for design of normal structures, for the strain induced in soil is much
higher.
1.6 FIELD TESTS
The most common field tests that are carried out at site for evaluation of dynamic
shear modulus or shear wave velocity are
1 Block Vibration Test
2 Seismic cross hole
1.6.1 Block vibration test

In block vibration test as shown in Figure 1.6.1, an oscillator is placed on a concrete
block of size 1.5 m × 0.75 m × 0.7 m resting at foundation level and induces dynamic

Oscillator
Lx Fdn Level
Propagating waves
H=0.6 to 1.2 m
Figure 1.6.1 Schematic diagram for block vibration test.
loading on the soil. Two geo-phones are placed at a distance to pick up the signal from
the oscillator.
Once the oscillator induces dynamic force on the soil the geo-phones pick up this
signal and transfer them to an oscilloscope which shows an elliptical figure of Lissajous.
The operating speed of the oscillator is varied till the time the natural frequency of the
soil and the operating frequency of the oscillator matches (the Lissajous’ figure in the
oscilloscope becomes a perfect circle).
The shear wave velocity of the site is then given by
vs = 4fLx (1.6.1)
where vs = shear wave velocity of the soil; f = operating frequency of the oscillator
in cps; Lx = distance between the two geo-phones.
For arriving at meaningful results usually high frequency oscillators (>100 cps) are
put to use for which the waves generated are of the order of 0.6 to 1.2 m.
Thus results obtained from this test only influence soil of depth 0.6 to 1.2 m below
the depth of foundation and should not be used where piles or other types of deep
foundations having influence area propagating much deeper is used.
Trying to induce lower frequency calls for much heavier oscillators which make the
test uneconomical compared to other types of tests.
1.6.2 Seismic cross hole test

As shown in the schematic sketch in Figure 1.6.2, a probe is placed in a bore hole
to the desired depth and shear wave is generated in the soil by hitting it hard with a
hammer.
The waves are picked up by a geo-phone entrenched firmly to the casing of another
bore hole located at a known distance (Lx ) from the first hole.
The time taken to pick up the signal is measured by the oscilloscope.

Oscilloscope
Hammer
Ground level
Bore Hole casing Probe
Geo-Phone
Lx
Figure 1.6.2 Schematic diagram for seismic cross bore-hole test.
The dynamic shear modulus Gdyn is then obtained from the expression
2
ρ Lx
Gdyn = (1.6.2)
g t
where, Gdyn = dynamic shear modulus of the soil; ρ = weight density of soil; g =
acceleration due to gravity; Lx = distance between the two bore holes, and t = elapsed
time.
One of the major advantage with this test is that dynamic shear modulus can be mea-
sured to any desired depth and can very well be an integrated part of a SPT program.
The test is very effective in case the soil is layered in nature where visual inspection
of each layer is possible based on SPT test.
However, the strain range for test is again restricted to 10−4 % which is normally
less than the strain range experienced by machine foundations and earthquake analysis
and needs to be corrected to arrive at the design value of G.
1.6.3 How do I co-relate dynamic shear modulus when

I do not have data from the dynamic soil tests?
A not so uncommon phenomenon, that even puts an experienced engineer under dif-
ficult situation at times. In many cases it has been observed that no dynamic test has
been carried out during the geo-technical investigation – especially if it is a building

project. Though not unusual, but should not happen as a rule, for this shows the lack
of foresight on the part of the engineer while submitting the technical and commercial
proposal for a project. Even at the proposal stage the process involved in a plant is well
known to the bidder and all the concerned civil engineer has to do is to check with his
process department and find out if rotating machines are part of the process or not.
On the other hand knowing the location of a particular site one can easily find out
from the codes how active this zone is seismically and if felt reasonable all he has
to do is to include this additional cost of dynamic geotechnical investigation in his
commercial bid. People suffer from misnomer that dynamic tests are expensive-which
is actually not true, for an average dynamic test in international market takes roughly
US$ 20,000–25,000 which would however be 0.25% of a small petrochemical refinery
and possibly 0.1% of a combined cycle 350 MW power plant.
Lack of these tests can land up some of the equipments operating in such projects
into serious problem whose cost itself would constitute 30–40% of the whole
project cost!
So one has to decide on the risk involved – and come to a conclusion of its worth.
Though theoretical co-relation exist for evaluation of dynamic shear modulus of soil
from static soil test (which has been successfully used in project works), it is always
preferable to have these dynamic tests carried out at site, for not only does it imbibe
more confidence in the design process but engineer should also be aware that “theo-
retically co-related values have also varied widely with respect to actual field data, and
should be mellowed with judgment.” Considering the uncertainty prevalent in soil, is
surely not an easy task to accomplish.
1.7 THEORETICAL CO-RELATION FROM OTHER SOIL

PARAMETERS
The most outstanding work in establishing theoretical co-relation for evaluating the
dynamic property of soil has been done by Hardin, Drnevich, Richart, Seed, Idriss
to name a few27 . The expressions suggested by them have been successfully used for
many real projects by the engineers in the past. We are going to have a look at some
of them hereafter and understand their limitations if any.
1.7.1 Co-relation for sandy and gravelly soil

1.7.1.1 Hardin and Richart’s Formula
For rounded grained soil having void ratio less or equal to 0.8 the dynamic shear
modulus is given by (Hardin and Richart 1963)
2630(2.17 − e)2 √
G= σ0 in psi (1.7.1)
1+e
27 This is by no mean to ignore other researchers who have contributed significantly to this difficult study.
We only name a few, which are popular in practice.

For angular grained soil the dynamic shear modulus is given by
1230(2.97 − e)2 √
G= σ0 in psi. (1.7.2)
1+e
where, G = dynamic shear modulus of the soil in psi, e = in-situ void ratio of the soil
sample, σ0 = mean effective stress in psi = 0.333σv (1 + 2K0 ), σv = vertical effective
stress in psi, σh = horizontal effective in psi = K0 σv , K0 = earth pressure at rest, and
is a function of the plasticity index and the over-consolidation ratio.
The relationship between plasticity index, over-consolidation ratio and K0 is as
shown in the following figure.
Figure 1.7.1 Value of the K0 after Brooker & Ireland (1965) Reproduced by permission of the National
Research Council of Canada from the candian geotechnical Journal Vol-2 (1965).
1.7.1.2 Seed and Idriss Formula

The formula for dynamic modulus in this case, Seed and Idriss (1970) have been related
to relative density of sand which can usually be quantified from SPT test and is given by
√
G = 83.3K2 σ0 in psi (1.7.3)
Here K2 is a function of the relative density of the sand which can again be estimated
from the SPT value.
The relationship between SPT value and the relative density is as given Table 1.7.1.

Table 1.7.1 Soil properties with SPT values.
SPT value Compactness Relative density Angle of friction
0–4 Very loose 0–15 <28

4–10 Loose 15–35 28−30
10–30 Medium 35–65 30−36
30–50 Dense 65–80 36−41
>50 Very dense >85 >41
Table 1.7.2 Values of K2 versus relative density at

strain of 10−3 % (Seed and Idriss 1970).
Relative density (%) K2
90 70
75 61
60 52
45 43
40 40
30 34
For case of computer programming K2 can also be represented by the expression
K2 = 0.6Dr + 16 (1.7.3a)
It is to be noted that in this case to determine the relative density, the observed SPT
value has to be corrected for the overburden pressure and dilatancy to arrive at the
design SPT value before it is co-related with the above table.
1.7.1.3 Corrections to SPT value

Though available in standard textbooks of Soil Mechanics and Foundation Engineer-
ing, for brevity we present the correction expressions as mentioned hereafter.
For dilatancy correction if the observed SPT value (N0 ) is greater than 15 then the
corrected SPT value N is given by (Terzaghi and Peck 1967).
1
N = 15 + (N0 − 15) (1.7.4)
2
The overburden correction as per Peck et al. 1980 is given by
2000
N = 0.77N log10 for p ≥ 25 kPa (1.7.5)
p
For p ≤ 25 kPa, as per Murthy (1991)
4N
N = (1.7.6)
2 + 0.034p

in which, N = corrected SPT value for overburden, N = corrected SPT value for
dilatancy, p = gross overburden pressure in kN/m2 .
1.7.1.4 Ohsaki and Iwasaki’s formula

Ohsaki and Iwasaki (1973) have given co-relation for dynamic shear modulus directly
co-related to SPT value and is expressed as
G = 12000 N 0.8 in kPa (1.7.7)
Here N = design SPT value at the site after relevant corrections.
Example 1.7.1
As shown in Figure 1.7.2 is a small site having dimensions 18 m × 6 m which
would be supporting a Compressor unit and a few pumps, for which four
boreholes were dug at four corners as shown. The soil was found to be cohe-
sionless in nature and SPT values observed at the four bore holes are as tabled
hereafter
Depth BH1 BH2 BH3 BH4

(meter) (SPT value) (SPT value) (SPT value) (SPT value)
2 4 6 4 3
4 8 6 6 5
6 12 9 11 8
8 15 12 16 11
10 20 18 24 16
14 22 24 28 20
18.0
BH1 BH2
6.0
BH3 BH4
Figure 1.7.2

Based on Laboratory and field analyses following parameters were further

established:
Ground water table = 1.6 m, below grade level
Saturated density of soil = 22 kN/m3
Void ratio e0 = 0.58; Plasticity Index = 0.0; Poisson’s ratio = 0.32
Determine the best estimate of dynamic shear modulus (G) of soil at 10.6
meter below ground level presuming no dynamic soil test was done during
geo-technical investigation.
Solution:
Average observed SPT value at a depth of 10.0 meter
20 + 18 + 24 + 16
= = 19.5 = 20 (say)
4
Average observed SPT value at a depth of 14.0 meter
22 + 24 + 28 + 20
= = 23.5 = 24 (say)
4
At a depth of 10.6 meter below ground level based on linear inter-polation
average observed SPT Value
24 − 20
= × 0.6 + 20 = 20.6 ∼
= 21 (say)
4
The above observed SPT value has now to be corrected for dilatancy and
overburden pressure
1 Correction for dilatancy

As per Terzaghi, corrected SPT (N ) value is given by
1
N = 15 + (N − 15) for N > 15; (1.7.8)
2 0
1
or N = 15 + (21 − 15) = 18 (1.7.9)
2
2 Correction for overburden pressure

As per Peck
2000
N = 0.77N log10 for p ≥ 25 kPa (1.7.10)
p
N = Corrected SPT value for overburden; N = Corrected SPT value for

dilatancy; p = Gross overburden pressure in kN/m2 .

Here p = 22 × 10.6 = 233.2 kN/m2
2000
Substituting above in Peck’s formula we have, N = 0.77 N log10 = 13
233.2
Thus, corrected design SPT value = 13

Referring to table 1.7.1 for N = 13, Dr (Relative density) = 39.5%
Net overburden pressure at 10.6 meter level is expressed as
σv = (22 − 10) × 9 + 22 × 1.6 = 131.6 kN/m2 (18.718 p.s.i)
As there is no previous history of loading on the site O.C.R. = 1.

Thus for P.I. = 0.0 and O.C.R = 1 as per Brooker and Ireland’s curve we
have K0 = 0.48
Considering confining pressure
18.718
σo = 0.333σv (1 + 2K0 ); we have σ0 = (1 + 2 × 0.48) = 12.22 p.s.i.
3
As per Hardin and Richart’s formula
2630(2.17 − e)2 √
G= σ0
1+e
2630(2.17 − 0.58)2 √
G= 12.22 = 14710.5 p.s.i. (101426 kN/m2 )
1 + 0.58
(1.7.11)
As per Seed and Idriss formula
Referring to the chart given above for Dr = 39.5% and strain in the range of
10−3 % (usually valid for machine foundation) K2 = 40.
√
And as G = 83.3 K2 σ0 we have
√
G = 83.3 × 40 × 12.22 = 11647 p.s.i. (80308 kN/m2 )
Thus taking average value of G based on Hardin and Seed’s method
101426 + 80308
Average G = = 90867 kN/m2
2
As per Ohsaka and Iwaski’s formula

G = 12000 N 0.8 in kPa ➔ G = 12000 × (13)0.8

= 93397.6 kpa (93398 kN/m2 )
Thus it will be observed that variation with average G obtained based on

Hardin, Seed’s and Ohsaka’s formula is not significant and is of the order of
2.7%28 .
1.7.2 Co-relation for saturated clay

1.7.2.1 Hardin and Drnevich formula
Hardin and Drnevich (1973) have given the following formula applicable to clayey
soil as
(2.973 − e)2
Gmax = 1230 (OCR)k (σ0 )0.5 in psi (1.7.12)
(1 + e)
where, e = void ratio; OCR = over consolidation ratio; σ0 = mean effective stress in
psi = 0.333 (σv + 2σh ); σv = vertical effective stress in psi; σh = horizontal effective
stress in psi = K0 σv , K0 = earth pressure at rest, and is a function of the plasticity index
and the over-consolidation ratio.
k = is a function of the plasticity index (PI) of the soil and is given as
k = −5 × 10−8 (PI)3 −4 × 10−5 (PI)2 + 0.0092(PI) + 0.0025 (1.7.12a)
It is to be noted that Gmax as obtained above corresponds to a shear strain range of

0.25×10−4 % and needs to be modified for the appropriate strain range as appropriate
for a problem in hand based on the expression
Gmax
G= (1.7.13)
(1 + ψ/ψr )
Here ψ = desired strain range; ψr = reference strain range and is expressed as
τmax
ψr = × 100 and
Gmax
28 The point we are trying to make here is not to go by one formula, but check with possibly all of them
and comparing them to arrive at result which would possibly be best fit and hopefully be most realistic.
Here again it is to be noted that we had not used the angular sand formula of Hardin, if the soil
description does not reflect it or the soil has both rounded and angular grains an intermediate value has
tobe chosen judiciously.

$ %2 $ %2 0.5
1 + K0 1 − K0
τmax = (σv − u) sin φ + c cos φ − (σv − u) (1.7.14)
2 2
in which, σv = total vertical stressing in soil; u = pore pressure; c = cohesion of soil;

φ = angle of friction of soil, and K0 = coefficient of earth pressure at rest.
Example 1.7.2
It has been decided to place foundation of an industrial structure at 4.0 meter
below the existing ground level.
Based on laboratory and field tests it has been found that the Ground water
table is at a depth of 1.0 meter below GL.
Unconsolidated undrained triaxial tests reveal the sample to have the following
values:
• Cohesion value c = 0.21 kg/cm2

• Angle of resistance = 18 degrees
• Pore pressure = 0.0 kg/cm2
Consolidation tests reveal that it had a history of pre-consolidation pressure

of 200 kN/m2 :
• Initial void ratio = 0.61

• Plasticity limit PI = 35
• Saturated unit weight of soil = 19 kN/m3
The site has a history of moderate to severe earthquake when from previous
record it is observed to generate a strain range up to 0.1%.
Calculate the dynamic shear modulus of soil for this predicted strain
range.
Solution:
For foundation located at 4.0 meter below the ground level net vertical pressure
σv = 19 × 1.0 + (19 − 10) × 3.0 = 46 kN/m2 (6.54 psi)
200
OCR = = 4.34, for plasticity index of 35 from Brooker and Ireland’s,
46
chart K0 = 1.1

Thus considering σo = 0.333σv (1 + 2K0 ), we have
σo = 0.333 × 6.54(1 + 2 × 1.1) = 6.976 psi
(2.973 − e)2
Gmax = 1230 (OCR)k (σ0 )0.5 in psi
(1 + e)
Here k = 0.27 for PI = 35 as per Equation 1.7.12a

Thus substituting the values we have
(2.973 − 0.61)2
Gmax = 1230 (4.34)0.27 (6.976)0.5
(1 + 0.61)
= 16746 psi (115465 kN/m2 )
Calculation for Shear stress
$ %2 $ %2 0.5
1 + K0 1 − K0
τmax = (σv − u) sin φ + c cos φ − (σv − u)
2 2
$ %2 $ %20.5
1 + 1.1 1 − 1.1
or, τmax = (6.54) sin 18 + 3.0457 cos 18 − (6.54)
2 2
= 5.00 psi.
ψr = Reference strain range and is expressed as
τmax 5.00
ψr = × 100 = × 100 = 0.0299%
Gmax 16746
Thus for 0.1% strain. . .
Gmax 115465
G= ➔G= = 26577 kN/m2
ψ 0.1
1+ ψr 1+ 0.0299
It is thus observed that dynamic shear modulus is 23% of the theoretically calc-
ulated data.
Based on the above example it would perhaps be not difficult to realize that how
important role does the strain range plays on the design value of dynamic shear
modulus of soil.

1.8 ESTIMATION OF MATERIAL DAMPING OF SOIL
Damping plays a significant part in the overall response of soil structure system. While
for structural members material damping plays a significant part (mostly considered
as Rayleigh damping), for soil, two types of damping are basically involved.
• Radiation damping
• Material damping
Radiation or geometric damping of a soil foundation system is a mean by which

the energy is dissipated by means of radiation from the source and is a function of
mass and inertia of the system29 . Material damping of the soil foundation system is
a mean by which the energy is dissipated by hysteresis and is an inherent property of
the constituting material of the soil.
This can very well be found from resonant column test in the laboratory when
after the soil has been vibrated the exciter is stopped and successive amplitudes are
measured. If a1 and a2 are two successive amplitudes then
&
a2 a2 2
Dm = ln 4π + ln
2 (1.8.1)
a1 a1
The total damping ratio of a soil foundation system is sum of radiation and material
damping. It is generally observed that material damping has a significant magnitude
relative to radiation damping specially in rotational modes. In such cases total damping
rather than geometric damping should be used to obtain the response of the structure
foundation system.
For translatory mode, on the contrary material damping plays an insignificant role
and may be neglected in the analysis. Thus for tall narrow structures like chimney,
Boiler structures, tall buildings where the coupled horizontal and rocking mode could
play significant role it would perhaps be realistic to also consider the material damping
of soil in order to have a meaningful response.
1.8.1 Whitman’s formula

Whitman (1973) has suggested that total damping (geometric + material) for a
machine foundation can be obtained from the expressions
Horizontal Mode
0.31
Dh = (1.8.2)
M
ρr30
29 We have dealt this detail in Chapter 2 (Vol. 2) – “Design of Machine foundations”.

For vertical mode
0.49
Dv = (1.8.3)
M
ρr30
For rocking mode

⎡' (0.5 ' ( ⎤−1
Iθ Iθ
Dθ = 0.05 + 0.1 ⎣ 1+ ⎦ (1.8.4)
ρr50 4ρr50
Here M = mass of foundation plus structure or machine vibrating; Iθ = mass

moment of inertia of foundation plus machine/structure about a horizontal axis
through the base of the foundation perpendicular to the plane of rocking; r0 =
equivalent radius of footing, and ρ = mass density of soil.
1.8.2 Hardin’ formula

Hardin (1965) has expressed material damping of sandy soil by the expression
0.985ψr0.2
Dm = √ (1.8.5)
σ0
Here notations are same as expressed earlier except the fact that the confining pres-
sure σ0 is expressed in kPa. The equation is valid for shear strain amplitude of 10−6
to 10−4 with a confining pressure of 24 kPa to 144 kPa.
For a particular strain range the value obtained above can be corrected based on the
expression
Dc ψ/ψr
= (1.8.6)
Dm 1 + ψ/ψr
Example 1.8.1
For the example as shown in Example 1.7.2, estimate the damping ratio of the
soil as per Hardin’s formula. The soil properties remain same as given in Example
1.7.2.
Solution:
Based on the solution furnished in Example 1.7.1 value of dynamic shear
modulus is given by
G = 90867 kN/m2 ; σ0 = 12.22 p.s.i. (85.9 kN/m2 ); K0 = 0.48

For design SPT value N = 13, φ = 31◦ , and σv = 131.6 kN/m2
$ %2
1 + K0
Considering, τmax = (σv − u) sin φ + c cos φ
2
$ %2 0.5
1 − K0
− (σv − u) , we have
2
$ %2 $ %2 0.5
1 + 0.48 1 − 0.48
τmax = 131.6 sin 31 − 131.6
2 2
= 36.67 kN/m2
τ 36.67
As ψ = × 100, we have ψr = × 100 = 0.0404%
G 90867
0.985ψr0.2
Considering Dm = √ , we have
σ0
0.985(0.0404)0.2
Dm = √ = 0.056
85.9
Thus material damping ratio is estimated as 5.6%.
1.8.3 Ishibashi and Zhang’s formula

Ishibashi and Zhang (1993) has proposed an expression for the damping ratio of
plastic and non-plastic soil and is given by
2
1 + exp(−0.0145PI1.3 ) G G
ζ = 0.333 0.586 − 1.547 +1 (1.8.7)
2 Gmax Gmax
The notations for the above expression are already explained in earlier formulas.
We show below variation of damping ratio with G/Gmax for different Plasticity
Index based on the above formula.
It will observed (Figure 1.8.1) that as G/Gmax reduces, as damping ratio goes
on increasing meaning thereby that as strain increases damping ratio goes on
increasing. Variation of Damping with strain vide Equation (1.8.6) is shown in
Figure 1.8.1a.

Variation of damping ratio

0.35
0.3
PI = 10
PI = 20
0.25
PI = 30
Damping ratio
PI = 40
0.2
PI = 50
PI = 60
0.15
PI = 70
0.1 PI = 80
PI = 90
0.05 PI = 100
0
7
9
1
1
0
0.
0.
0.
0.
0.
0.
0.
0.
0.
G/Gmax
Figure 1.8.1 Variation of damping with plasticity index as per Ishibashi and Zhang (1993).
1.2
0.8
D/Dr
0.6
0.4
0.2
0
0 0.01 0.1 1 10 100 1000
Strain Ratio
Figure 1.8.1a Variation of damping ratio with strain under cyclic loading.
Example 1.8.2
For the clayey soil sample as shown in Example 1.7.2, determine the damping
ratio for the strain range level of 0.1% based on Zhang’s formula. Consider all
soil properties same as Example 1.7.2?
Solution:
Based on earlier example we have seen that plasticity index PI = 35.

In Example 1.7.2 we have already calculated that for 0.1% strain G/Gmax =
0.230 Substituting the above in Ishibashi and Zhang’s formula we have
1 + exp(−1.47)
ζ = 0.333 [0.586(0.23)2 − 1.547 × 0.23 + 1] = 0.1382
2
Thus estimated damping ratio is 13.82%.
1.9 ALL THINGS SAID AND DONE HOW DO WE ESTIMATE

THE STRAIN IN SOIL, SPECIALLY IF THE STRAIN IS LARGE?
We acknowledge at the very outset that posing the question, though easy, is not very
easy to answer. The uncertainties involved are so widely varying that it would be
difficult to give a precise answer to this issue. To the best of our knowledge there is
no straight forward answer to this problem and the best one can achieve is a reason-
able estimate or can possibly study a range of values and try to predict the overall
behavior.
For high speed centrifugal machine foundation it does not pose a serious problem
for at the low strain range a few percent here and there does not contribute a significant
variation to these values.
But for impact type of machines (hammer foundations) and slow speed machines
(coal mill foundations, reciprocating compressors) induced strain could be larger than
strain developed during field test, for which the correct estimation of Gdyn and damping
becomes important.
For earthquake of course the strain would invariably be larger than measured during
test, even for a moderate earthquake when as the strain range increases, degrada-
tion in soil stiffness becomes significant and has a major contribution to the overall
response.
It is obvious that strain induced in soil will depend upon the strength of dynamic
loading, the geological condition of the site, stress history of soil and a number of
other factors. So the point remains that if there exists no previous records of strain
from similar machine in same site or from previously occurring earthquake data how
does one rationalize the strain?
We discuss below some of the techniques which could be used for evaluation of the
strain induced in the soil.
1.9.1 Estimation of strain in soil for machine

foundation
For machine foundations the present practice of arriving at strain dependent dynamic
shear modulus and damping can be structured as follows:

• Start with the field observed/lab obtained data for Gdyn and damping as furnished
in the soil report which would usually correspond to the strain range of 10−4 to
10−3 %.
• Calculate natural frequency of the soil-foundation system based on free vibration
analysis.
• For rotating mass type calculate the transmissibility factor based on the expression
!
r2 1 + (2ζ r)2
Tr = ! (1.9.1)
(1 − r2 )2 + (2ζ r)2
• For constant force excitation (like in hammer foundation) calculate transmissibil-

ity factor based on expression
!
1 + (2ζ r)2
Tr = ! (1.9.2)
(1 − r2 )2 + (2ζ r)2
where r = ωωmn , ωm = operating frequency of the machine; ωn = natural frequency

of the foundation; ζ = damping ratio of the soil.
• Find out the pseudo-static force by multiplying the vertical unbalanced force of
the machine by the transmissibility factor as mentioned above.
• Find out the dynamic stress induced in the soil by dividing the above force by the
foundation plan area.
• The approximate shear strain in the soil is given by the expression.
12qdyn
ψ(%) = (1.9.3)
G
• Verify the strain obtained against the initial value. If they vary significantly find
out the new G value based on the calculated strain and repeat the process as
mentioned above till it converges.
The above method is surely non-rigorous but generates an answer which will give
reasonably accurate results for practical analysis of machine foundations. For more
complicated soil with varying properties a more rigorous analysis based on Finite
element analysis is possible. This will be discussed later on.
The above technique is now explained based on a suitable numerical example.
Example 1.9.1
A centrifugal turbine driven compressor has foundation dimension of 6 m ×
3.2 m × 2.5 m. The weight of the compressor is 300 kN. The unbalanced mass

on the shaft is 3.5 kN · sec2 /m rotating at an eccentricity of 0.4 mm having

operating frequency of 1800 rpm.
The soil investigation has revealed the soil data as follows
• SPT = 13 (After dilatancy and overburden correction)

• Plasticity Index (PI) = 0
• Poissons Ratio = 0.3
Calculate the correct value of dynamic shear modulus and damping.
Solution:
Based on Ohsaka’s formula G = 12000 × (13)0.8 = 93397.65 kN/m2
Assumed strain level = 1 × 10−4 %
Weight of foundation = 6 × 3.2 × 2.5 × 25 = 1200 kN; Weight of machine =
300 kN, Total weight = 1500 kN
Mass of foundation + machine = 1500/9.81 = 152.9052 kN · sec2 /m

6 × 3.2
Equivalent radius of the foundation (r0 ) = = 2.472155 m
π
Equivalent vertical spring stiffness of soil30 .
4Gr0 4 × 93397.65 × 2.472

Kz = = = 1.32 × 106 kN/m
1−υ 0.7
Kz 1.32 × 106
ωn = = = 92.89 rad/sec;
m 152.9
1800 × 2 × π
ωm = = 188 radian/sec r = ωm /ωn = 2.029.
60
0.4
Pdyn = m · e · ωm
2 = 3.5 × × (188)2 = 49.7428 kN.
1000
Considering transmissibility as
!
r2 1 + (2ζ r)2
Tr = ! we have, Tr = 0.65078.
(1 − r2 )2 + (2ζ r)2
30 Refer to Chapter 5 (Vol. 1) – Basic Concepts of Soil Dynamics, for details of this formula.

0.65078 × 49.7428
Equivalent static force on foundation = = 1.686 kN/m2
6 × 3.2
12qdyn 12 × 1.686
Considering ψ(%) = = = 2.17 × 10−4 %
G 93397.65
Gmax
Considering G = ψ
(1 + ψr )
93397.65
we have, New G = = 29497.88 kN/m.
2.17×10−4
1+ 1×10−4
We proceed with second cycle of iteration with this new value of G.
Shown below is such iteration for 14 cycles
Cycles 1 2 3 4 5 6 7
Gdyn 93397.65 29497.88 22553.77 20646.63 20025.37 19811.83 19737.08

Damping 0.012987 0.189764 0.21998 0.228656 0.231517 0.232505 0.232851
Kz 1.32 × 4.17 × 3.19 × 2.92 × 2.83 × 2.80 × 2.79 ×
10+06 10+05 10+05 10+05 10+05 10+05 10+05
ωn 92.89143 52.20392 45.64753 43.67493 43.01282 42.78288 42.7021
r(wm /wn ) 2.029203 3.610755 4.129371 4.315875 4.382311 4.405864 4.414199
Tr 0.65078 0.298033 0.255622 0.243327 0.239243 0.23783 0.237334
Pdyn 49.74281 49.74281 49.74281 49.74281 49.74281 49.74281 49.74281
Peq 32.37161 14.82497 12.71537 12.10377 11.90063 11.83032 11.80565
qdyn 1.686022 0.772134 0.662259 0.630405 0.619824 0.616163 0.614878
Strain 2.17 × 3.14 × 3.52 × 3.66 × 3.71 × 3.73 × 3.74 ×
10−04 10−04 10−04 10−04 10−04 10−04 10−04
Cycles 8 9 10 11 12 13 14
Gdyn 19710.75 19701.45 19698.17 19697.01 19696.60 19696.45 19696.40

Damping 0.232973 0.233016 0.233031 0.233037 0.233039 0.233039 0.23304
Kz 2.78 × 2.78 × 2.78 × 2.78 × 2.78 × 2.78 × 2.78 ×
10+05 10+05 10+05 10+05 10+05 10+05 10+05
ωn 42.6736 42.66353 42.65998 42.65872 42.65827 42.65812 42.65806
r(wm /wn ) 4.417147 4.418189 4.418558 4.418688 4.418734 4.41875 4.418756
Tr 0.237159 0.237097 0.237075 0.237068 0.237065 0.237064 0.237064
Pdyn 49.74281 49.74281 49.74281 49.74281 49.74281 49.74281 49.74281
Peq 11.79696 11.79388 11.7928 11.79241 11.79228 11.79223 11.79221
qdyn 0.614425 0.614265 0.614208 0.614188 0.614181 0.614179 0.614178
Strain 3.74 × 3.74 × 3.74 × 3.74 × 3.74× 3.74 × 3.74 ×
10−04 10−04 10−04 10−04 10−04 10−04 10−04
We show hereafter the variation of strain, damping and Gdyn per cycle in Figs. 1.9.1 to 1.9.3.

Variation of strain per cycle

4.00E-04
3.50E-04
3.00E-04
2.50E-04
Strain(%)
2.00E-04
1.50E-04
1.00E-04
5.00E-05
0.00E+00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Number of cycles
Figure 1.9.1 Variation of strain (%) per cycle.
Variation of damping ratio with strain

0.35
0.3
0.25
Damping ratio
0.2
0.15
0.1
0.05
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Number of cycles
Figure 1.9.2 Variation of material damping per cycle.

Variation of Gdyn with strain

100000
80000
Gdyn(kN/m2)
60000
40000
20000
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Number of cycles
Figure 1.9.3 Variation of dynamic shear modulus per cycle.
From the tables and the above plots it is observed that the value becomes
constant after 7th cycle of iteration based on which we conclude that design
values are as follows
Gdyn = 19700 kN/m; Material Damping ratio = 0.23, and Estimated strain
range = 3.74 × 10−4 %.
Thus actual design of the foundation shall be carried out based on this corrected
value instead of the initial values as mentioned in the soil report.
1.9.2 Estimation of soil strain for earthquake

analysis
For earthquake analysis things are surely more complicated for not only the forces
induced in the soil is much more complex, the behavior itself is different from machine
foundations.
While in machine foundation the force is induced in the soil from the structure
in earthquake the force is induced within the soil where the soil first start vibrat-
ing based on the waves propagating through it. Thus acceleration, it is excited to,
depends on the free field vibration of the site. This acceleration induced in the soil
generates shear strain on which the stiffness degradation and damping ratio would
depend.
Though the analysis shown hereafter is based for isotropic homogenous medium
it can well be extended to layered soil having variable property based on weighted
average.

Propagation of Earthquake
Figure 1.9.4 Schematic diagram of an industrial site with propagating earthquake waves.
Shown in Figure 1.9.4 is a schematic diagram of an industrial site with propagating

earthquake waves. The depth of the site (H) is considered to the bedrock level from
where the waves are presumed to be propagating31 .
The waves propagating at bedrock level travels upward and hits the surface (z = 0)
when the site surface undergoes a motion. However as surface is free, it is free to shake
as such no strain energy develops at the surface.
The motion of such elastic waves propagating through an elastic medium can be
defined by the partial differential equation
∂ 2u ∂ 2u
= vs2 2 (1.9.4)
∂t 2 ∂z
Here u = displacement of the soil and is a function of time t and depth z, vs = Shear
wave velocity of the soil.
Considering u(z, t) = φ(z)ψ(t), (1.9.5)
31 For site having no bedrock this level is usually considered at the depth where shear wave velocity of
the site is greater or equal to 600 m/sec. Based on SPT value this can be considered as the depth where
design SPT value is greater than 50.

we have substituting in the equation above
φ(z)ψ̈(t) = vs2 φ̈(z)ψ(t)

ψ̈(t) φ̈(z)
or = = −p2 (say)
vs ψ(t)
2 φ(z)
The above on separation gives two homogenous equations
φ̈(z) + p2 φ(z) = 0 and ψ̈(t) + p2 vs2 ψ(t) = 0 (1.9.6)
The above gives solution
φ(z) = A cos pz + B sin pz (1.9.7)
at z = 0 as there will be no shear strain hence du/dz = 0
φ̇(z) = −Ap sin pz + Bp cos pz = 0, at z = 0
The above gives the constant B = 0, from which we deduce, φ(z) = A cos pz
At z = H as the soil is confined, hence we have, u(z, t) = 0 → A cos pH = 0
(2n − 1) π
p= (1.9.8)
2H
Considering, ψ(t) = C cos λt + D sin λt (1.9.9)
we have, λ = p2 vs2 , where λ is the eigen-value of the problem.

Knowing from our fundamental knowledge of vibration that
(2n − 1)2 π 2 2 (2n − 1) π

ωn2 = λ = vs ➔ or ωn = vs rad/sec (1.9.10)
4H 2 2H
2π
Considering T = ωn , we have
4H
Tn = secs (1.9.11)
(2n − 1) vs
Here Tn is known as the free field time period of the site for n numbers of mode.

The corresponding eigenvectors are given by
(2n − 1)πz
φ(z) = cos (1.9.12)
2H
The displacement vector is given by
u(z) = κi φi (z)Sd (1.9.13)
where Sd = Displacement spectrum of the site which can again be represented as
Sa
u(z) = κi φi (z) (1.9.14)
ωn2
where Sa = acceleration spectrum of the site and is a function of the free field time
period of the site32 .
In which,

m i φi
κi = Modal mass participation factor = (1.9.15)
mi φi2
The modal participation factor can thus be considered as
)H )H
mi φi πz * πz
κi = = γ z cos γ z cos2 (1.9.16)
mi φi2 2H 2H
0 0
8
The above on integration by parts gives, κi = (1.9.17)
π +2
Thus for the present problem
32βSa H 2 (2n − 1)π z

u(z) = cos (1.9.18)
(2n − 1)2 π 2 (π + 2)vs2 2H
ZI
Here β = 2R the IS code factor33
32 This response spectrum is usually available as site response spectra in absence of which charts furnished
in National codes are usually followed.
33 Presently code does not have any guidline for R for soil. It has been observed that a value between R = 2
to 3 usually gives realistic results.

∂u
Considering shear strain γz = ∂z we have
16βSa H (2n − 1)π z

γz = − sin (1.9.19)
(2n − 1)π(π + 2)vs
2 2H
Considering G = ρvs2 we have
16βSa Hρ (2n − 1)π z

γz = − sin (1.9.20)
(2n − 1)(π + 2)πG 2H
Here G = dynamic shear modulus of the soil, ρ = mass density of the soil.
For foundation at a particular depth below the free surface for which we have
obtained the dynamic shear modulus based on field or lab test34 . We start initially to
find out the shear strain in the soil based on this value considering a strain range of
10−3 /10−4 %.
The steps that are followed subsequently to arrive at the corrected G and damping
value are furnished hereafter (Chowdhury 2008).
1 Identify the bedrock level (H) of the site

2 Find out the shear wave velocity from the expression G = ρvs2
4H
3 Find out the free field time period of the site from the expression Tn = (2n−1)v s
4 Based on the site response spectra/spectra given in code and damping value as
obtained in soil report obtain the acceleration Sa
5 Obtain shear strain for the soil profile based on the expression γz =
16βSa Hρ
− (2n−1)π(π+2)G sin (2n−1)πz
2H
6 Check if this strain is near or equal to the initial strain(10−3 to 10−4 )%35 .
Gmax
7 If there exists a significant variation correct G based on the equation G = (1+ψ/ψ r)
8 Find out the ratio G/Gmax
9 Obtain new damping ratio based on Zhang’s expression
2
1 + exp(−0.0145PI1.3 ) G G
ζ = 0.333 0.586 − 1.547 +1
2 Gmax Gmax
10 Repeat the steps as mentioned from 2 to 7 till the strain is same as previous cycle.
The value for which the strain becomes constant is the corrected Dynamic shear
modulus of the soil.
The above steps will now be further elaborated by a suitable problem.
34 Or from theoretical co-relation.

35 In absence of this input in soil report consider 10−3 % for soft soil and 10−4 % for medium stiff soil.

Example 1.9.2
For a particular site susceptible to earthquake it was observed based on soil
investigation that bed rock exists at 20 meters below ground level. Seismic cross-
hole test reveals average dynamic shear modulus of the soil to be 154897 kN/m2
at a reference strain of 1 × 10−5 .
Considering density of soil as 19 kN/m3, and plasticity index as 35. Calculate
the corrected dynamic shear modulus of soil and damping at 2.5 meter below
GL where foundation of a particular structure will be placed. Consider IS 1893
curves to evaluate the acceleration pertaining to a particular time period.
Solution:
Depth of soil over bedrock = 20 m, Density of soil = 19 kN/m3 , Thus mass
19
density of soil (ρ) = 9.81 = 1.936 kN · sec2 /m, Dynamic shear Modulus of soil
(G) = 154897 kN/m . 2
G
Shear wave velocity of soil (vs ) = = 282.8 m/sec.
ρ
4H
Considering Tn = we have for fundamental mode
(2n − 1)vs
4 × 20
T1 = = 0.283 sec
282.8
Considering Zhang’s formula
2
1 + exp(−0.0145PI1.3 ) G G
ζ = 0.333 0.586 − 1.547 +1
2 Gmax Gmax
Taking G/Gmax = 1 for first cycle and PI = 35 we have, ζ = 0.798%

For damping @ 0.798% and time period of 0.283 sec Sa = 5.75 m/sec2 from
IS code.
For this case the code factor Z is considered as 0.24, I = 1.2, R = 3. The value
of R is chosen as 3 in this case because for PI = 35 it is assumed that the soil has
high plasticity and thus has reasonable ductility. As a matter of fact there is no
guideline at moment prevalent in the code and an engineer has to use his own
judgment here.
Substituting the above data in the expression
2Sa Hρ (2n − 1)πz

γz = − sin
(2n − 1)πG 2H

We have at depth of 2.5 meter below ground level γz = 2.5145 × 10−5
Gmax
Considering G = ψ
we have G = 44073.2 kN/m2
(1 + ψr )
We proceed with next cycle of iteration with this new value of G.

In table below we show how the data converges for 10 successive cycles
Cycles 1 2 3 4 5
GkN/m2 154897 44073.21001 39512.54 37131.85 35405.98

Vs m/sec 282.7998707 150.8499063 142.8319 138.4621 135.206
T sec 0.282885561 0.53032847 0.560099 0.577775 0.59169
Sa m/sec2 5.75 1.9 1.85 1.85 1.8
Strain 2.51454 × 10−05 2.9202 × 10−05 3.17 × 10−05 3.37 × 10−05 3.44 × 10−05
G/Gmax 1 0.284532367 0.255089 0.23972 0.228578
Damping
ratio(%) 0.798 12.425 13.167 13.562 13.852
Cycles 6 7 8 9 10
GkN/m2 34857.41 34437.41 34113.86 33863.42 33668.85

Vs m/sec6 134.1545 133.3438 132.716 132.2279 131.8475
T sec 0.596327 0.599953 0.602791 0.605016 0.606762
Sa m/sec2 1.8 1.8 1.8 1.8 1.8
Strain 3.5 × 10−05 3.54 × 10−05 3.57 × 10−05 3.6 × 10−05 3.62E−05
G/Gmax 0.225036 0.222325 0.220236 0.218619 0.217363
Damping
ratio(%) 13.945 14.016 14.071 14.114 14.147
Thus based on above calculation we may take, corrected G value =

33600 kN/m2 ; Damping ratio = 0.14. Variation of shear modulus and damping
ratio (%) with number of cycles are shown in Figs. 1.9.5 and 1.9.6.
Varaition of Gdyn at foundation level
200000
150000
100000 G
50000
0
1 2 3 4 5 6 7 8 9 10
Iteration num ber
Figure 1.9.5 Variation of shear modulus with number of cycles.

Varaition Damping ratio(%) at foundation level

Damping Ratio(%) 15.000
10.000
5.000
0.000
1 2 3 4 5 6 7 8 9 10
Number of iterations
Figure 1.9.6 Variation of damping ratio (%) with number of cycles.
1.9.3 What do we do if the soil is layered with varying

soil property?
Till now the theories we have presented assumes soil as a homogenous isotropic
medium but in reality in all possibility the soil encountered at a particular site will
be layered in nature.
Shown in Figure 1.9.7, is a typical stratified soil profile where the shear modulus,
density of soil and Poisson’s ratio vary with depth.
For most of the cases taking a weighted average is the normal practice where the
average dynamic property may be taken as
G1 · H 1 + G 2 · H 2 + G 3 · H 3 + G 4 · H 4
Gav = (1.9.21)
H1 + H 2 + H 3 + H 4
and same principle be applied for mass density and Poisson’s ratio.
However for very important structures or site susceptible to major earthquakes
methods based on finite element analysis may be applied to arrive at a design dynamic
modulus and damping value36 .
Shown in Figure 1.9.8, is the finite element model of a site having layered soil
property. In this case the soil is modelled as plane strain element to the bedrock
boundary and each individual layers having different properties can very easily be
catered to.
To start with we assume G value as obtained from soil report and consider the
damping ratio based on Zhang’s formula considering G/Gmax = 1 at the strain level
of 10−3 /10−4 % say.
Suppose the previous earthquake history shows that shaking has taken place for
duration of 3 sec maximum, we select duration of 6 sec for analysis.
36 In such cases preferably site response spectra of the particular should be used. Moreover some previous
history of shaking and its duration should be available for analysis.

Figure 1.9.7 Layered soil strata with varying soil property.
Super Structure
Layered Soil
Bed Rock
Figure 1.9.8 Finite element model of a site having layered soil property.
Next for duration of 6 sec we input the Sa/g curve for the particular damping and
perform a time history analysis of the system for 6 sec.
From the output for each layer we find the average shear strain. Based on the output
average strain, we correct the value of Gdyn for the next cycle and also the damping
ratio and do a second cycle of time history for 6 sec. We repeat this process for a
couple of times till the values have stabilised with respect to the previous cycle.
The value of G and damping considered in last cycle where the strain has stabilised
are the dynamic shear modulus and damping of the soil.

Calculation of shear modulus of soil based on the free field time period is an effective
tool for assessing the dynamic shear modulus of soil. However, there is a possibility
that the time period obtained by this method could be higher than the reality unless
proper consideration are given for the confining effect of the surrounding soil and
proper judgement of the depth is made. ATC (1982) has defined Hmax as the depth
limited to 183 m having low strain shear wave velocity of 760 m/sec.
1.9.4 Checklist of parameters to be looked in the

soil report
Based on above discussion, the parameters which require particular attention in a soil
report from the engineer are summarised as follows:
1.9.4.1 Field test

• Has SPT test been carried out?
The obvious intention is to find out N on which G value depends. This can also be
utilised to cross check the field observed dynamic data.
• If SPT values are furnished are the observed data or corrections need to be done?
A point to be checked for field observed data as shown earlier needs to be corrected.
While the soil consultant will do this correction during his own calculation of bearing
capacity of soil for foundation recommendations, usually furnishes observed field data
while furnishing the bore log detail in the report. So for your calculation this data needs
to be corrected. If you are not too sure you can back calculate it from recommended
φ value.
• Has Ground Water Table been established during boring?
Usually provided in a soil report but better to check for this has significant effect on
the net vertical stress.
• Has any dynamic field test carried out?

◦ Block Vibration test
◦ Seismic cross hole test
One of the tests should be a part of the soil report. But do not take the values
furnished sacrosanct. Back check with theoretical co-relation to establish if the order
is close, if not you do have the right to ask your soil consultant why there is this
discrepancy. There could be special geological condition which could result in such
discrepancy and you should be clear about it.
• If the above tests are carried out, what is the strain range induced in the soil during
the test?
This is something usually not supplied by the soil consultant who usually would
recommend a unique G value. This should not be acceptable to you.
You should clear it at the very outset when providing him the specification for
Geotechnical investigation that this is an input you are looking for and it should be

a part of his report. It is more realistic to start with this value rather than guessing a
theoretical value of 10−3 /10−4 %.
1.9.4.2 Laboratory tests

• Check Atterberg’s limit – gives values of liquid limit, plastic limit, plasticity
index etc.
Generally speaking37 as a ritual, structural engineers/analysts ignore this topic. Our
suggestion would be, do not disregard this for this is the basic data which gives you
the first insight into the fact as to how the soil behaves. Moreover plasticity index
being an important property it is all the more important that you should pay attention
to this.
• Triaxial test gives values of c, φ and pre-consolidation history

Again given a back handed treatment by the analyst who without going through
the test data would prefer to pick up the numbers which concerns him (ca and φ). We
suggest go through the test and develop enough skill to interpret the pre-consolidation
stress. Make sure to ask during enquiry to the consultant to supply this data. For Over
Consolidation Ratio (OCR) plays a very important role in arriving at the correct value
of Gdyn .
During interpretation if need be, seek help of a geotechnical specialist to make sure
what has been understood is correct – this will save a lot of headache in the long run.
• Bulk density and void ratio of soil

• Grain distribution to check if the soil is gap graded, uniformly graded, or poorly
graded. Relative density of cohesion less soil is highly dependent on this.
1.10 EPILOGUE
The technology described in this chapter to our perception is still in its infancy and we
are optimistic that with time and research that is being carried out all over the world,
we shall be in a better position in future to predict more realistically the dynamic
properties of soil which affect the response of structure.
Whatever we have presented here is what we believe is simple to apply, provides
reasonably realistic results and practical for day to day design office practice.
There is hardly any comprehensive text which gives a defined picture on this
issue. Most of the techniques developed herein are based on research papers
(names furnished in the reference) and typical practices followed in some design
offices38 .
We urge the readers to go through these papers which we believe will give them
further insight to the problem.
37 Exceptions are always there. . ..

38 Even consultants who require to use these type of technology is very limited.

The ideas presented in this chapter is to make the reader aware of the limitations
prevalent with soil and also to caution him on the fact that without these values
realistically estimated, the whole analysis related to dynamic soil structure interaction
could become a questionable exercise.
So be aware and use your judicious best to furnish a meaningful design.
SUGGESTED READING39
1 Cohen, M. & Jennings, P., ‘Silent Boundary Methods For Transient Analysis’, Computa-
tional Method in Transient Analysis – Computational Method in Mechanics, Vol. 1, North
Holland.
2 Dasgupta, S.P. & Kameswara Rao, N.S.V.K. 1976, ‘Some finite element solutions in the
dynamics of circular footings’, Proc 2nd International Conference on Numerical Methods
in Geomechanics, Blacksburg USA.
3 Dasgupta S.P. & Kameswara Rao, N.S.V.K. 1978, Dynamics of rectangular footings by
Finite elements, Journal of GT Division ASCE, Vol. 104, No. 5.
4 Gazetas, G & Tassoulas, A.L. 1987, ‘Horizontal Stiffness of Arbitrarily shaped embedded
foundation’, Journal of GT Division, ASCE, Vol. 113, No. 5.
5 Kameswar Rao, N.S.V. 1977, ‘Dynamic soil structure system – A Brief Review’, J. Struct.
Engg., India, Vol. 4.
6 Lysmer, J. & Kuhlemeyer, R.L. 1969, ‘Finite dynamic model of infinite media’, J.EM.Divn,
ASCE, EM4.
7 Segol, G., Abel, J.F. & Lee, P.C.Y. 1975, ‘Finite element Mesh Gradation of surface waves’,
J. GT Division, ASCE, Vol. 101, GT 11.
8 Wolf, J.P. 1985, Dynamic Soil Structure Interaction, Prenctice-Hall Englewood Cliffs, NJ.
9 Wolf, J.P. 1988, Dynamic Soil Structure Interaction in Time Domain, Prenctice-Hall,
Englewood Cliffs, NJ.
10 Wolf, J.P. 1994, Foundation Vibration Analysis: Using Simple Physical Model, Prentice-
Hall, Englewood-Cliffs, NJ.
11 Whitman, R.V. 1970, Soil Structure Interaction – Seismic design for Nuclear power plants,
The MIT press, Cambridge, Massachusets.
39 This topic being relatively new, there are not much reference books (other than reference 8, 9 & 10)
which deal this topic comprehensively. Many literatures though have mentioned the interaction effect
in their work. The references suggested are thus mostly restricted to research papers, which we would
request you to get hold of and rummage through patiently.

SW1451 c001 PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

SW1451 c001 PDF

Hochgeladen von

Copyright:

Verfügbare Formate

Chapter 1

Dynamic soil structure interaction

• Static soil structure interaction

1.1.1 The marriage of soil and structure

∗∗ For theoretical background on these topics please consider Volume 1.

© 2009 Taylor & Francis Group, London, UK

1.1.2 What does the interaction mean?

2 The reason for these effects we will discuss subsequently.

© 2009 Taylor & Francis Group, London, UK

Node iii Node j

Soil Spring Ki Soil Spring Kj

Figure 1.1.1 Equivalent beam element connected to soil springs.

Figure 1.1.2 Mathematical model of the equivalent beam element.

and the displacement vector is given by

{δ} = <δ1 θ1 θ2 δ2 θ3 θ4>T (1.1.2)

© 2009 Taylor & Francis Group, London, UK

1.1.3 It is an expensive analysis do we need to do it?

Vs ≤ 600 m/sec. (1.1.7)

© 2009 Taylor & Francis Group, London, UK

Considering, I = Ar2 and m = ρ · A, where A = area of cross section; r = radius of

Shear Wave Velocity for Soil-Structure interaction for

Shear Wave velocity

800.00 steel chimney

© 2009 Taylor & Francis Group, London, UK

1.1.4 Different soil models and their coupling

1 Equivalent soil springs connected to foundations modeled as beams, plates, shell

1.2 MATHEMATICAL MODELING OF SOIL & STRUCTURE

1.2.1 Lagrangian formulation for 2D frames or stick-models

© 2009 Taylor & Francis Group, London, UK

Figure 1.2.1 2D Mathematical model for soil structure interaction.

The equation for kinetic energy of the system may be written as

© 2009 Taylor & Francis Group, London, UK

© 2009 Taylor & Francis Group, London, UK

Figure 1.2.3 Spring supported beam.

Figure 1.2.4 Two degrees of freedom of a beam element.

The element matrix for such case is given by

Assembling the element matrix for the two beams we have

⎡ 12EI 6EI −12EI 6EI ⎤

© 2009 Taylor & Francis Group, London, UK

Figure 1.2.5 Bridge girder across abutments.

© 2009 Taylor & Francis Group, London, UK

Figure 1.2.6 Idealisation of the bridge girder ignoring soil effect.

The unconstrained combined stiffness matrix as

Substituting the values we have

© 2009 Taylor & Francis Group, London, UK

76800 96000 −76800 96000 0 0 0 0 0 0

160000 −96000 80000 0 0 0 0

Lumped mass at each node is given by → Mii = 25 × 5 × 2.5/9.81 = 31.85

© 2009 Taylor & Francis Group, London, UK

Figure 1.2.7 Idealisation of the bridge girder considering soil effect.

Considering the equation

Eigen value 692 1328 2684 4897 7448 7787 11722

Here r0 = 3.38 m and for G = 2500 kN/m2 and ν = 0.3 Kz = 48285.71

160000 −96000 80000 0 0 0 0 0

The Mass matrix remains same as derived earlier.

Eigen-values 75 692 2684 3045 7067 7448 9489 11722

© 2009 Taylor & Francis Group, London, UK

1.2.2 What happens if the raft is f lexible?

• λL ≤ π4 the raft will behave as rigid raft

1.2.2.1 Calculation of spring constant for rigid raft

7 Refer Example 4.6.1 in Chapter 4 (Vol. 1) for further details.

© 2009 Taylor & Francis Group, London, UK

1.2.2.2 Calculation of spring constant for f lexible raft