Dynamic Soil Structure Interaction of Bridge Piers Supported On Well
Dynamic Soil Structure Interaction of Bridge Piers Supported On Well
A R T I C L E I N F O A BS T RAC T
Keywords: For both steel and RCC Bridges passing rivers or creeks, common practice in many countries is to provide
Well foundation concrete wells to support the bridge girders. For many bridges that are strategically important in terms of
Bridge Pier defense or trade, it is essential that they remain functional even after a strong earthquake hits the structure. The
Modal analysis present state of the art for design of well foundation is still marred with a number of uncertainties where a
Puzrevsky functions
simplistic pseudo static analysis of its response only prevails, though it is a well-known fact that loads from
DSSI
super structure, character of soil and its stiffness plays an important role in defining its dynamic characteristics.
The present paper is thus an attempt to present a dynamic analysis model trying to cater to a number of such
deficiencies as cited above and also provide a practical model (amenable to design office application) that can be
used to estimate the pier, well and soil's dynamic interaction
1. Introduction Design and construction of such well foundations and piers are
usually carried out as per recommendations of the codes of respective
Well foundations otherwise called caissons are often deployed to countries and some of the most commonly used codes are IRC 6,45 and
support a number of important bridges around the world. Verrazano 78 [1–3], AASHTO [4], CALTRANS [5], and Eurocode 8 Part2 & 4 [6].
Narrows, San Francisco-Oakland bay bridge in USA, Rokko Island The design procedure adapted by many of these codes for seismic
Bridge in Japan, Mahanadi River and Kolaghat Rail Bridge in India are analysis is quite simplified (though popular perception is that it is
some of the bridges that have been built on large diameter well conservative), and overlooks a number of crucial issues like:
foundations.
In all these cases the super-structure or the top deck is supported 1. Ignoring self weight of pier while calculating its time-period.
on massive piers, which in turn are supported on large diameter 2. Ignoring the shear deformation characteristics of the pier column, as
caissons transferring the load to foundation soil. in many cases, especially for flyovers having variable height depend-
Because of large diameter (5–12 m) and depth (15–30 m or even ing on slenderness ratio of the pier, this can well dominate the pier's
more) of the caissons, it has long been assumed that well foundations dynamic behavior.
are far too massive and stiff to be affected by any vibration either due to 3. Inertial and kinematical interaction of the pier with well and its
moving traffic or earthquake. Thus most of the codes of practice surrounding soil.
assumes the bridge pier supporting the superstructure to be fixed at
top of well. Present paper proposes two mathematical models based on which
However, observations on performance of some of the bridges in many of these limitations in terms of earthquake analysis can be
recent earthquakes like Loma Prieta (USA) 1989, Kobe (Japan) 1999, overcome to arrive at a realistic result. The analysis is analytical in
Nepal 2015 (in Nepal and India), it is found that assuming the pier as nature and does not require any special purpose software to be used
fixed at its base is certainly not realistic. Despite being huge, well and can well be carried out in general purpose utility software like,
foundations are significantly affected by the propagating waves during MATHCAD, MATLAB or even a spread sheet if necessary.
an earthquake that affects the response of the bridge pier in turn.
⁎
Corresponding author.
E-mail addresses: [email protected] (I. Chowdhury), [email protected] (R. Tarafdar), [email protected] (A. Ghosh),
[email protected] (S.P. Dasgupta).
http://dx.doi.org/10.1016/j.soildyn.2017.03.005
Received 16 March 2016; Received in revised form 17 January 2017; Accepted 7 March 2017
Available online 22 March 2017
0267-7261/ © 2017 Elsevier Ltd. All rights reserved.
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
2. Practice as in trend from superstructure considered as a lumped mass at its head (see
Fig. 1). The inertial mass of the pier itself is ignored.
Before we present the mathematical models for analysis of such Above expression is similar to what has been proposed in [4,5] that
system, the practice as in trend [1–3], and other codes used inter- recommends an expression
nationally [4–6] are briefly reviewed.
Present recommendations in [1–3] for design of bridges under D
T = 0.32
seismic force are based on a study by Murthy and Jain [7] to bridge the F (2)
gap between [1] and state of the art international practice as prevalent
and made a significant improvement/modification to [1], compared to here D and F are same as defined in Eq. (1), except that values are in
its previous version. FPS unit.
The study and subsequent recommendation are in line with The basis of these expressions is actually T = 2π m / k where m is
procedure as furnished in [4,5] and is almost analogous in philosophy. the lumped mass considered at top of pier and k the stiffness of pier in
In the present study, though a number of recommendations were fundamental mode.
made on different aspect of bridge design under seismic loading like Japanese code JRA Part V [8] also recommends an expression
adapting different response reduction factor (R) for different parts of similar in nature and is given by.
the bridge, development of plastic hinges, damping properties etc., the T = 2.01 δ . Here δ= static deflection at the tip of pier due to lateral
overview is restricted to the seismic response of the bridge pier part load in meter. This can well be derived from the expression
T = 2π m / k → 2π δ / g . Assuming g=9.81 m/s2. One can easily arrive
and its foundation only.
at the equation as proposed in the Japanese code.
As per [1], time-period of pier in fundamental mode is computed
This again shows that the idealized model used to compute time
from the expression
period in this case is also a single degree lumped mass inverted
D pendulum type, like one used in [1–6].
T=2 Fig. 1 shows typical bridge piers deployed in practice to support the
1000F (1)
superstructure of a bridge/flyover.
In Eq. (1), D = Dead load reaction from superstructure in kN. F = For the lumped mass model as shown in Fig. 1, ignoring the inertial
Horizontal force in kN, to be applied at center of mass of the effect of pier, one may argue is conservative, as the added mass of pier
superstructure for one mm horizontal deflection of the bridge along will only go on to elongate the time period, which will either reduce the
considered direction of horizontal force. spectral acceleration or it may even remain invariant-depending on
In this context, it should be noted that code has not stated whether pier geometry and superstructure load. However, as deflection, mo-
we use an Euler-Bernoulli type beam or a Timoshenko type beam for ment and shear are proportional to square of the time period will be
computation of F. The practice is usually to use beams ignoring the lower bound. The pier mass should also be incorporated in the time
shear deformation. period expression to arrive at a realistic result.
The basis of Eq. (1) is obviously assuming the bridge pier as a In a particular flyover, supporting piers (having same diameter/
member having single degree of freedom fixed at its base, with load width) may have different height. They are usually tallest at the center
252
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
Fig. 1. Typical bridge piers deployed in practice and its analytical model.
of the bridge while progressively gets shorter at both ends. Thus shall be taken into consideration.
depending on height and diameter of the pier, shear deformation can In reference [6] Part 4, vide clause 5.4.2, it has also recommended
significantly affect the dynamic response vis-à-vis amplitude moment that pier supported on caissons (well foundations), appropriate spring
and shear. This effect has been completely ignored in Eqs. (1), (2) and stiffness and damping to be considered for analysis of both inertial (II)
the Japanese code. and kinematical interaction (KI) effect. Sub-Clause 6P of section 5.4.2
For better structural stability and also being aesthetically more has also cited conditions under which considering KI is mandatory.
pleasing, in many countries tapered shear walls like one shown in However what would be appropriate values of spring and damping,
Sections 2–2 (Fig. 1) are being increasingly used. Computing F for such and how to consider the II and KI effect, code has not given any specific
shear walls becomes difficult for adaptation in Eqs. (1–2). directive.
Finally, considering the pier is an inverted pendulum type struc- Thus, [6] though provides a much superior recommendation than
ture, computing moments and shears based on fundamental mode only [4] and [5] in terms of DSSI effect, the issue yet remains somewhat
may not be adequate, as higher mode contribution can be noteworthy. open. When possibly a more specific and explicit guideline of how to
All the references cited above have no recommendations/guidelines on consider the effect of II and KI, what should be the appropriate spring
how to compute higher time- periods, possibly leaving it to a computer and damping stiffness to be considered for a caisson foundation like
to do the slogging. As far as dynamic soil-structure interaction (DSSI) one cited in [20] and [23] would certainly improve upon the procedure
is concerned, both [4] and [5] have made no direct reference to it, suggested.
though both have possibly tried to consider the local site amplification
by extending the peak zone of the response spectrum curve as a 3. Proposed method
function of soil stiffness. The same principle has been adapted in [1]
also. Before getting into DSSI aspect of the problem, fixed base dynamic
The response spectrum curve followed in [1] is as shown in Fig. 2 response of the pier is evaluated first, and in the process it is examined
for three-type soil. The curves are plotted for 5% structural damping. if it is possible to improve upon the expression as cited in Eqs. (1) and
As seen from the curve a bridge pier supported on foundation (2).
resting on soft soil (N ≤10) where N=number of blow counts in a SPT For a bridge pier like as shown in Fig. 1, considering the pier fixed
test, is expected to have a spectral acceleration 50% more than a same at base (i.e. at the top of foundation), it may be assumed to behave as a
foundation resting on rock (N ≥30). The amplification is expected to be cantilever beam whose equation of free vibration under flexure can be
20% more when the foundation is resting on medium soil (N ≥15). expressed as
However, there are ambiguities in this approach. It is observed
from Fig. 2 that till T=0.7 s, structures resting on rock, intermediate or ∂ 4u ∂ 2u
EI 4
= ρA 2
soft soil (non-liquefiable), all are subjected to same acceleration. A ∂z ∂t (3)
number of studies carried out previously on DSSI have shown that stiff
here ρ= mass density of beam material, A=cross sectional area E=
structures resting on soft soil shows significant amplification (Dowrick
Young's modulus of beam, I =Moment of inertia of the beam, u=
[9]).
displacement in lateral direction. Let u(z, t ) = Y (z )q(t ), then based on
Similarly for a fixed based structure having fundamental time
separation of variable technique, Eq. (3) can be separated into two
period T=0.8 s say is expected to experience a spectral acceleration of
linear differential equation and one of which is
1.5g with 5% damping on rock. However, if placed on softer soil
considering soil structure interaction will show an elongation in time
period (depending on the soil stiffness) and the spectral value could
well reduce, depending on the time period and level of damping
imposed by the soil.
Other than this, no provision has been kept for DSSI in [1,4] and
[5], where code has recommended special investigation for near field
sites (mentioning the effect of DSSI herein), but what and how to carry
out this investigation, code has remained tactically reticent.
Reference [6] on the contrary has dealt the issue more realistically,
where time and again in different place they have stressed the
importance of DSSI. In Clause 4.1.4 of [6] the code has specifically
recommended that - under action of a unit load, when the soil flexibility
contributes to more than 20% deflection of the pier tip, DSSI effect Fig. 2. Response spectrum curve as per IRC 206 (2010) for different soil.
253
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
⎛ d 4Y ⎞ Table 1
EI ⎜ 4 ⎟ − λ 4Y = 0 Value of factor CTb for first three modes**.
⎝ dz ⎠
Mode Ω=1 Ω=2 Ω=4 Ω=6 Ω=8 Ω=10
where
1 4.033 5.422 7.461 9.053 10.403 11.598
ρAω2
λ4 = 2 0.384 0.394 0.399 0.401 0.402 0.402
EI (4) 3 0.121 0.122 0.122 0.122 0.122 0.122
Generic solution to this equation is given by
** Intermediate values can be linearly interpolated without any significant errors
Y = C1 sin λz + C2 cos λz + C3 sinh λz + C4 cosh λz (5)
Values of CTib for different values of Ω are furnished in Table 1 for
Imposing the appropriate boundary conditions for a cantilever
first three modes.
beam (Fig. 1):
Eq. (15) will give an accurate value of time period of the bridge pier
We have the shape function as
which is tall and slender, when flexural deformation predominate, but
μm z μm z ⎛ μz μ z⎞ would underestimate the response, when pier height is short and rigid
Ym = sin − sinh − αm⎜cos m − cosh m ⎟
H H ⎝ H H ⎠ (6) where shear deformation would predominate or for intermediate
height when both flexure and shear deformation plays almost an equal
here m=number of modes 1,2,3,……….∞
2m − 1 role in its dynamic behavior.
μm = 1.875, 4.694, 7.855, 2 π for m ≥4 and
sin μm + sinh μm 3.1. Shear deformation behavior of the pier
αm =
cos μm + cosh μm (7)
For a cantilever shear beam (which is usually short and wide) fixed
Eq. (6) is actually the eigenvectors of the beam and is expressed as
at base of uniform area A, equation of free vibration is expressed as
μi z μi z ⎛ μz μz⎞
fib (z ) = sin − sinh − αi⎜cos i − cosh i ⎟ ∂ 2u ∂ 2u
H H ⎝ H H⎠ (8) G = ρc 2
∂z 2 ∂t (16)
Here superscript ‘b′ stands for bending and subscripts i stands for
mode numbers 1,2,3…. ∂ 2u 1 ∂ 2u
→ 2
= 2 2
Based on Rayleigh's technique as per Hurty and Rubenstein [10], it ∂z Vs ∂t (17)
can be shown that stiffness matrix of the pier can be expressed as
here
H
b″ G = Dynamic shear modulus of the bridge pier material, ρc= Mass
[K ] = EI ∫0 fi (z )f jb″ (z )dz
(9) density of bridge pier material, Vs=Shear wave velocity of the bridge
Considering the flanged pier cap plus dead and live load from top pier material, G =ρc. Vs2.
deck acts as a mass (Wd/g), lumped at the tip of pier, the mass matrix Solving Eq. (17), it can be shown (Chowdhury and Dasgupta [13])
can be expressed as (Meirovitch [11]) that Eigen vector of the problem can be expressed as
(2i − 1)πz
γcA H Wd b fis = cos
[M ] =
g
∫0 fib (z )f jb (z )dz + f (H )f jb (H )
g i (10)
2H (18)
here superscript s stands for the term shear and subscript i stands for
here
mode 1,2,3….
E = Young's modulus of bridge pier material, A = Area of cross
Based on Rayleigh's technique it has been shown [13] that stiffness
section of pier, I= Moment of inertia of pier cross section, γc = Weight
matrix of the system for such shear beams can be expressed as
density of pier material, H = Height of pier, Wd= Load coming from the
H
top deck on the pier, g = Acceleration due to gravity, i= Mode number
1,2,3….
[K] = ηGA ∫0 fis ′ (z ) f js ′ (z )dz
(19)
Considering ξ= z/H, we convert Eqs. (9) and (10) from global to
Eq. (19) in natural co-ordinate ξ=z/H can be expressed as
natural co-ordinate when we have
ηGA 1
[ K] =
EI
∫0
1
fib″ (ξ )f jb″ (ξ )dξ
[ K] =
H
∫0 fis ′ (ξ ) f js ′ (ξ )dξ
(20)
H3 (11)
Mass matrix in natural co-ordinate can be expressed as
γcAH 1 Wd b
[M] =
g
∫0 fib (ξ )f jb (ξ )dξ + f (1)f jb (1)
g i (12)
γcAH ⎡ 1 ⎤
[M] =
g ⎣
⎢ ∫0 fis (ξ ) f js (ξ )dξ + Ωfis (0) f js (0)⎥
⎦ (21)
γcAH ⎡ 1 ⎤
→[M] =
g ⎣
⎢ ∫0 fib (ξ )f jb (ξ )dξ + Ωfib (1)f jb (1)⎥
⎦ (13)
In Eq. (20), η is the shear correction factor for the pier cross section
and is dependent on shape of the pier cross section. Values of η, for
Here Ω is ratio of the mass lumped at its tip (Wd/g) to that of mass of different type of pier cross sections can be obtained from various
pier itself. standard references for e.g. Shames and Dym [14].
Considering the general Eigen value problem [K]{φ} = λ[M]{φ} and Considering the general Eigen value problem [K]{φ} = λ[M]{φ} and
solving the same by Jacobi's method (Bathe [12]), one can obtain [λ], solving the same by Jacobi's method, one can obtain [λ], and knowing
and knowing ω = λ and T = 2π / ω , when we finally have ω = λ and T = 2π / ω , we finally have
2π γcA γcH2
Tib = H2 Tis =
2π
λib EIg (14) λis ηGg (22)
γcA H
→Tib = CTibH2 →Tis = CTis
EIg (15) Vs η (23)
254
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
Table 2 solution for time periods for the fixed based bridge pier and is more
Value of factor CTs for first three modes**. accurate then the formulation furnished in code which also restricts
itself to fundamental mode only.
Mode Ω=1 Ω=2 Ω=4 Ω=6 Ω=8 Ω=10
As per modal response analysis
1 7.171 9.383 12.712 15.337 17.574 19.558
Sa
2 1.731 1.789 1.823 1.836 1.842 1.846 Sd =
3 0.909 0.916 0.92 0.922 0.922 0.923 ω2 (30)
here Sd = modal amplitude, Sa= spectral acceleration corresponding to
** Intermediate values can be linearly interpolated without any significant errors.
time period T, and ω= natural frequency of the system when ω=2π/T.
Thus as per norms as prescribed in IS 1893 [18] amplitude of
here Vs= Shear wave velocity of concrete and values of CTis for different
vibration is expressed as
values of Ω are as furnished in Table 2.
Time-period of flexure and shear deformation can now be combined Sai
Sdi = Tei 2
as explained hereafter. 4π 2 (31)
If a load vector P is applied on a body, displacement of the body due
Sai T
to bending as well as shear deformation [14] can be expressed as, usi = κiβ Tei 2[φib] {fib (ξ )}
4π 2 (32)
δt = δb + δs (24)
⎧ 2 b ⎫
EI ⎛ SaiTei
2⎞
T ⎪ d f (ξ ) ⎪
where, δt = total displacement of the system; δb = displacement due to Mzi = − κiβ ⎜⎜ ⎟⎟[φib] ⎨ i 2 ⎬
bending; and δs = displacement due to shear. H2 ⎝ 4π 2 ⎠ ⎪
⎩ dξ ⎭
⎪
(33)
Eq. (24) can be expressed as
⎧ 3 b ⎫
EI ⎛ SaiTei ⎞ b T ⎪ d fi (ξ ) ⎪
2
P P P Vzi = − κiβ 3⎜
⎜ ⎟[φ ] ⎨
2 ⎟ i 3 ⎪
⎬
= + H ⎝ 4π ⎠ ⎪
⎩ dξ ⎭ (34)
Ke Kb Ks (25)
Substituting Eqs. (15) and (23) in Eq. (28), one can have the 4. Dynamic soil structure interaction of pier with well
combined effective time period Tei as foundation
⎡ ⎛ C s ⎞2 2(1 + ν ) ⎤
WpH3 ⎢ ⎥ Investigation on soil-structure interaction effect between well and
Tei = CTib 1 + ⎜⎜ Tib ⎟⎟
EIg ⎢⎣ ⎝ CTi ⎠ ηrs 2 ⎥⎦ pier has not drawn much attention, though considerable research has
(29)
been carried out on behavior of rigid caissons/cylindrical foundation
here embedded in soil under earthquake force.
Wp = weight of the pier, γc.A.H, rs = slenderness ratio of pier (H/R), Thakkar et al. [19] studied the inertial effect of well on overall
R = radius of gyration of pier cross section (I =AR2), ν= Poisson's ratio response of the pier-well system. They used a lumped mass stick model
and η = shear correction factor. for this study considering both the pier and the foundation.
Validity of coupling of the type as shown in Eqs. (24) to (29) has Gerloymos and Gazetas [20] used a Winkler type model for
already been tested against detailed two and three-dimensional finite dynamic soil structure interaction of rigid caisson foundation under
element analysis (Chowdhury et al. [17,40]) and has been found to be seismic loading for linear analysis. In this work the focus was more on
in excellent agreement. estimating a realistic value of the soil springs that can be used to
Eq. (29) which is the general equation of time-period considering estimate the foundation's dynamic response. The validation of soil
combined effect of flexure and shear is of particular interest, as it springs were carried out by comparing the results against three
explains the effect of shear deformation that goes on to elongate the dimensional finite element analyses. Mandal and Jain [21] made a
time period of the pier. It is observed from Eq. (29) that it is strongly study on pier and well interaction in non-linear domain to assess the
dependent on slenderness ratio rs. As R reduces, the component rs load capacity of pier. Chowdhury et al. [22] used a finite beam on
increase rapidly (in terms of its square), and for large values of rs elastic foundation without considering the shear deformation effect to
makes the second term within the parenthesis insignificant when determine the effect of well and soil on overall dynamic response of the
flexural mode dominates the response. pier. The results showed that considering pier well and soil interaction
One can also say with certainty that Eq. (29) is a more general there was an amplification in moment and shear values in pier cross
255
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
Table 3
Value of factor κi for first three modes**.
• A simplified model ⎡ M +m
⎢ f i (Mf Zb + mihi ) ⎤⎧ uxi ⎫
⎥⎨ ⎬
• A generalized solution ⎢ (M Z + m h ) J + M Z 2 + m h 2 ⎥⎩ θi ⎭
⎣ f b i i θ f b i i ⎦
In this case mathematical model perceived is as shown in Fig. 4. In Eq. (37) Mf= Mass of foundation, Jθ = Mass moment of inertia of
The well foundation is assumed as a rigid cylinder partially embedded the foundation, Zb is c.g. height of foundation only from center of
256
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
here Z, R and I are code factors and Sai is spectral acceleration of the p (t ) ⎛ ⎧ 1−ζ
qθ (t ) = ∼ θ ⎜⎜1 + e−ζθωθt ⎨
⎪
(52)
[M]{φi}qï (t ) + [C]{φi}qi̇ (t ) + [K]{φi}qi(t ) = {P} (41)
As per theory as proposed by Veletsos and Meek [33] it has been
Pre multiplying Eq. (41) by {φi}T we have shown by Chowdhury et al. [34] that dynamic amplification due to
DSSI can be finally expressed as
{φi}T [M]{φi}qi(t ) + {φi}T [C]{φ}qi(t )
ux θ
+{φi}T [K]{φi}qi(t ) = {φi}T {P} (42) AF = 1 + +h
us us (53)
Based on theory of vibration Eq. (42) can be expressed as
In Eq. (53) us= Fixed based amplitude of the pier at its c.g. height
2
qï (t ) + 2ζiωiqi̇ (t ) + ωi qi(t ) = pi (t ) hi , ux is the translational displacement of the foundation and θ=
(43)
Rotational amplitude of the foundation.
Eq. (43) gives two uncoupled equations Thus for the present problem
φxx qx (t ) φxθ qx (t )
qẍ (t ) + 2ζxωxqẋ (t ) + ωx 2qx (t ) = px (t ) (44) AFxi = 1 + + hi
usi usi (54)
And
Eq. (54) can be expressed in an expanded format as
qθ̈ (t ) + 2ζθωθ qθ̇ (t ) + ωθ 2qθ (t ) = pθ (t ) (45)
φxx ⎡⎢ px (t ) ⎛ ⎧ 1−ζ
⎜1 + e−ζxωxt ⎨
⎪
AFxi = 1 + ∼ ⎜
x
Solution of Eq. (44) will give usi ⎢⎣ kx (ωi ) ⎝ ⎩ 1 + ζx
⎪
⎛ ⎞ ⎫⎞⎤
qx (t ) = e−ζxωxt ⎜A cos ωx 1 − ζx 2 + B sin ωx 1 − ζx 2 t ⎟ ×sin ωx 1 − ζx 2 t − cos ωx 1 − ζx 2 t ⎬⎟⎥
⎝ ⎠ ⎭⎠⎦
+px (t )/ ωx 2 (46) ⎛ ⎞ φ ⎡ p (t ) ⎛ ⎧ 1−ζ
M
+⎜⎜ bi ⎟⎟ xθ ⎢ ∼ x ⎜1 + e−ζxωxt ⎨
⎪
x
⎢ ⎜
⎩ 1 + ζx
⎪
Imposing the boundary conditions (i) at t=0, qx (t)=0 and (ii) at t=0 ⎝ bi ⎠ si ⎣ x i ⎝
V u k ( ω )
dqx (t )/ dt = Sdωx finally gives ⎫⎞⎤
×sin ωx 1 − ζx 2 t − cos ωx 1 − ζx 2 t ⎬⎟⎥
⎭⎠⎥⎦
px (t ) ⎛ ⎧ 1−ζ (55)
⎜1 + e−ζxωxt ⎨
⎪
qx (t ) = 2 ⎜
x
sin ωx 1 − ζx 2 t
⎩ 1 + ζx
⎪
ωx ⎝ In Eq. (55) the subscript x to the term AF depicts translational
⎫⎞ mode of the foundation while superscript i depicts ith mode of the
− cos ωx 1 − ζx 2 t ⎬⎟ superstructure.
⎭⎠ (47) For rocking mode proceeding in identical fashion we derive
φθx ⎡⎢ pθ (t ) ⎛
Now imposing the frequency dependency soil springs on the ⎧ 1−ζ
⎜1 + e−ζθωθt ⎨
⎪
⎛ ⎧ 1−ζ
qx (t ) =
px (t )
⎜1 + e−ζxωxt ⎨
⎪
x ⎫⎞⎤
⎜ ×sin ωθ 1 − ζθ 2 t − cos ωθ 1 − ζθ 2 t ⎬⎟⎥
⎩ 1 + ζx
⎪
2 2 2 ⎝
ωx 2
(1 − rx ) + (2ζxrx ) ⎭⎠⎥⎦
⎫⎞ ⎛ ⎞ ⎡ p (t ) ⎛ ⎧ 1−ζ
× sin ωx 1 − ζx 2 t − cos ωx 1 − ζx 2 t ⎬⎟ M φ
+⎜⎜ bi ⎟⎟ θθ ⎢ ∼ θ ⎜⎜1 + e−ζθωθt ⎨
⎪
θ
⎭⎠ (48) ⎢ ⎩ 1 + ζθ
⎪
⎝ bi ⎠ si ⎣ kθ (ωi ) ⎝
V u
∼
Now considering the denominator outside the bracket as kx (ωi )- the ⎫⎞⎤
×sin ωθ 1 − ζθ 2 t − cos ωθ 1 − ζθ 2 t ⎬⎟⎥
frequency dependent stiffness of the system in uncoupled mode Eq. ⎭⎠⎥⎦ (56)
(47) can be expressed in a more condensed form as
257
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
258
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
Table 4
Values of coefficient CT for various soil profiles. 5. For G=G0(1+z/H)2
Mode → 1 2 3 ⎡ πz 3πz 5πz ⎤
u1̈ = κnβSa1⎢1.0 cos + 0.109 cos + 0.00955 cos ⎥
⎣ 2H 2H 2H ⎦ (73)
G=G0(z/H) 5.19 2.133 1.228
G=G0(z/H)0.5 4.486 1.668 1.005
The modal mass participation factor κn for first three modes are as
G=G0(z/H)2 7.826 3.301 1.66
G=G0(1+z/H) 3.094 1.095 0.66 furnished in Table 5.
G=G0(1+z/H)2 2.421 0.906 0.545
3. For G=G0(z/H)2
⎡ πz 3πz 5πz ⎤
u1̈ = κnβSa1⎢1.0 cos + 0.427 cos + 0.177 cos ⎥
⎣ 2H 2H 2H ⎦
(71)
4. For G=G0(1+z/H)
⎡ πz 3πz 5πz ⎤
u1̈ = κnβSa1⎢1.0 cos + 0.051 cos + 0.00161 cos ⎥
⎣ 2H 2H 2H ⎦
(72)
Table 5
Values of κn for first three modes.
Mode → κ1 κ2 κ3
259
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
α 2 α 2 u 4EIαe 4 1
C2 V1(α ) + C3 V2(α ) = H Kw = ∫0 φ″e (ξ )2dξ
H H H (81) H3
η
+ksDeff He ∫ e φe 2(ξ )dξ (91)
Eq. (80) and (81) can be expressed in matrix form as 0
⎡ V2(α ) V ( α ) ⎤⎧ ⎫ ⎧ uH ⎫ 4EIαe 4 ⎡ 1
⎥⎨C2 ⎬
3
⎢ = ⎨ uH ⎬ →K w = 2
⎢∫ [Ae V0(αeξ ) + BeV1(αeξ )] dξ
⎢⎣ α 2 V1(α ) α 2
V ( α ) ⎥⎦⎩ C3 ⎭ ⎩H⎭ H3 ⎣ 0
H H 2 (82)
η ⎤
+ksDeff He ∫ e [Ae V 2(αeξ ) + B e V 3(αeξ )]2 dξ ⎥
Solution of Eq. (82) finally gives 0 ⎦ (92)
⎡α 2 V (α ) ⎤ u here ηe is the ratio of embedded depth to full depth of well He/H, where
C2 = ⎢ V2(α ) − 3 ⎥ H
⎣ H H ⎦Δ (83) 0 ≤ ηe ≤ 1.0 .
Eq. (93) can be written in a simplified compact form as
and
4EIαe 4
⎡ −α 2 Kw = [I1 + ηeI2] → ksDeff Heηe 3[I1 + ηeI2]
V (α ) ⎤ u H3 (93)
C3 = ⎢ V1(α ) + 2 ⎥ H
⎣ H H ⎦Δ (84)
Eq. (93) can be expressed as
Here 4EI b
Kwb = χ
α 2 2 H3 (94)
Δ= [V2 (α ) − V1(α )V3(α )]
H (85) b 4
where χ = α (I1 + ηeI2 ) is a stiffness coefficient and the superscript b
stands for the term bending. Here
Now substituting the value of C2 and C3 in Eq. (81) we finally have 1 η
I1 = ∫ [Ae V0(αeξ ) + BeV1(αeξ )]2 dξ and I2 = ∫ e [Ae V 2(αeξ ) + B e V 3(αeξ )]2 dξ
0 0
u (z ) ⎡ α 2 V (α ) − V (α ) ⎤ ⎛ αz ⎞ are integration constants that can be computed numerically.
= ⎢ 2 3
⎥V2⎜ ⎟
uH ⎣ HΔ ⎦ ⎝H⎠
⎡ V (α ) − α 2 V (α ) ⎤ ⎛ αz ⎞ 4.4. Stiffness due to shear deformation
+⎢ 2 1
⎥V3⎜ ⎟
⎣ HΔ ⎦ ⎝H⎠ (86)
Considering the large diameter of well the well shaft can have
Eq. (96a), is a dimensionless term and is the shape function of well significant stiffness in shear deformation. Equation of equilibrium in
deformation and can be expressed as this case can be expressed as
260
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
261
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
⎡M + M MwZb + Ms(h + H ) Mw ⎤
⎢ w s
⎥
⎢ MwZb + Ms(h + H ) Jθ + MwZb 2 + Ms(h + H )2 MwZb ⎥
⎢M Mw ⎥⎦
⎣ w MwZb
⎧ x ⎫ ⎡ C 0 0 ⎤⎧ x ⎫
⎪ ⎪ ⎢ x ⎥⎪ ⎪
×⎨ θ ⎬ + ⎢ 0 Cθ 0 ⎥⎨ θ ⎬
⎪ u w ⎪ ⎢ 0 0 C ⎥⎪ ⎪
⎩ uw ⎭
⎩ ⎭ ⎣ w⎦
⎡ Kx 0 0 ⎤ x ⎧Vb ⎫
⎢ ⎥⎧⎪
⎫
⎪ ⎪ ⎪
+⎢ 0 Kθ 0 ⎥⎨ θ ⎬ = ⎨VbH + Mb⎬
⎪
⎩ u ⎪
⎭ ⎪ ⎪
⎣⎢ 0 0 Kw ⎥⎦
e w ⎩Vb ⎭ (111)
In Eqs. (109)–(111) Ms= mass of super-structure or pier, Zb is c.g.
height of base and is expressed as
1
∫0 ξφeb(ξ )dξ
Zb = 1
H
∫0 φbe(ξ ) (112)
h = Mb / Vb , the c.g. height of pier. Jθ= moment of inertia of the well
Fig. 7. Bridge pier supported on caisson foundation.
8Gr0 8Grθ 3 Gρr0 4
foundation about its c.g. Kx = 2−ν
,Kθ = ,C
3(1 − ν ) x
= 3.054 ,
2−ν
Gρrθ 8
Cθ = 1.33 Table 6
2−ν
r0 and rθ are radius of well foundation G= dynamic shear modulus Fixed based time period of the pier.
of soil ρ= mass density of soil.
Mode 1 2 3
Considering we are using the SRSS value of the superstructure
moment and shear the dominant frequency of the applied load is IRC 06 (2010) 0.256 – –
obviously the fundamental mode. Thus computing the displacement by Proposed method considering flexure only[Vide Eq. 0.264 0.014 0.0043
standard modal analysis technique cited earlier, total displacement at (14)]
Proposed Method considering combined flexure and 0.271 0.017 0.0061
c.g. height of well is given by
shear[ Vide Eq.(29)]
uw(tot ) = uw(t ) + x (t ) + Zbθ (t ) (113)
Thus amplification factor for the moment and shear in well is
expressed as
x (t ) Z θ (t )
AFw(t ) = 1 + + b
u w (t ) u w (t ) (114)
Thus while designing the section the forces in well and has to be
multiplied by the peak value AFw. Moment and shear along the well
shaft is expressed as
2EIαe 2 ⎡ ⎛α z⎞ ⎛ α z ⎞⎤
Mz = ⎢C2V0⎜ e ⎟ + C3V1⎜ e ⎟⎥ Fig. 8. Bending moment diagram of the fixed based pier IRC-06 versus proposed
H2 ⎣ ⎝H⎠ ⎝ H ⎠⎦ (115) method.
2 2 EIαe 3 ⎡ ⎛α z⎞ ⎛ α z ⎞⎤
Vz = ⎢ −C2V3⎜ e ⎟ + C3V0⎜ e ⎟⎥
H3 ⎣ ⎝H⎠ ⎝ H ⎠⎦ (116)
Fig. 9. Comparison of bending moment with and with shear deformation effect.
where
⎡ MbH2 VbH3 ⎤
⎢ 2EIα 2 V0(αe ) − V1(αe )⎥
⎣ e 2 2 EIαe3 ⎦
C2 = and
V0 2(αe ) + V1(α e )V3(αe )
⎡ M H2 3 ⎤
⎢ b V3(αe ) + VbH V0(αe )⎥
⎢⎣ 2EIαe2 2 2 EIαe3 ⎥⎦
C3 =
V0 2(αe ) + V1(α e )V3(αe )
Fig. 6. Mathematical model for kinematical interaction of semi rigid well foundation.
262
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
Solving Eq. (117) one can find out uw and us from which the nodal
forces can be computed from the expression
⎧Vw ⎫ ⎡ Kwe + kp −kp ⎤⎧ uw ⎫
⎨ ⎬=⎢ ⎥⎨ ⎬
⎩Vp ⎭ ⎣⎢ −kp kp ⎥⎦⎩ us ⎭ (118)
Fig. 13. Moment diagram of the complete pier and well in tandem.
263
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
6. Conclusion
Table 7
Deflection at pier tip.
An analytical solution is presented herein that demonstrate the
Foundation type Tip deflection limitations as prevalent with state of the art as suggested in codes.
The paper also emphasizes on the importance of considering soil
Rigid 12 mm
Semi rigid 26 mm structure interaction for such deeply embedded foundations that
significantly affects the superstructure behavior.
It also furnishes a general solution to cater to the deformation in
hereafter in Fig. 8. well foundation when depending on well geometry the well will behave
Fig. 8 clearly shows that method proposed in [1,4,5] for first mode as either a rigid, semi-rigid deformable body for which no solution
response would all underestimate the moment in pier. exists till date.
Code recommendation underestimates the moment by as much as The model can also take into cognizance of scenario analysis if
40% at the base of pier compared to proposed method. Considering liquefaction occurs at soil surface/scour line as the stiffness formula-
only first mode dynamic analysis, it underestimates the moment by tion of the well foundation considers partial embedment.
about 14%, this possibly emphasis the importance of considering Non-linear behavior of soil subjected to site-specific spectra analy-
effects of higher modes as well as inertial effect of pier, in computation sis, effect of hydrodynamic force above scour line and that for saturated
of moments and shears. soil, would highly add to our knowledge base on such foundations and
The variation in moment considering flexure only and considering is the authors' future endeavor.
both flexure and shear deformation is shown in Fig. 9.
It is observed that in this particular case as slenderness ratio of the References
pier is about 14, shear deformation affects both shear and moment
significantly. Considering only the flexural effect the dynamic moment [1] IRC:06. Standard specifications and code of practice for road bridges. Section II
is underestimated by about 17%, and shear force is underestimated by Loads and Stresses. Indian Roads Congress; New Delhi India; 2010.
[2] IRC:45. Recommendations for estimating the resistance of soil below the maximum
about 6%. scour level in the design of well foundations for bridges. Indian Roads Congress;
Considering soil structure interaction the amplification factor New Delhi India; 1992.
obtained by Eqs. (55), (56) and (57), for first three modes varying [3] IRC:78. Standard specifications and code of practice for road bridges, foundations
and substructures. Indian Roads Congress New Delhi India; 2000.
with time is shown in Fig. 10. [4] AASHTO. Standard Specification for seismic design of Highways Bridges. American
The maximum amplification factors obtained for the first three Authority of State Highway and Transport Officials. Washington USA; 1992.
modes are 1.05, 1.567 and 1.52 respectively. The predominant rocking [5] CALTRANS. Bridge design specification. California Department of Transportation
Sacramento California USA; 1991.
in second and third mode gives higher amplification in this case [6] Eurocode 8 Part 2and 4 . Design of structures for earthquake resistance-
however lower modal mass participation still makes the fundamental bridgesBelgium. Brussels; 2010.
mode finally governing. [7] Murty CVR, Jain SK. A study on seismic design codes for highway bridges. Research
report #3 Department of Civil Engineering, Indian Institute of Technology Kanpur
The variations of moments in pier considering all the methods
India; 1997.
considered herein are as shown in Fig. 11. [8] JRA part V Design Specifications of Highway bridges for seismic design in Japan.
It is evident from Fig. 12 that DSSI augments the fixed based [9] Dowrick DJ. Earthquake resistant design and risk reduction. NY, USA: John Wiley
response. In this case the amplification on SRSS value is almost two Publication; 2009.
[10] Hurty W, Rubenstiein MF. Structural dynamics. India, New Delhi, India: Prentice
times compared to code values. In Fig. 11 MDSSI case-1 is when the Hall; 1967.
well foundation is considered as rigid and case-2 depicts when well [11] Meirovitch L. Elements of vibration analysis. New Delhi, India: Allied Publishers;
foundation is considered flexible. From Fig. 11 it is apparent that 2001.
[12] Bathe KJ. Finite element analysis in engineering procedures. New Delhi: Prentice
MDSSI (case-1) > MDSSI (case-2) > MSRSS the fixed based response. Hall of India; 1980.
Flexibility in the foundation reduces the moment and shear force to [13] Chowdhury I, Dasgupta SP. Dynamic Earth Pressure on Rigid Unyielding Walls
certain extent, however the system still experience more force than a under Earthquake Forces Indian Geotechnical Journal Vol-37 No-2 April pp. 81–
93; 2007.
fixed based case which goes on to highlight the importance in [14] Shames IH, Dym CL. Energy and finite element method in structural mechanics.
considering DSSI effect for pier- well combination. New Delhi, India: New Age Publishers; 1995.
The variation of shear force in pier is as shown Fig. 12. [15] Dunkerley S. On The Whirling and Vibration of Shafts. Philosophical Transactions
of the Royal Society, London, UK 185: pp. 279–360; 1894.
In Fig. 12 it is again observed that shear force values are amplified [16] Chowdhury I, Singh JP. Do DSSI attenuate Dynamic response of building?
due to DSSI effect. Proceeding 14th International symposium on earthquake engineering Roorkee
Considering the well as a semi rigid body the complete moment and India; 2010.
[17] Chowdhury I, Tarafdar R, Ghosh A, Dasgupta SP. Dynamic response of cylindrical
shear over the length of pier and well are shown in Figs. 13 and 14.
structures with coupled soil structure interaction Bulletin of Earthquake
In Figs. (13) and (14,) z/H > 0 represents the pier part and z/H < 0 Engineering Europe. May; 2016.
represents the well foundation part. [18] IS-1893. Code of practice for earthquake resistant design of buildings. Bureau of
The kinematical interaction gives a very interesting insight to the Indian Standards. New Delhi India; 2002.
[19] Thakkar SK, Dubey RN, Singh JP. Effect of Inertia of embedded portion of well
problem. The tip deflections of the pier, considering the well behaving foundation on seismic response of bridge structure. 12th symposium on earthquake
as a rigid and semi-rigid foundation are given in Table 7.
264
I. Chowdhury et al. Soil Dynamics and Earthquake Engineering 97 (2017) 251–265
engineering, I.I.T. Roorkee, India; pp. 1055–1061; 2002. [31] Tsigginos C, Gerolymos N, Assimaki D, Gazetas G. Seismic response of bridge pier
[20] Gazetas G, Gerolymos N. Winkler model for lateral response of rigid caisson on rigid caisson foundation in soil stratum. J Earthq Eng Eng Vib 2008;7:33–44.
foundations in linear soil. J Soil Dyn Earthq Eng 2006;26:347–61. [32] Tajimi H. Dynamic analysis of a structure embedded in an elastic stratum.
[21] Mandal G, Jain SK. Effect of Non-Linearity in Pier and Well Foundation on Seismic Proceedings IVth world conference in earthquake engineering Chile; 1969.
Response of Bridges 14th World conference on earthquake engineering, Beijing, [33] Veletsos AS, Meek JW. Dynamic behavior of building foundation systems. Research
China; 2008. Report #20, Department of Civil engineering Rice University Houston USA; 1974.
[22] Chowdhury I Singh JP, Tilak R. Seismic response of well foundation with dynamic [34] Chowdhury I, Tarafdar R, Ghosh A. An analytical solution to kinematic and inertial
soil structure interaction. Proceedings 15th World Conference in Earthquake interaction of building with deep basements. , In: Proceedings of the 23rd
Engineering 2012 Lisbon Portugal; 2012. international conference in structural mechanics in reactor Technology Manchester
[23] Varun Assimaki D, Gazetas G. A simplified model for lateral response of large U.K; 2015.
diameter caisson foundations - Liner elastic formulation. J Soil Dyn Earthq Eng [35] Chowdhury I, Tarafdar R, Ghosh A, Dasgupta SP. Stiffness degradation and
2009;29:268–91. damping augmentation of soil under earthquake loading. Electronic Journal of
[24] Novak M, Beredugo YO Novak M. Coupled horizontal and rocking vibration of Geotechnical Engineering, Vol.20 Bundle9 Oklahama USA; 2015.
embedded footings. Canadian Geotechnical Journal 1972 .9 #4; 1972. [36] Waas G, Hartmann HG. Seismic analysis of pile foundations including pile soil pile
[25] Kausel E, Rosset JM. Dynamic stiffness of circular foundation. J Eng Mech Div interaction. Proceedings VIIIth world conference in earthquake engineering, San
ASCE 1975, 98(SM12) 1975:770–85. Francisco, USA; 1984.
[26] Wolf JP. Dynamic Soil Structure Interaction in Time Domain. N.Y.,USA: Prentice [37] Wolf JP, Von Arx GA. Horizontally ravelling waves in a group of piles taking pile-
Hall; 1985. soil pile interaction into account. J Earthq Eng Struct Dyn 1982;16(2):235–7.
[27] Dominguez J. Dynamic stiffness of rectangular foundations Research Report 78-20 [38] Novak M. Pile under Dynamic Loads. In: Proceedings of the 2nd international
MIT USA; 1978. conference in recent advances in geotechnical earthquake engineering St Louis USA
[28] Gazetas G. Analysis of machine foundation vibrations; State of the art. J Soil Dyn Paper # SOA14; 1991.
Earthq Eng 1983;1983(2):2–42. [39] Karnovsky I, Lebed O. Formulas for structural dynamics. N.Y., USA: McGraw-Hill
[29] Gazetas G, Dobry R. Dynamic response of arbitrarily shaped foundations. J Geotech publication; 2001.
Eng ASCE 1986 112 #GT2 1986. [40] Chowdhury I, Singh JP. Behaviour of Gravity type retaining wall with generalized
[30] Mita A, Luco JE. Dynamic response of square foundations embedded in elastic half back fill [USAMay]Journal of Earthquake Engineering. Taylor and Francis
space. J Soil Dyn Earthq Eng 1989;8(2):54–67. Publication; 2015.
265