You are on page 1of 10

Chapter 9 GT103

Chapter 9 DESIGN OF EARTH-RETAINING STRUCTURES ANDSLOPES FOR EARTHQUAKES


Debasis Roy
Department of Civil Engineering, IIT, Kharagpur.

1. INTRODUCTION Case histories abound the literature reporting failure of retaining walls and slopes due to earthquakes, one of the most publicized among which is the failure of waterfront structures of the Port of Kobe during the 1994 Hyogo-ken Nambu Earthquake of Japan (Figure 1a) and the failure of the upstream slope of Lower San Fernando Dam during the 1971 San Fernando Earthquake (Figure 1b).

Figure 1. Seismic failure of earth-retaining structure and earth-dam For satisfactory performance during an earthquake, earth-retaining structures and soil-slopes must withstand cyclic inertial load because of rapid cyclic nature of the ground motion in addition to other permanent and incidental loads. The performance earth-retaining structures and soil-slopes may be affected additionally by earthquake-related reductions in the shear strength of the backfill soil or those within the slope as well as the underlying foundation layers. 2. DESIGN OF EARTH-RETAINING STRUCTURES FOR EARTHQUAKES 2.1. Rigid, Yielding Retaining Wall

Rigid-yielding retaining walls are usually designed using a procedure that is essentially an extension of the Coulomb earth-pressure theory that incorporates the inertial forces due to a constant-amplitude ground-motion while assessing the stability of the active wedge (Okabe 1926, Mononobe and Matsuo 1929). The procedure is commonly referred to as the Mononobe-Okabe (M-O) method. The procedure involves inclusion of the inertial forces due to a constant-amplitude ground-motion represented by the horizontal and vertical seismic coefficients, kh and kv, respectively. The vertical and the horizontal seismic coefficients are essentially representative values of the horizontal and

NPCBEERM, MHA (DM)

Chapter 9 GT103

vertical ground acceleration within the active and passive wedges divided by the acceleration due to gravity, g, respectively (Figure 2).

Figure 2. Forces on the active and passive wedges in M-O analysis

2.1.1.

Active Thrust

Stability considerations for the active wedge lead to the following expression for the active thrust during an earthquake:

PAE = 0.5K AE H 1 (1 k v )
2

(1)

The active earth pressure coefficient for earthquake loading, KAE, is calculated from:

K AE =

cos 2 ( 1 ) sin ( + )sin ( 1 ) 2 cos cos 1 cos( + 1 + )1 + cos( + 1 + ) cos(1 1 )


2

(2)

where

, = tan 1 {k h (1 k v )}. For meanings of other symbols refer to Figure 2. It is

evident from Equation 2 that earthquake in effect increases the magnitude of the active thrust compared with its static value. This procedure has been modified by Seed and Whitman (1970) to account for the fact that earthquake-related earth pressure acts at an elevation higher than a third of wall-height above wall-base as implied in Equation 1. The incremental contribution of the earthquake load to active thrust over the its static value, PAE , using:

PAE = PAE PA
where PA is Coulombs active thrust for static condition. The height at which

(3)

PAE acts is given by:


(4)

h = [PA H 1 3 + PAE 0.6 H 1 ] PAE

NPCBEERM, MHA (DM)

Chapter 9 GT103

2.1.2.

Passive Thrust

Passive earth pressure for rigid, yielding retaining wall can be similarly estimated using:

PPE = 0.5K PE H 2 (1 k v )
2

(5)

The passive earth pressure coefficient for earthquake loading, KPE, is calculated from:

K PE =

cos 2 ( + 2 ) sin ( + )sin ( + 2 ) 2 cos cos 2 cos( 2 + )1 cos( 2 + ) cos( 2 2 )


2

(6)

The meanings of the symbols used in Equation 6 are explained in Figure 2. It is evident from Equation 6 that an earthquake in effect reduces the magnitude of the passive thrust compared with its static value. Towhata and Islam (1987) proposed an approach for finding the line of action of PPE involving partitioning similar to that in Equation 3. 2.1.3. Saturated Backfill

If the backfill is saturated, in drained analyses the modified M-O procedure is used for the soil skeleton after setting = in Equations 1 and 5 as appropriate and adding the thrusts due to the hydrostatic and hydrodynamic pressures to the contribution of the soil skeleton. The hydrostatic pressure has a triangular distribution. Consequently, line of action of the thrust due to the hydrostatic pressure is at a third of the height of the retaining wall above wall base. For retaining walls with near vertical face, thrust due to hydrodynamic pressure, Phd is usually approximated by (Zangar 1952):

p hd = C (a max g ) w h
where

(7)

is the unit weight of water and C is a coefficient obtained from Figure 3.

Figure 3. Hydrodynamic coefficient, C 2.1.4. Limitations

Except that the modified M-O procedure does not predict the phase of the cyclic thrust force for flexible walls, the modified M-O procedure has been found to compare very well with the results of more realistic analyses (e.g., Steedman and Zang 1990) and model and finite element study (Whitman 1994).

NPCBEERM, MHA (DM)

Chapter 9 GT103

2.1.5.

Seismic Coefficients

Seismic coefficients should reflect the average earthquake acceleration level within the active or passive wedge. The choice of the seismic coefficients should therefore be such that it can reproduce reasonably the effects of ground motion whose amplitude is likely to vary with height measured from the base of the wall. The seismic coefficients should also reproduce reasonably the effects of a ground motion with time-varying amplitude. No rigorous, analytically-based guideline is available for estimating the seismic coefficients. Choice of these coefficients is mainly based on local experience. A horizontal seismic coefficient, kh, of 0.5 times the peak horizontal acceleration at the top of the retaining wall is often considered satisfactory (Hynes-Griffin and Franklin 1984). Considering two-fold amplification of ground motion between the base and the top of the retaining wall, k h = a max g is sometimes considered appropriate, where amax is the peak horizontal ground acceleration (PHGA) at the base of the retaining wall. Seed and Whitman (1970) also found that non-inclusion of kv in the calculation usually does not introduce an error of more than 10% in the magnitude of the active thrust. As a result, kv is often ignored in earthquake design of retaining structures. 2.2. Rigid, Non-yielding Retaining Wall

Massive gravity walls or basements designed as braced concrete boxes are unlikely to yield sufficiently to mobilize the shear strength of the soil supported by it. In such cases, active and passive pressures would not develop. Following elastic solutions, provided by Woods (1973), can be used for estimating the thrust, PE , and the overturning moment about the base of the wall, M E for designing non-yielding walls:

PE = H 2 k h F p
M E = H 3 k h Fm

(8) (9)

where H is the height of the wall and the backfill. These equations can be used if the wall is smooth, the backfill is without any surface slope, frequencies of the input motion is less than one-half of the fundamental frequency of the backfill soil column, Vs (4 H ) , where Vs is the velocity. For backfill extending horizontally to a distance of 4H or more from the back of the wall, dimensionless thrust factors, Fp and Fm, vary with the Poissons Ratio of backfill soil as shown on Figure 4. For L < 4 H , Fp and Fm are approximately proportional to L/H, where L is the horizontal distance to which the backfill extends measured from the back of the wall. 2.3. Softening of Backfill and Foundation Soils

Cyclic loading during an earthquake causes many types of soils to soften. The soils considered susceptible to a significant loss of shear strength during earthquakes include liquefiable loose cohesionless deposits, sensitive clayey soils, organic soils such as peat and loess. The soil properties used in limit equilibrium slope stability assessment should reflect the potential for reduction of shear strength during earthquakes. A conservative approach for checking the stability of an earth-retaining structure for liquefaction or softening is to use residual shear strength for the soils susceptible to softening and check the undrained stability of the wall using the limit equilibrium method such as those used in slope stability assessment or bearing capacity estimation.

NPCBEERM, MHA (DM)

Chapter 9 GT103

Figure 4. Dimensionless thrust factors for

L/ H 4

The approach is illustrated with an anchored sheet pile wall shown on Figure 5 underlain by a thin liquefiable layer. A possible mechanism for instability in this case is surface ABCD, segment BC of which is likely to provide little resistance against earthquake-related inertial forces. A limit equilibrium slope stability analysis computer program, e.g., XSTABL (Interactive Software Designs, Inc. 1994), is often used to check the overall stability involving a deep-seated failure such as that illustrated on Figure 5.

Figure 5. Overall stability of retaining walls underlain by liquefied soils

NPCBEERM, MHA (DM)

Chapter 9 GT103

2.4.

Factor of Safety

The design procedure for retaining walls is largely the same as the static design procedures except that thrusts due to an earthquake need to be estimated and a reduced material strength need to be used if appropriate in stability assessment as described in the preceding sub-sections. In design of retaining walls for earthquake a factor of safety of 1.1 to 1.2 is usually considered satisfactory.

3. DESIGN OF SOIL SLOPES FOR EARTHQUAKES Compared with static slope stability under static condition, stability of slope during earthquakes is influenced by following two additional factors: Inertial stresses induced by the ground motion and Earthquake-related softening of the stress-stain behavior of materials that the slope is comprised of as well as those below the base of the slope.

Limit equilibrium procedures modified to accommodate inertial effects are usually employed in assessing the stability of slopes during earthquakes. Two such methods, the pseudostatic method and the sliding-block method, are discussed in the following sub-sections. The more rigorous method of dynamic analysis is also discussed later in brief. 3.1. The Pseudostatic Method

In the pseudostatic method, the effects of an earthquake are represented by constant horizontal and/or vertical accelerations similar to the M-O procedure discussed earlier. In this method the limit-equilibrium factor of safety is computed using an appropriate static method, e.g., the Modified Bishop or Janbu methods, applying the seismic load in proportion of the total weight of the potential failure block consisting of soils above the slip surface (Figure 6). The constants of proportionality in the horizontal and vertical directions, i.e., the horizontal and vertical seismic coefficients, kh and kv, are are chosen largely based on experience as discussed earlier. The similarity between seismic and static limit equilibrium slope stability assessment is apparent from the information presented in Figure 6 except that in assessment of seismic slope stability, the potentially unstable mass (Slice n: Figure 6) is subjected to inertial forces due to earthquake ground motion (khWn, kvWn: is the weight of Slice n) in addition to usual static forces (inter-slice forces, En, En+1, Xn, Xn+1; weight, Wn; shearing resistance, Sn; and normal force at slice base, Nn). Another major difference between limit equilibrium slope stability assessments for static and earthquake conditions is that earthquake loading is usually too fast for pore water pressure dissipation. As a result, most soils behave in an undrained manner during earthquakes irrespective of their grain size distribution. Therefore stability assessment of slopes for earthquake loads usually involves estimation of the factor of safety using undrained soil properties in terms of total stress. The third major difference between limit equilibrium slope stability assessments for static and earthquake conditions is that cyclic loading during an earthquake may cause many types of soils to soften as discussed earlier. The soil properties used in limit equilibrium slope stability assessment should reflect the potential for reduction of shear strength during earthquakes.

NPCBEERM, MHA (DM)

Chapter 9 GT103

Figure 6. Limit equilibrium slope stability assessment for circular slip surface

Olson and Stark (2003) provide the following guidelines for estimating undrained shear strength, su, of cohesionless soils from in-situ tests:

0 = 0.205 + 0.0143 q c1 su v 0 = 0.205 + 0.0075 (N 1 )60 su v


The corresponding estimates of post-liquefaction shear strength are:

(10) (11)

0 = 0.03 + 0.0143 qc1 su v 0 = 0.03 + 0.0075 ( N 1 )60 su v


In Equations 11 to 14, symbol

(12) (13) represents the pre-earthquake value of the effective vertical stress,

qc1 is the stress-normalized value of the cone tip resistance in MPa, ( N1 )60 is the stress-normalized

0 v

value of the SPT blow count corrected for input energy. The stress-normalized value of the cone tip resistance is calculated using:

0 ) qc1 = qc (Pa v

(14)

where qc is the cone tip resistance, Pa is the atmospheric pressure and n is an exponent whose value is 0.5 for sands and 1.0 for clayey soils. The following equation is used for stress-normalization and energy correction of SPT blow count:

(N1 )60 = N (Pa v0 )0.5 ER

(15)

where N is the observed SPT blow count and ER is the energy input divided by 60 % of theoretical input energy. For cohesive soils, peak and residual undrained shear strengths are usually estimated from Field Vane Shear Test (FVST). The pseudostatic approach is traditionally not used for slopes containing and/or underlain by soils susceptible to softening in excess of 20 %, although a conservative estimate of the factor of safety may be obtained if residual or post-liquefaction shear strength is used for soil units that are likely to soften or liquefy during the design earthquake. The conventional pseudostatic slope stability assessment involves estimation of the limit equilibrium factor of safety considering appropriate NPCBEERM, MHA (DM)

Chapter 9 GT103

seismic coefficients with 20 % reduction in soil shear strength. Slopes are considered stable if a limit equilibrium factor of safety of 1.1 or more is obtained from these calculations. The horizontal seismic coefficient is usually assumed to be about one-half of the estimated PHGA at the crest of the slope divided by g. Assuming two-fold amplification between the base and the crest of the slope, the horizontal seismic coefficient becomes equal to the PHGA at the base of the slope divided by g. The vertical seismic coefficient is usually assumed to be zero in pseudostatic analysis. 3.2. The Sliding-block Method

The sliding-block method approximates the potential sliding mass as a rigid block on a rigid, sloping base undergoing earthquake ground motion. The contact between the block and the sloping base is treated as rigid plastic. The procedure essentially involves estimation of the yield acceleration, ay, from pseudostatic analysis. The yield acceleration is defined as the horizontal seismic coefficient multiplied by g that makes the slope to be marginally stable. The ratio of the yield acceleration to the PHGA at the crest of the slope has been related to the deformation of the slope for an easy application of this procedure (Makidisi and Seed 1978, Hynes-Griffin and Franklin 1984). One such relationship is presented in Figure 7. If a deformation of 1 m or less is estimated for the computed yield acceleration using the upper bound relationship, the slope is interpreted to remain stable after the design earthquake.

Figure 7. Permanent deformation as a function of yield acceleration (Hynes-Griffin and Franklin 1984)

3.3.

Example Illustrating the Sliding-block Procedure

This example illustrates estimation of deformation using an embankment shown on Figure 8a for a Magnitude 7 earthquake with a peak horizontal ground acceleration of 0.3g. The foundation condition includes a non-liquefiable crust of clayey silt extending to 1.5 m below ground surface underlain by a 1.5-m thick liquefiable layer and a non-liquefiable soils. The embankment is also constructed with well compacted granular soils not expected to liquefy during the design earthquake. Using post-liquefaction undrained shear strength of the liquefied layer and the values of peak undrained shear strength of non-liquefied soils estimated from Equations 10 to 13 a y = 0.05 g (Figure 8b). Computer program XSTABL Version 5.20 (Interactive Software Designs, Inc. 1994) was used in the limit equilibrium slope stability computations. Hence, a y (PHGA ) = 0.16 , for which the upper-bound estimate of potential deformation becomes 0.95 m (from Figure 7). Estimated deformation of less than 1 m from a sliding-block analysis is usually considered acceptable.

NPCBEERM, MHA (DM)

Chapter 9 GT103

Figure 8. Example illustrating the Sliding-block Procedure

3.4.

Dynamic Analysis

Rigorous dynamic analysis of slopes is sometimes undertaken for critical structures or in situations where adoption of the simplified pseudostatic or the sliding-block approach is inappropriate. The procedure for dynamic analysis may be based fully-coupled, pseudo-coupled or uncoupled. Also, the analysis may be carried out in terms of total or effective stresses. Dynamic analysis usually requires setting up of an elaborate numerical model usually based on finite-element or finite-difference. It requires a reasonable numerical reproduction of the anticipated material behavior through a reasonable stress-strain model and adequate material testing allowing proper calibration of the stress-strain model. Additionally, a suite of earthquake time histories are also needed to assess the earthquake behavior of the model. Such an elaborate undertaking is usually too expensive to be feasible except in very large or critical projects.

NPCBEERM, MHA (DM)

Chapter 9 GT103

REFERENCES Hynes-Griffin, M.E. and Franklin, A.G. 1984. Rationalizing the seismic coefficient method. Miscellaneous Paper GL-84-13, US Army Corps of Engineers Waterways Experiment Station, Vicksburg, Mississippi, USA. Interactive Software Designs, Inc. 1994. XSTABL an integral slope stability analysis program for personal computers, Version 5. Reference Manual. Moscow, Idaho 83843, USA. Makidisi, F.I., and Seed, H.B. 1978. Simplified procedure for estimating dam and embankment earthquake-induced deformations. Journal of Geotechnical Engineering. ASCE. 104(7): 849-867. Mononobe, N. and Matsuo, H. 1929. On the determination of earth pressures during earthquakes. Proceedings, World Engineering Conference. 9. Okabe, S. 1926. General theory of earth pressures. Engineering. 12(1). Journal of the Japan Society of Civil

Olson, S.M. and Stark, T.D. 2003. Yield strength ratio and liquefaction analysis of slopes and embankments. Journal of Geotechnical and Geoenvironmental Engineering. ASCE. 129(8): 727-737. Seed, H.B. and Whitman, R.V. 1970. Design of earth retaining structures for dynamic loads. Proceedings, ASCE Specialty Conference on Lateral Stresses in the Ground and Design of Earth Retaining Structures. 103-147. Steedman, R.S. and Zang, X. 1990. The seismic response of waterfront retaining walls. Proceedings, ASCE Specialty Conference on Design and Performance of Earth Retaining Structures. STP 25. 872-886. Towhata, I. and Islam, S. 1987. Prediction of lateral movement of anchored bulkheads induced by seismic liquefaction. Soils and Foundations. 27(4): 137-147. Whitman, R.V. 1994. Earth pressure under dynamic load. Seminar: November 8, 1994. University of British Columbia, Vancouver, British Columbia, Canada. Woods, J. 1973. Earthquake-induced soil pressures on structures. Report EERL 73-05. California Institute of Technology, Pasadena, California, USA. Zangar, C.N. 1952. Hydrodynamic Pressures on Dams due to Horizontal Earthquake Effects. USBR Engineering Monograph No. 11.

NPCBEERM, MHA (DM)

10

You might also like