You are on page 1of 6

Missouri University of Science and Technology

Scholars' Mine
International Conferences on Recent Advances in 1981 - First International Conference on Recent
Geotechnical Earthquake Engineering and Soil Advances in Geotechnical Earthquake Engineering
Dynamics & Soil Dynamics

Apr 26th - May 3rd

Analysis for Liquefaction: Empirical Approach
M. K. Yegian
Northeastern University, Boston, MA

B. M. Vitelli
Northeastern University, Boston, MA

Follow this and additional works at:
Part of the Geotechnical Engineering Commons

Recommended Citation
Yegian, M. K. and Vitelli, B. M., "Analysis for Liquefaction: Empirical Approach" (1981). International Conferences on Recent Advances
in Geotechnical Earthquake Engineering and Soil Dynamics. 5.

This Article - Conference proceedings is brought to you for free and open access by Scholars' Mine. It has been accepted for inclusion in International
Conferences on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics by an authorized administrator of Scholars' Mine. This
work is protected by U. S. Copyright Law. Unauthorized use including reproduction for redistribution requires the permission of the copyright holder.
For more information, please contact

Vitelli Graduate Research Assistant. Specific appli- cations of the proposed procedure are described and an example analysis is presented. In the past few years the need LPI < 1. LPI. In addition. M. Sc LPI (1) intensity. K. and c and c are constants. B. ov is the total of case histories is presented in this paper. Northeastern University. The form 1 2 173 . to describe the earthquake stress pa~ameter. Mass. which is inversely Yegian and Whitman (1978) proposed an empirical related to factor of safety against liquefac- procedure for liquefaction which is based on tion. the recent earthquakes has been apparent. but will v be published separately. which strength parameter Sc has more commonly been used by other investi- gators to describe earthquake intensity. Northeastern University. R. INTRODUCTION Throughout the past decade. The If list of case histories used by Yegian and Whit- man to develop the liquefaction criteria they proposed was that which was published earlier LPI > 1. with some additional data points. The method is based on an expanded list of case histories of liquefaction and no-lique- faction and employs earthquake magnitude and hypocentral distance to describe the intensity of shak- ing at a site. is not presented in this paper. Boston. various laboratory testing procedures have been developed to supply soil parameters needed for analytical techniques that predict liquefaction potential at a site. SYNOPSIS An improved analytical method to calculate the likelihood of soil liquefaction is de- scribed. and hypo- central distance. A simple analytical technique for liquefaction where M is the earthquake magnitude. The paper discusses the merits of using M and R as opposed to acceleration. LPI. parameter LPI was employed and an expression thors of this paper completed such an overall for it was developed as follows: the stress review of all the case histories discussed and parameter was assumed to have the form presented by the original investigators. liquefaction is not likely to occur for re-evaluation and expansion of the currently used list to include data obtained from more In the investigations reported herein. Deterministic and probabilistic understanding of liquefaction have evolved from applications of the method are discussed and an extensive laboratory studies. which includes a total of about 322 data points corre. Yegian Associate Professor of Civil Engineering. The new list of case histories is compiled from a complete re-evaluation of previ- ously published case histories and field data observed in more recent earthquakes. av is the effective vertical A new criterion for liquefaction is proposed which employs earthquake maqnitude and distance stress. Considerable gains toward an condition. The au. liquefaction is likely to occur by Seed and Idriss (1971). (2) sponding to 80 locations and 19 different earth- 0 quakes. vertical stress.Analysis for Liquefaction: Empirical Approach M. for the evaluation of the liquefaction potential at a site. penetration test results to represent the soil merous studies. can be expressed as: interpretation of field data using earthquake parameters such as magnitude. Data from field observa. Mass. inary investigations of liquefaction. R is the based on field data compiled in the new list hypocentral distance in KM. Due to space limitations this updated list. liquefaction due to to describe the seismic event. the parameter Liquefaction Potential Ind8x. and standard earthquake shaking has been the subject of nu. ANALYSIS FOR LIQUEFACTION tions of liquefaction have been utilized to study the phenomenon as it occurs in the field Yegian and Whitman (1978) proposed the use of and to develop empirical procedures for prelim. M. Boston. example analysis is included.

corrected for the overburden pres. 035 + (0. c . Assuming that the earthquake parameters are sure as suggested by Seed (1976) 'given'. (SPT). the equation for the mean value of LPI u = (9) can he written as: 0 lnLPI -0 4 where. The rest of the terms in Eq. and the depth of the water table respectively. 6 yields the mean value of LPI using mean or 'expected' sc ( 3) values of the parameters which define LPI. an analysis of liquefaction potential involves many uncer- tainties.1. VLPI' is given by: N N (1. 174 for the strength parameter was assumed to be: these uncertainties in the analysis are essen- tial for a realistic assessment of the likeli- hood of liquefaction.dw (7) 0 v 0 v LPI ( 5) 0 where Var. 6 is very similar to that of peak ground acceleration. Thus.25log 0v(tsf)) ( 4) c I where N is the SPT recorded in the field.50. of the constants c were assumed and the differ- ence between lnS and lnS was computed for the lower value corresponds to the uncertainty c c only in the liquefaction analysis procedure each case history. 4 shown that measured ground acceleration is log- 0. Quantification and incorporation of . liquefaction is 'ex. the coefficient of variation of LPI. it is necessary to compute the coeffi- cient of variation of LPI in order to make pre- where Nc is the standard penetration test val- dictions of the probability of liquefaction. otherwise i t was discarded. LPI is also assumed c3 to be lognormally distributed. 6.16) Va\t + I 1 + bining Eqs. Eq. ue. P [ LIQ. iterative approach was employed in which values Typical values of VLPI range from 0.N. The conditional 0. 6). can provide an estimate The best estimate of the constants c were of the conditional probability of liquefaction. LPI can be expressed as: N L Yw 2 + (=--) Var. total unit weight. lnLPI pected' to occur. In. Com. However.464 normally distributed. case was positive and the case was a 'no' liquefaction or negative and the case was a 'yes' liquefaction. Donovan (]973) has c2 =-0. Thus. The constant The values of the constants c . scribe the uncertainties in the soil parameters linear multi-regression analysis following an used in the liquefaction analysis for any site. 2 and 3. Var. Yegian and Whitman (1978) suggested the log- normal distribution for LPI. 7 de- tion and no-liquefaction case histories.4 probability of liquefaction is then determined c4 by computing first the standardized variable U. c and c 1 2 3 4 term in Eq.2 to 0. J (8) pretation technique and the data. The form of LPI cl = 0. The minimum of the sum of the squares is a measure of the uncertainty in both the inter. LPI ( 6) 0.dw are the variances v of the field blow counts. addition. then this difference was The use of VLPI' together with the mean value of squared and saved.2 as given in Eq. l. 6).y and Var. 7 is due to the uncertainty in the were evaluated using simulation techniques and c parameters which define the equation for LPI the 322 data points corresponding to liquefac- (Eq. 0. The result of these investigations shows that the best Evaluation of P [LIQJ requires an assumption estimates of the values of the constants are regarding the probabllity density function. evaluated by minimizing the sum of these defined as squared differences for the entire case history list. If the difference for a which is proposed herein (Eq. A non. and yw is the unit weight of water. mlnLPI Hence.464 N°• 4 0 c v The above equation for LPI can be used for a and particular site to deterministically evaluate the liquefaction potential: when the computed 2 2 0 ln (VLPI + 1) LPI is greater than 1. LPI computed from Eq. e 0 · 2 M(R + 25) .

7. and are com- The procedure proposed above differs from the plicated and expensive to apply. sands.crv the laboratory strength data when plotted on 0. response with normalized laboratory cyclic mined in a more refined and rigorous manner. and on the laboratory pore the conditional probability of liquefaction can pressure data on cyclically-loaded specimens of be determined. to 0. The model employs LPI to define a Furthermore. this procedure predicts larger pressure response. but does not indicate inary study of liquefaction for selected design the level of excess pore pressure that might seismic events. levels causing liquefaction may be of such mag- nitude as to reduce the effective stresses in the soil to levels consequential to the dynamic response of the deposit. still be generated during that particular event. The parameter B is the slope of LPI = e { R + 25} . propagation theory. Based on this review. with an average value of 0. Values of B \vere estimated from Fig. The liquefaction analysis procedure described in this paper can be used to develop such a model for excess pore pressure prediction as a func- tion of earthquake magnitude and distance. Definition of possible ranges for these parameters was attempted on the basis of published laboratory strength and pore pres- sure data. saturated herein is based on the new.2-0. This plot can LPI less than 1 may indicate that the sand is be used i~ enginePring practice as a prelim.2M -0. 7). ru.0 (13) where.4 0.0 1. can the sand under study has liquefied. not the pore pressure response is greater than 1 for a given seismic event. Fig.10 and 0. ru' can be related to LPI as: 1 2 r u 1T arc sin (LPI) 2aS LPI < 1. The pore liquefaction. and to the settlement of a structure founded on the deposit.0 number of cycles causing the parameter. behavior curves.0.0 describes a determinis.25. but yields smaller coefficients of vari- ation due to the larger number of data points used. 1 shows results of a tion involves the determination of whether or probability calculation made assuming the typ. not likely to liquefy.5 and 1. Note that the curve corres. a computed ical range 0. the parameters and the constants threshold event causing 100% pore pressure defining LPI in this investigation were deter.4 be equal to 1. Such increases in excess pore pressure below tic analysis of liquefaction. ponding to VLPI = 0. a pore pres- sure response of 100% (ru = 1) indicates that As described above. pore pressure generated during events smaller In comparison to the previous criterion for than the event causing liquefaction. for a between 0. application of the one-dimensional shear wave .50 for VLPI. A similar study of published data on Pore Pressure Prediction excess pore pressure plotted against the normal- Procedures currently used for estimation of ized number of cycles yielded a range of values pore pressures in sands are based on the earth. a is a curve-fitting parameter used to relate laboratory pore pressure data for a par- ticular soil to the ratio of the number of 0 equivalent cycles of stress application to the 0 0. 1 Conditional Probability of Liquefaction laboratory strength curves suggested by various versus LPI investigators for different types of tests and sands. L'lu r (12) u a v APPLICATIONS in which L'lu is the excess pore water pressure Liquefaction Analysis for level ground conditions. ru' is defined as: LPis. Eq. Such procedures for pore pressure pre- diction involve many uncertainties. in order to predict the excess thus increasing the reliability of the model. expanded data bank. In engi- neering practice.464 Nc cry log-log paper. Yegian (1980) showed that the pore pressure response parameter. values of B ranged between 0. 6 together with the coefficient of variation of LPI (Eq. 175 Using the computed u and the normal tables.19.5 1. sands. common analysis of liquefac- be used to predict the conditional probability of liquefaction. Yegian (1980) previous model developed by Yegian and Whitman proposed an empirically developed model for the (1978) in that the liquefaction criterion used prediction of pore pressures in loose. Thus. Thus.5 2. Seed and Booker quake-induced shear stresses obtained from the (1977) recommended a typical value of 0.

2 / / / Liquefaction risk analysis can provide informa- "'"'w o. 9 TCM the seismic source data and an attenuation law relating the seismic parameter of interest to earthquake magnitude and distance.464 N (} c v -----UPPER 8 LOWER BOUNDS 0 9 --AVERAGE I~ . 2 Excess Pore Pressure Ratio versus Liquefaction Potential Index.8 (factor of safety of 1. o. LPI k e 2 (R + 25) 3 (15) 1 (T) STRENGTH where FACTOR OF SAFETY= (T) EARTHQUAKE (} 0 0 1 v (\J kl 0. Liquefaction Risk Analysis M 5.4 LPI • e (R. The attenuation law specified in seismic hazard analysis is usually .2M -0.5 0. the output is the annual w / a: / probability of liquefaction. 0..5 Probabilistic seismic hazard analysis has received increased attention in the past de. plots of pore pressure response versus earthquake magnitude. k . of the form.1 0.2 0. The analysis can also be used to study the degrees of influ- 0.4 3 i= 0 6 <( a: Using these parameters to define the attenuation w 0.8 ::::> <l 0 7 0 and k =-0.8 0. R = 50 km cade. 6). 2. (14) A erated as shown in Fig. distance and soil strength are gen. LPI.. k and k are constants. gether with the equation for LPI (Eq. For a given sand deposit.25) ence of the various parameters involved to identify the major factors contributing to 0.f.0 in the analysis. same form as the attenuation law used in seismic tion of pore pressures for preliminary studies hazard analysis. 176 Using the ranges for B and a given above. a similar way liquefaction risk analysis can be performed since the equation for LPI has the Thus the model described enables quick evalua. 0 3 / as 1. 0 c. 2 demonstrates that while an LPI of 0.5 / a: :> law in the computer program one can compute the / annual probability of LPI exceeding a certain "'"'w 04 / / value... EXAMPLE Fig.. 1 2 3 event.4 0.7 0.4 likelihood of liquefaction occurring at a site.464 N~.9 1.. LPI and provides an opportunity to combine future will be given by field data and laboratory data on pore pres- sures in a simple.4 10 0. In pore pressure response of up to 50%. Fig.3 0.0 risk to a constructed facility. The input information to such an analysis includes y = I.25) may imply The output of the analysis is the annual proba- safety against liquefaction.. 3 Example Site evaluate an overall annual probability of liquefaction at a site. This plot can be used in preliminary studies to determine ex- pected buildup of excess pore pressure in a particular sand deposit during a given seismic where. Computer programs are now available to compute annual probability of a certain seismic N = 10 blows/ ft parameter exceeding a specified value. logical and consistent k M k manner. If the value to be exceeded is assigned a: c. to. / tion enabling comparisons between the various u X w 0 factors contributing to the overall seismic 0 0. The details of such an analysis are beyond the scope of this paper and are reviewed by Yegian (1979) These computer programs can also be used to Fig. there may be a bility of A exceeding a certain value 'a'.6 0.

3 shows the pertinent data for the example REFERENCES site studied herein..39 Pore Pressure Prediction in Sands".y 0. "Evaluation of Soil Lique- age value of the LPI is computed from Eq.5 and distance 50 km. N. (1978). (1979). u. (1971). ru' for level GT7. 7 tion Potential". Considering a seismic event of magnitude 5. "Simpli- The coefficient of variation of LPI is computed fied . 93. (1973). and Booker.. Rome. Application of the method to evaluate liquefac- tion potential and to estimate earthquake- generated excess pore pressures are discussed. American Society of Civil Engineers. An empirically developed criterion for liquefaction is presented which is based on a new. I.035 + (0. M.04 10 4. K. Seed. Assuming the following variance data quakes. Earthquake Engineering.Procedure for Evaluating Soil Liquefac- from Eq. Division. 6 to faction Effects On Level Ground During Earth- be 0. and Idriss. burden is 1. July. September. H. the probability of liquefaction Yegian. Vol. Army Engineer Waterways Exper- imental Station. The exact estimation of ru for this site will require the evaluation of these parameters from laboratory tests on samples of the sand under study.B. (1977). Journal of the Geotechnical though liquefaction is "not expected" to occur. expanded list of case histories. .. Engineering Division. VLPI = 0. c. July. B. the pore pressure response.28.9 tern. "A Statistical Evalua- 10 blows/ft." State-of-the-Art Paper. Report 13. Pennsylvania. H. "Risk of the sand layer given the postulated seismic Analysis for Earthquake-Induced Ground Failure event. CONCLUSIONS A simple analytical method to evaluate the likelihood of liquefaction at a site is des- cribed. Pro- about 14 blows/ft. American Society of Civil Engineers. even by Liquefaction".5)2)0.85 and quake Engineering.4 using average values of a and B parameters described earlier in this paper. the aver. Seed. "Stabil- ization of Potentially Liquefiable Sand Var. using Eq. American for the site parameters: Society of Civil Engineers convention.. B. American Society of Civil Engineers. employing Fig. H. "Probabilistic Seismic Hazard Analysis". is about 0.. The water table is at the ceedings of the Fifth World Conference on ground surface and the unit weight of the over. (1976). GT7..32. Phil- adelphia.04 Deposit Using Gravel Drains". K.92 Yegian. Journal of the Geotechnical. M. September. The average SPT-value for the two meter sand layer shown in the figure is Donovan. May. M. K. During this event.5 1. Var. R. J.16)~ + (1 + (9. ground condition is estimated to be 0. An example study is included. Proceed- ings of the Seventh World Conference on Earth- Thus. "Empirical Procedure for VLPI = 0.N 4 Seed. 0.. Turkey. (1980). State-of-the-Art for Assess- ing Earthquake Hazards in the United States. 177 Fig. The SPT-value corrected for the tion of Strong Motion Data Including the Feb- effect of the overburden stress. 1 with LPI = 0. Yegian. 4 is ruary 9.s. M. and Whitman. 1971 San Fernando Earthquake". V.85. R. SM3. Journal of the Soil Mechanics and Foundations Division.