Proceedings of Ninth World Conference’on Earthquake Engineering [August 2-9, 1988, Tokyo-Kyoto, JAPAN (Voll) 415 OVERCONSOLIDATED MEXICO CITY CLAY ‘UNDER CYCLIC LOADING J. Abraham DIAZ-RODRIGUEZ Professor of Civil Engineering, Universidad Nacional Autonoma de Mexico, MEXICO suomaRy This paper describes two series of undrained cyclic triaxial teste on undi turbed soil samples of Mexico City clay. The tests were conducted to determined theundvained shear strength and stress-strain characteristics of normal consoli- dated and overconsolidated of Mexico City clay. Sqpirieal relationships about equivalent Young's aodulus, equivalent damping ratio and cyclic stress level asso chated to cyclic strain level, vere obtained. OF particular concern uas the pos sible loss in undrained shear strength caused by cyclic loadings. ‘DNTRODUCTION The intense ground shaking during the 19 Septesber 1985 Mexico City earthquake, magnitude 8.1 (Ma) and intensity 9 in parts of the city, caused many Foundation to undergo excessive settlenent and tilting, resulting in either col~ Lapse or substantial danage of superstructures (Ref. 1). It is believed that the characteristics of soil in Mexico City played a key role in these disastrous events; of particular concern is the possible loss in undrained shear strength caused by cyelie loading. Previous studies had shoved that the strength of cohesive soils under monot— onous loading conditions increased as the speed of loading increases. Other stu~ dies consistently revealed that the behavior of cohesive soils subjected to re~ peated cycles of loading and unloading way differ considerably from their behavior Guting a single loading cycle, It is therefore useful to etudy the effect of re- peated loading under unérained conditions in the laboratory. Tn practice, an earthquake may produce oscillations of buildings and their foundations. In’ simple way, a soil elenent below a foundation may be consid~ ered to be subjected to an initial sustained strees together with euperimosed vertical eyelie stresses. The strength of the soil under those conditions difters from that determined by a normal type of compression test. Thus it vas decided to conduct laboratory studies by cubjecting each specinen to a sustained stress, allowing the sample to reach equilibrium and then applying a series of transient stress applications representative of an earthquake. SOIL DESCRIPTION ‘The undisturbed soil samples used in the present investigation were obtained J. As DIAZ-RODRIGUEZ m-29with shelby tubes from 11.30 m depth, The soil was the following average proper- ties: Specific gravity, 2.24; natural water content, 366 25 Liquid limit, 4563 plastic linit, 92 2; votd ratio, 8.76 and unit weight, 1-10 t/m3. ‘TESTING PROGRAM, ‘The olution co several practical problems require a knowledge of the strain induced by cyclic loading and the peak strength of the soil after a Linited number of repeated loading cycles. Thus an specific testing program was followed, (Bef. 2). ‘The final trimmed size of the test specimens was 39 am in diameter and 95 sm im height. No side drains were used and regular end platens with porous stone at top and botton of the specimen vere used. Two series of undrained cyclic triaxial tests were conducted, the details are outlined in Table 1. TABLE 1. Summary of Cyclic Loading Program qeitiai_ | fina | Single | cyette ‘test consolidation —] stress | stress stress ssplieude | ‘ratio Jrsber | —F 7 Sea | ce ne Now | ween? | keen? | refea” [2 <8 y-|42 ots | 7 3g Syl a3 2 3 ons | Ee Balas bs e 0:25 29 Sy 2 gl as 0 S 0:32 | 38 FS? a6 0:43 50 aby [az [ous 7 16 ga Bs] a3 : 2 0.23 20 an pues bd Ey 0.25 B S'yogl as S S 032 28 #55) a6 0:40 35 ‘Test series A consisted of one static test and five cyclic tests, Fach sam ple was isotropically consolidated to an effective stress equivalent to the ex- ieting effective overburden pressure of 0.43 Ke/en? using a back pressure of 1.7 Rg/en?. Test series B consisted of one static test and five cyclic tests. Overcon— solidated specinene were obtained by Firet consolidating the specimens to a maxi~ mim stress of 0.86 Kz/en? sing a back pressure of 1.7 kg/cm? and then unloading to the final stress of 0.43 Kg/en2, thus inducing an overconsolidation ratio of 2. ‘The undrained normal static strength tests (CU-static test) were conducted to provide a reference for interpreting the results of the undrained cyclic triaxial teste with etress controlled loading (CU-cyclic tests) at various stress Levels; one hundred deviatoric stress cycles (sinusoidally varing loading at 0.5 Hs) vere applied to the eamples, If the sample did not fail during the cyclic Loading then a conventional. static undrained compression test using controlled stress was carried out on the sane sample. No drainage was allowed during any Part of eyclie or post-cyclic tests. -30BEHAVIOR DURING UNDRAINED CYCLIC LOADING In all specinens, Series A and B, the level of repeated stress did not cause failure even under 100 stress cycles and an equilibrium state was achieved in which the stress-strain curves followed closed hysteresis loops with no addition fal changes in pore water pressure or strain occurring with load repetition. The relationship between cyclic stress ratio, R, in percent defined as devia tor eyelic stress, qq, divided by static strength, qge, versus axial cyclic strain, cc, dn percent was plotted in Fig, 1 and geonetric regression analysis wag run obtaining the folloving empirical relationships: R30 ee for oc = 1 R= 27.8 co!" for oR = 2 @ As show in Fig. 1, normally consolidated specimens were less deformable that overconsolidated ones. ‘The equivalent Young's modulus, B, vas determined from the slope of a line through the end poines of the loops. The equivalent damping ratio, 1, was calcu lated using the well known expression: “L Area of hysteresie loop S A= Ty fees of triangle Fig. 2 shows the relationships between E/qggs in percent versus single sxisl cyclic strain amplitude, cc, in percent. Te has been shown (Refs. 3,4) that, for a cyclic stress level below a criti- esl value s aon-failure equilibrium state is reached, closed stress-strain hys~ teresis loops occur and the aoii behavior ie essentially elastic. The specimens of Mexico City clay exhibited elastic behavior in spite of irs very high water content. On the basis of a geonetric regression analysis, the following equation for equivalent Young's modulus vere obtained: Bdge = 37-85 eZ" for OCR = 1 @ Blagg * 27078 e[*8 — for oR = 2 o ‘The relationship between equivalent damping ratio, 4, in percent and single axial cyclic strain anplicude, ee, in percent is show in Pig. 3. A Linear re~ Bression analysis was run obtaining the folloving expirical relationships: = 9,05 + 2.43 6 for OoR = 1 6) Ae 8.22 + 2.14 e¢ for OCR = 2 o The applications of a shear stress can set up a negative pressure in the water in the pore space between the particles which persists even after the renov al of the stress. Measurenents were nade of the excessive pore pressure during tthe cyclic loading. In the normally consolidated specinens the residual pore pressure was positive, in all overconsolidated specimens except one (sample B 6) the residual pore pressure was negative. STRENGTH AFTER CYCLIC LOADING It has been shown (Ref, 5) that the strength deteriorating effect due to 1-31cyclic loading becomes nore pronounced as the proportion of the cyclic stress am plitude increases relative to the static component of stress applied initially to the specizen. A convenient way of expressing the possible loss in undrained shear strength 4s to use the failure ratio, Rf, defined as deviator stress failure after cyclic loading, def» divided by the static strength, qef- A non-dimensional plot of failure ratio, RE, versus cyclic stress ratio, Ry ie shown in Fig. 4. The re~ sults obtained by Diaz-Rodrfguez, (1988) vere drawing as a reference for the re- sults of this paper, it is clear that the strength loss with the cyclic stress ratio 1s practically zero for R < 50%. ‘coNcLusros The result presented above support the following conclusions concerning to the behavior of the Mexico City clay tested under cyclic loading conditions. It is realized that the observations apply strictly only to the type of soil and conditions of test employed in the experiments. Normally consolidated specinens were less defornable that overconsolidated 2. Cyclic loading can produce a sustantial decrease in the modulus of soil. BL Post-cyelic static tests indicate that the strength was essensibly unchanged due to prior cyclic loading Af 8 < 50%. ACKNOWLEDGMENTS ‘The author wishes to express his gratitude to Sofl Dynamics Group at the Graduate School of Engineering, at the Universidad Nacional Autonoma de México, in special to Mabel Carcfa and Moises Herrera during the testing progran. The author gratefully acknosledges the cooperation of Eng, Guillermo Springail in providing the soil samples. REFERENCES Zeevaert, Le, "Seisno-soil dynamics response of the ground gurface and build ing foundations in Mexico City earthquake, Septenber 19, 1985", Terzaght Lecture. ASCE Convention, Anaheim, Cel. 27 October, 1987. 2. Garcfa, M.y "Efecto de 1a carge ciclica en muestras isotropicanente consoli: das de’1a Ciudad de Mexico", Master thesis, directed by J.A., Dfaz-Rodriguez, Graduate School of Engineering, UNAM, Mexico, (1987). Sangroy, D-A., Henkel, D.J. and Esring, M.I., “The effective stress response of a saturated clay soil to repeated loading", Canadian Geotechnical Journal, 6: 241-252, (1969). 4, Dfaz-Rodrfguez, JuA., "Behavior of Mexico City clay subjected to undrained repeated leading". To be published. Seed, H.B. and Chan, C.K., "Clay strength under earthquake loading condi~ tions", Journal of the Soll Mechanics and Found. Div., Proc. ASCE, Vol. 92, No. SZ, pp. 53-78, (1966). m-32_/Mgqr Percent Cyclic stress ratio, R Equivalent Young's modulus/q,,, percent 005 10 15 20 Axial cyclle strain, €¢, percent Fig Cyclic stress ratio v5 axial oyctic strain 90; eof Experimental [® OCR= | rok |} Points 0 OcR=2 60 i Flag sot \\ 49) Bop . 20}-Eayy227. 707° lob 3 00810 Axial cyélic strain, €¢, percent Fig 2 Normalized Young's modulus vs axial cyclic strain m-33Equivalent damping ratio, X, percent 10 ce i A20.224 2.14 ey A= 9.08 +2.43 €, 4 6 4} Experimental fe OCR =1 points © OcR=2 2 ° Os 10 1s 20 = er/Aats percent Failure ratio, Ry Axial cyclic strain, €,, percent Fig 3 Equivalent damping ratio vs axial cyclic. strain Experimentol 020. 40 +60 80 100 120 Cyclic stress ratio, R = a,/ayp percent Fig 4 Failure ratio vs cyclic stress ratio m-34