BEHAVIOUR OF REINFORCED EARTHEN EMBANKMENTS DURING EARTHQUAKES

Abstract

Ductile reinforced earth (R.E.) capable of taking compression and tension is ideal for use in geotechnical earthquake engineering. This investigation attempts to understand its behaviour in (earthen) embankments during earthquakes. Large R.E. embankments were tested and analyzed under dynamic loads by treating it to be homogeneous and elastic for simplicity. Objective is to understand dynamic behaviour of R.E. embankments to study: (a) Preparation of test embankments on shake table (b) Embankment response (c) Characterizing factors affecting response and analysis of test data (d) Determination of strain dependent dynamic shear moduli by back analyses of test data (e) Determination of dynamic pullout resistance of fabric from back analyses of frequency-response test data (f) Idealizing R.E. to be homogeneous in FEM analysis to obtain response using shear moduli from tests and by using excitations used in tests (g) Comparison of analytical and experimental embankment responses. Air dry sand, gcotextile reinforcements and sinusoidal excitation were used since seismic excitation may be represented by its sinusoidal equivalent (Seed and Idriss, 1971). This analysis in time domain can also consider earthquake excitation. Following were studied in this study : (a) Forming uniformly dense embankment by using a specifically developed device (b) Developing stress control setup to obtain pullout resistance of fabric (c) Construction of test embankments 1.5 m long and 0.75m wide with different reinforcement arrangements. Developing technique to create plane IV strain conditions in transverse section by restraining longitudional embankment strains (d) Exciting embankments with maximum acceleration up to 0.32g and frequency in range of 5-20 Hz (e) Measuring embankment response at different points (f) Evaluating strain dependent shear moduli for R.E. at different strains (g) Evaluation of dynamic pullout resistance coefficient mobilized along fabric and comparison with static values (h) Analysis in time domain of R.E. embankments with excitation used in tests by FEM by assuming R.E. to be homogeneous to compare with experimental responses (i) Comparing response of plain sand embankments of height, top width and density same as those for R.E. embankments but with different side slopes with R.E. embankment response to highlight merits of R.E. This study investigated the following :(a) Shear modulus and shear strain (b) Frequency of excitation, natural frequency, and excitation force ratio (c) Shear wave velocity, time lag, phase difference, phase angle and acceleration (d) Inertia at the level of each fabric (e) Coefficient of dynamic pullout resistance for fabrics of each layer and coefficient of average and maximum dynamic pullout resistance (f) Displacement, confining pressure and settlement (g) Response and dynamic shear stress by FEM. Sand rain apparatus (SRA) developed produced uniform deposits over entire test bed which are better than those from setup reported so far. Strain control setup fails to study time dependent pullout displacement. Stress control setup developed enables such a study also and hence is superior to strain control setup. Lab technique developed for creating plane strain conditions is an important research contribution. Based on analytical and test results, following conclusions were drawn: a. Linear variation of shear strain with excitation force ratio for all the three R.E. embankments indicates their elastic behaviour within excitation range employed, even when excited at resonance. b. Seismic coefficients recommended by Richardson and Lee (1975) are smaller than those obtained by testing R.E. embankments and fail to predict nonlinear seismic coefficient-base acceleration relationship for stronger excitation and for different embankment stiffnesses. c. Values of F by method proposed by Richardson et. al. (1977) for R.E. embankments are higher than those obtained from tests. Hence, their expression is not valid for all R.E. embankments. d. Continuous reinforcements reduce response near embankment top. e. For safety, ^advmx ^avdmx and dimensionless disturbing force may be obtained for xMISi at different OME. For stiffer R.E. embankment, these variables are lower than those for weaker R.E. embankment. At resonance, y < reduces sharply with change in r^. f. Dimensionless disturbing force, Madvmx '*avcmix at different 0ME and rf values remain nearly the same for stiffer embankment and appreciably different for weaker embankment. g. R.E. embankments do not fail even when lateral displacements are more than 0.005 H at which plain soil fail. h. Maximum displacement occurs at about 0.6H from top which is also reported by other investigators. i. Computed post-vibration confining pressures are in agreement with field data of a 15 m high R.E. embankment and with test data reported by Richardson and Lee (1975) and with test data of Fairless (1989). Continuous fabrics not failing in tension are more effective than discontinuous ones designed for pullout resistance. VI j. Post vibration settlements are small even after experiencing resonance. k. Damping ratio-shear strain relationship proposed by Seed et.al. (1984) for plain sands is also valid for reinforced earth. 1. For R.E. embankments, shear strain and damping ratio are of the order of 10 and 0.18 which are much higher than corresponding values of 10 and 0.05 respectively for plain sands in elastic domain. Since damping greatly affects shear stress, it helps to reduce peak dynamic shear stress within elastic domain. This is a great advantage of R.E. m. For R.E. embankments, response by FEM analysis by idealizing reinforced earth to be homogenous and by using proper damping ratio are comparable to measured response for same excitation. Measured response is closer to computed one with 2-layer idealization which is better, n. For same top width, height and excitation, measured response and computed shear stress for R.E. embankment are much smaller than those for plain sand embankments in top 40% of the depth. Maximum dynamic shear stress in plain sand embankments is far more (2.3) than that of R.E. embankment. This highlights advantages of R.E. embankments. Analytical and test results clearly indicate that responses of R.E. embankment, specially that using continuous reinforcement, is far better than that of plain sand embankment for earthquake conditions. Also it is seen that, for the purposes of analysis, R.E. in (earthen) embankment can be treated to be homogeneous material with improved shear modulus without losing accuracy. It would be better to analyze embankment by considering it as a layered system with appropriate shear modulus for each layer.