Application of MLR Procedure for Prediction of Liquefaction-Induced Lateral Spread Displacement

I presented the Kenneth L. Lee lecture at the 2016 Queen Mary Seminar, ASCE Los Angeles Geo-Institute Chapter, and publish it herein. The topic was lateral spread problems I have encountered as a consultant. The first issue was how deep to bury a pipeline at stream crossings to mitigate the lateral spread hazard. My answer is twice the bank height (2H) beneath approaches or 1H beneath the channel. For shallow liquefiable layers, a shear zone would form well above and not harm the pipe. For deep liquefiable layers, nonliquefiable soil above the pipe should buttress the channel against lateral spread. The second issue was finding the thinnest liquefiable layer that is susceptible to lateral spread. In the database of Youd et al. database, the thinnest layer is 1.0 m. For case histories used by Zhang et al. to verify their procedure, the minimum thickness is 0.6 m. Extrapolation to thinner layers adds uncertainty and a tendency for overprediction. The third issue was how to apply multiple linear regression (MLR) at a site with insufficient SPT data. CPT data were used to create $\psi$ profiles; we then applied the criterion that $\psi <–0.08$ indicates material too dilative for lateral spread to develop. Summing layer thicknesses with $\psi >–0.08$ yielded estimates of ${\mathrm{T}}_{15}$ for use with MLR.