Learning from Field Measurements Through Heuristic Back Analysis

Abstract

Back-analysis involves the determination of input parameters required in computational models using field monitored data, and is particularly suited to underground constructions, where more information about ground conditions and response become available as the construction progresses. A crucial component of back-analysis is an algorithm to find a set of input parameters that will minimize the difference between predicted and measured performance (e.g., in terms of deformations, stresses or tunnel support loads). Methods of back-analysis can be broadly classified as direct and gradient-based optimization techniques. An alternative methodology to carry out the required nonlinear optimization in back-analyses is the use of heuristic techniques. Heuristic methods refer to experience-based techniques for problem solving, learning, and discovery that find a solution which is not guaranteed to be optimal, but is good enough for a given set of goals. Two heuristic methods are presented and discussed, namely, Simulated Annealing (SA) and Differential Evolution Genetic Algorithm (DEGA). SA replicates the metallurgical processing of metals annealing, which involves a gradual and sufficiently slow cooling of a metal from the heated phase, which leads to a final material with a minimum imperfections and internal dislocations. DEGA emulates the Darwinian evolution theory of the survival of the fittest. Descriptions of SA and DEGA, their implementations in the computer code Fast Lagrangean Analysis of Continua (FLAC), and uses in the back-analysis of the response of idealized tunnelling problems, and a real case of a twin tunnel in China are presented.