Use of digital image correlation to directly derive soil stress-strain response from physical model test data

摘要

Image correlation techniques generate large quantities of data from a physical model test that can be used to reconstruct the in-situ stress-strain response of the soil within the model. Such data can be used to significantly enhance the interpretation of the model test. This paper evaluates an energy based optimization approach for reconstructing the stress-strain response of a soil from a combination of external load-settlement data and internal strain field data. The core concepts are illustrated through application of the method to undrained compression, shear and bearing capacity tests. Utilising idealised artificially generated strain fields from finite element (FE) simulations, it is demonstrated that the original stress-strain response can be closely re-constructed. Key steps in the process that require careful control are highlighted and the issues involved in extending the method to deal with drained soils and noisy image data are discussed.