University of Sheffield
Reconstruction of Soil Stress-Strain Response Using Optimisation
Abstract
dc:description.abstractAn Identification Method is a methodology by which the properties of a material can be recovered by back analysis of experimental data. In the context of geotechnics, the relevant data would be the displacement field of a body of soil undergoing deformation along with the loading data that is causing the deformation. The recovered properties for a geotechnical problem that would be most useful can be represented with a stress-strain curve. Two Identification Methods are presented in this thesis. Both methods utilise optimisation such that a stress-strain curve can be recovered that minimises the gap between internal work, a function of the stress and strain fields, and external work, a function of loading and load displacement. The first method, an evolution of existing work, splits the unknown stress-strain curve into a number of segments, with the stress associated with each segment being optimised. The second method, novel for this project, defines the unknown curve with an arbitrary equation, the parameters of which are to be found through optimisation or alternatively a brute force approach. To validate the methods, a series of artificial datasets were produced using FEA and were found to function very well. The datasets were subjected to a variety of artificial degradation strategies such as the addition of random noise with the finding that the peak stress is lowered proportionally to the addition of noise. Laboratory tests were also carried out for validation purposes. A series of footing tests were performed on undrained clay, with Digital Image Correlation (DIC) data recorded, and the recovered curves were compared with triaxial and shear vane data. Depending on the quality of the individual data sets, the methods were found to be promising. More flawed data sets of course produced worse recovered curves, but higher quality datasets resulted in recovered curves that were within the range suggested by triaxial testing.
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Sheffield
- Year dc:date.issued
- 2020
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Charles, Jared
- Advisors dc:contributor.advisor
-
- Smith, Colin
- Black, Jonathan
- ORCID person:identifier.orcid
Rights
- Language dc:language.dcterms_iso639_2
- eng
Identifiers
dc:identifier.*- Identifier
- uk.bl.ethos.829697