ABSTRACT
Soil-parameter identification is a long-standing topic that has received considerable attention over the years. Traditional identification approaches usually follow the least-squares criteria. These approaches aim to find a single set of parameter values that produces the minimum absolute error between numerical predictions and measurements. However, these approaches have several drawbacks, particularly their inability to accommodate systematic modelling uncertainty. A recent data-interpretation methodology, Error-Domain Model Falsification (EDMF), is used to overcome these drawbacks. A key feature of EDMF is the explicit representation of systematic uncertainties originating from model simplifications. This paper presents the application and adaptation of this methodology in the area of soil parameter-identification. A synthetic excavation case is used as an illustration. Using EDMF, accurate soil properties can be identified with the inclusion of systematic uncertainties, which leads to improved predictions of the excavation performance. The residual minimization approach, however, returns inaccurate soil properties and biased predictions. The present study shows that EDMF offers a promising data-interpretation methodology that can yield multiple sets of admissible parameter values, which in turn help engineers to appreciate and understand the possible variations in soil properties and predictions that are compatible with field measurements.
