Inspection of two sophisticated models for sand based on generalized plasticity: Monotonic loading and Monte Carlo analysis. (8th December 2022)

• Record Type: Journal Article
• Authors: Yeh, Fu‐Hsuan; Tafili, Merita; Prada‐Sarmiento, Luis Felipe; Ge, Louis; Wichtmann, Torsten
• Is Part Of: International journal for numerical and analytical methods in geomechanics. Volume 47:Number 3(2023)
• Page Start: 425
• Page End: 456
• Publication Date: 2022-12-08
• Subjects: generalized plasticity -- granular soils -- monotonic loading -- Monte Carlo toolbox -- material parameters; Soil mechanics -- Mathematics -- Periodicals; Rock mechanics -- Mathematics -- Periodicals; 624.1510151
• DOI: 10.1002/nag.3476
• Languages: English
• ISSNs: 0363-9061

Abstract

Abstract: The generalized plasticity theory has evolved in the last decades and attracts the attention of different research groups, because of characteristics like the absence of an explicit flow surface formulation. This paper presents a detailed inspection of two sophisticated constitutive models for granular soils under monotonic loading: the Pastor‐Zienkiewicz‐Chan (P‐Z‐C) model[1] and the Manzanal‐Merodo‐Pastor (M‐M‐P) model.[2] Even though they are similar in mathematical framework, the P‐Z‐C model requires only eight parameters compared to the M‐M‐P model which requires 18 parameters to describe the material behavior under monotonic loading. Parameters of both models were calibrated based on the extensive experimental database available for monotonic triaxial tests performed on Karlsruhe fine sand by Wichtmann et al.[3] An objective discussion addressing the strengths and weaknesses of each model is presented. Finally, a procedure to identify the most relevant parameters that control different elements in the model like the stiffness, the evolution of dilatancy or the hardening variables is introduced, using the Monte Carlo Toolbox (MCAT) for simulations. Some interesting inferences could be obtained, like the conclusion that the MCAT is a helpful tool for identifying material parameters and their interaction, but it cannot replace the human factor, the experience and objective judgement of the modeller for the calibration of constitutive models.