Button to scroll to the top of the page.

Colloquium: Graduate Portfolio in Scientific Computation

The Fall Research Colloquium for the Graduate Portfolio in Scientific Computation details are below.






Masoud Behzadinasab Friday, November 16 1:00-1:30 PM GDC 7.402 "Predicting Failure of Metallic Structures with Peridynamics"
Chen Chen Wednesday, December 12 1:00-1:30 PM GDC 7.402  "Intrusive Polynomial Chaos Uncertainty Quantification on Stochastic Shallow Water System, Part (I): Deterministic Solver"

Masoud Behzadinasab

(PhD student in Engineering Mechanics, supervised by Dr. John Foster)

Title: "Predicting Failure of Metallic Structures with Peridynamics"

Abstract: Accurate structural life prediction of components in engineering, naval, and industrial applications is of prominent interest to engineers/scientists. Prediction of dynamic failure in metallic alloys with complex geometry involves lots of challenges. Damage accumulation along the plastic loading path governs the fracture initiation in ductile materials; thus, it is vital to accurately simulate deformations of a malleable structure before predicting its failure behavior. Over the past two decades, the peridynamic theory has been exploited for modeling dynamic problems involving fracture. While peridynamics has broadly been applied to brittle materials, its robustness in modeling ductile fracture has largely remained untested. Sandia Fracture Challenge 2017 (SFC3) provided an opportunity for the mechanics community to assess their modeling capability in predicting deformations and failure behavior in additively manufactured metal. We participated in the challenge to investigate the accuracy of the state of the art of peridynamic modeling of ductile fracture. Recent material and damage models have been implemented in Peridigm, an open-source massively-parallel computational peridynamics code. The framework was first calibrated by the data provided by Sandia National Laboratories. Following that, a blind prediction was performed on the challenge problem. Simulation results are compared with the experiments to assess the approach.

 Chen Chen

(PhD, Engineering Mechanics, Cockrell School of Engineering, supervised by Dr. Clint Dawson)

Title: "Intrusive Polynomial Chaos Uncertainty Quantification on Stochastic Shallow Water System, Part (I): Deterministic Solver"

Abstract: "In order to quantify uncertainty in water elevation prediction, it usually takes months for a research scientist to determine parameter space sampling strategy, run forward model hundreds of times, and build up a surrogate model. As we all know that the quality of the final surrogate model depends largely on the sampling strategy. However, sampling parameter space is the first step toward building a surrogate. If we happened to choose a sampling strategy which gave us unsatisfactory surrogate surface, we are forced to rerun each forward simulation which is computationally costy and time consuming. Hence in this paper we aimed at searching for a numerical method in order to quantify the uncertainty of water elevation prediction in real-time. We chose a method called intrusive polynomial chaos(IPC) to resolve this issue. As part (I) of the paper, we firstly need to choose a numerical scheme which can calculate deterministic surge elevation and can also be compatible with IPC. It has been found that the operator splitting finite element approach suits our need. Three test cases have been conducted to verify this model: the slosh, the hump, and theidealized inlet test cases. By comparing the model results against analytical solution and a well-known adaptive hydraulic model, and by conducting convergence test, we managed to verify our model. This paper serves as a foundation for the theory in part (II), which is to build up a stochastic finite element model in order to achieve instant uncertainty quantification in surge elevation."