Simulation of Granular Material using DVI and DEM

My current work is focused on the validation of the differential inequality approach (DVI) using experimental measurements and the well known discrete element method. Unlike penalty methods as in DEM, unilateral constraints are inforced as an inequality leading to an optimization problem. This approach can handle much bigger step sizes than DEM resulting in the possibility of simulating larger systems in less amount of time.

To carry out the validation I was writing my own DEM code first in MATLAB and then in C. A parallel version running on the GPU using CUDA is currently in progress.

The experiments focus on the flow of granular material.

Movies of different simulations and experiments can be found under Animations.

Parts of the work will be presented at the ASME 2010 IDETEC and CIE in Canada.

 


 

Electric Mining Shovel Simulation using a Co-Simulation environment

The goal of this project was to determine how the dig cycle time of a large above ground electric mining shovel was affected by varying the types of electric motors that actuate the shovel. The implementation was carried out by lab member Justin Madsen and myself. An accurate dynamics model of the shovel was created in MSC/ADAMS, and the corresponding motors and controls were modeled using MATLAB/Simulink. Through a co-simulation environment, we were able to successfully create an accurate simulation of the shovel on the system level. Dig cycle times were representative of actual measured values, and the simulations were able to show the advantage of on type of electric motor over the other. This project was done in collaboration with P&H Mining Equipment.

A movie can be found under Animations.


 

Surrogate Modelling

When dealing with computationally expensive simulation codes or process measurement data, surrogate modeling methods are firmly established as facilitators for design space exploration, sensitivity analysis, visualization, prototyping and optimization. Surrogate models are widely used for parametric studies, optimization, design-space exploration, visualization, prototyping, and sensitivity analysis. Following picture depicts an outline of two surrogate modeling projects done in SBEL. In these two projects we outlined an approach for speeding up the simulation of the dynamic response of vehicle model (full HMMWV model) that includes high fidelity FE tire model and FTire tire components. The method proposed replaces the tire models with a surrogate model that emulates the dynamic response of the actual tire. In the proposed methodology, training information generated with a reduced number of harmonic excitations is used to construct a surrogate model using “Support Vector Machine” for FE tire model and “Embedded Neural Network” for FTire model. The proposed approach has three stages: a system representation, a simulation stage which followed by a constructing of the learned model and the third stage of embedding the surrogate model in the full HMMWV model for the purpose of validation.