Date of Award

Spring 2020

Document Type


Degree Name

Master of Science (MS)


Mechanical Engineering

First Advisor

Voglewede, Philip

Second Advisor

Fleischmann, Jonathan

Third Advisor

Crovetti, James


A numerical pullout test was built using the discrete element method (DEM) to model and capture the pullout response of steel reinforcements and soil in mechanically stabilized earth (MSE) walls. Through numerical modeling, microscale phenomena showing aggregate behavior in response to the reinforcement can be used to gain insight into the macroscale structure. The general setup of the simulation is a steel specimen encased in a rectangular apparatus filled with particles. A normal pressure is applied to the top layer of particles while the strap is slowly removed from the box until it reaches a prescribed displacement.The simulation was created using YADE, an open-source DEM software, which allows for rapid scene construction via scripting. The numerical model uses an iterative approach to step through time while resolving contacts at each step and translating those contacts into forces to ultimately provide updated positions for each body at every time step. For this research, a non-cohesive, elastic-frictional Cundall-Strack contact model was employed to resolve interactions on an individual body basis. Test parameters were largely based on the experimental setup of pullout tests performed by Weldu. Particle packings for the pullout simulation were calibrated to the aggregate used in Weldu’s experiments by setting up a simple triaxial compression simulation within YADE to derive the correct microscale particle friction angle such that it produced the proper macroscale behavior.Using the numerical model, three sets of experiments from Weldu’s research were reproduced with particle uniformity coefficients of 1, 2, and 3. Simulations sets were run at various normal pressures and included 400,860 particles at the upper end. The numerical tests resulted in an encouraging degree of correlation to the laboratory experiments, with pullout residuals being as close as 2% different and an average of 14% different. In addition, this thesis discusses some of the microscale data extracted from the simulations, such as force chains and rolling characteristics, and how numerical simulations could be used in the future to help guide pullout testing and MSE wall design.

Included in

Engineering Commons