(1.) “Modeling crystallization pathways of polymorphic materials”
Tesia D. Janicki, Z. Wan, S. E. Babcock, T. F. Keuch, P. G. Evans, J. R. Schmidt University of Wisconsin–Madison
We explore the application of a collective variable derived from the Debye scattering function to examine amorphous-crystalline polymorphic phase transitions in Lennard-Jones argon and alumina (Al2O3). These results suggest that metadynamics can provide not only thermodynamic, but also mechanistic and kinetic, insight into complex crystallization phenomena.
(2.) “Flexibility from Networks of Data Centers: A Market Clearing Formulation with Virtual Links”
Weiqi Zhang, UW-Madison, Line A. Roald, UW-Madison, Andrew A. Chien, University of Chicago, Argonne National Lab, John R. Birge, University of Chicago, Victor M. Zavala, UW-Madison
We formulate an electricity market clearing model that captures load shifting flexibility from data centers. This model shows how data center flexibility can help mitigate price volatility in the markets.
(3.) “SynChrono: An MPI-Based, Scalable Physics-Based Simulation Framework for AI Studies of Autonomous Vehicles Operating in Off-Road Conditions”
Jay Taves, Asher Elmquist, Radu Serban, Dan Negrut
SynChrono is an open source, physics-based simulation platform for modeling autonomous vehicles. It can scale up to hundreds of vehicles while remaining in real-time and includes physics-driven vehicle and terrain models that enable simulation and testing of autonomous vehicles in reinforcement learning environments
(4.) “Discovering Multiscale Structure Using Data-Driven Wavelets”
Daniel Floryan, Michael D. Graham
Multiscale structure is all around us: in biological tissues, active matter, oceans, networks, speech, and images. Identifying the multiscale features of these processes is crucial to our understanding and control of them. We have developed a method that rationally extracts localized multiscale features from data, which may be thought of as the building blocks of these processes.
(5.) “Topological Data Analysis: A Study of Molecular Simulations”
Alexander D. Smith, Spencer Runde, Alex Chew, Reid Van Lehn, Victor M. Zavala
In this work we discuss a new field of applied mathematics known as Topological Data Analysis (TDA). We present an application of TDA to Molecular Dynamics simulations and demonstrate how TDA can be used to extract simple, meaningful topological and geometric features from complex simulation data for use in statistical analysis.
(6.) “Predicting Energy Harvesting Efficiency in Two Tandem Oscillating Foils”
Bernardo L. R. Ribeiro, David Burkhart, Jennifer Franck (UW-Madison); Yunxing Su, Kenneth Breuer (Brown University).
Oscillating foils have shown to be an efficient option for energy harvesting. Using numerical simulations, the wake between foils is analyzed and a model is developed to predict efficiency in different array configurations.
(7.) “Finding Optimal Power System Frequencies”
David Sehloff, Line Roald, Giri Venkataramanan, UW-Madison
Developments in grid-scale power electronics have removed the necessity that power systems operate at a single fixed frequency, unlocking substantial advantages. This work quantifies these separate advantages of capacity, flexibility, and control, using a variable frequency circuit model and an AC optimal power flow formulation and solver, demonstrating value for power systems planning and operation.
(8.) “Predicting Chaotic Dynamics with Neural ODEs”
Alec Linot and Michael D. Graham
Trajectories of chaotic dynamical systems tend towards finite dimensional attractors. We use neural networks to find the coordinate transformation and dynamics on this manifold from sparse data giving excellent attractor reconstruction.
(9.) “Optimizing Electromagnetic Coils for use in Fusion Energy Devices”
Luquant Singh, T. Kruger, A. Bader, D.T Anderson (UW-Madison), C. Zhu, S. Hudson (Princeton Plasma Physics Laboratory)
Controlled nuclear fusion could provide a practically unlimited carbon-free energy source, so long as a thermonuclear plasma can be confined well enough by external electromagnetic coils for fusion reactions to take place. Coil models for such plasmas are discussed, and results from a new coil optimization implementation are presented.
(10.) “Advanced Control for Autonomous Driving”
Alex Pletta, Aaron Young, Evan Deutsch, Joshua Pulsipher, Professor Dan Negrut
Machine learning provides ways to enhance autonomous control systems by modeling highly non-linear system dynamics and adapting control policies for new data. Steps for developing and testing learning-based autonomous controllers are discussed, and a case study is conducted in a high-speed simulation race for the Indy Autonomous Challenge, an international university competition.
(11.) “Understanding the extraordinary stability of ZIF-71 RHO in humid SO2 environment via QM/MM methods”
Kai Cui, Schmidt Group
Most ZIFs are unstable in acidic environment but ZIF-71 RHO shows extraordinary stability in humid SO2 environment. We developed a novel QM/MM method to model the reactivity of ZIFs and used this method to rationalize the stability of ZIF-71
(12.) “Simulations of bio-inspired undulated cylinders through dynamic morphing of surface topography”
Mikihisa Yuasa, Kathleen Lyons, Jennifer A. Franck
Undulation on a cylinder, inspired by seal whiskers, lower the drag force and vibration in the flow when compared to smooth cylinders. This research introduces a new methodology to dynamically modify the complex whisker-inspired surface geometries during simulation to achieve fast and stable performance enabling exploration of a large parameter space of the undulated cylinder geometry.
(13.) “Distribution Grid Optimization: Mitigating Voltage Unbalance due to High Penetrations of Solar PV”
Kshitij Girigoudar, Line Roald
There has been growing penetration of distributed energy resources like rooftop solar PV in the recent past. The research work is focused on using optimization methods to address the power quality issue of voltage unbalance due to large-scale penetration of solar PV installations in the distribution grids.
(14.) “Spatio-Temporal Analysis of Liquid Crystal Responses for O3/Cl2 Detection”
Shengli Jiang, Department of Chemical and Biological Engineering, UW-Madison
Liquid crystal sensors have interesting spatio-temporal responses that are critical in characterizing the chemical environment to which they have been exposed. Compared with only using the light intensity, spatial-temporal analysis using color information uncovers more hidden signatures and establish new design principles for liquid crystal sensors.
(15.) “A Posteriori Chance Constraint Tuning for DC Optimal Power Flow”
The increased penetration of renewable energy has introduced uncertainty into power systems that can render solutions obtained from traditional, deterministic models insecure. In this poster, we present a computationally light-weight method for solving chance-constrained optimal power flow that iterates between solving an approximated reformulation of the optimization problem and using results from a posteriori sample-based tests.
(16.) “Predicting Critical Micelle Concentration of Surfactants Using Graph Convolutional Neural Networks”
Amy Qin, Herry Jin, Reid C. Van Lehn, Victor M. Zavala
We aim to predict the critical micelle concentration (CMC) values of all four major types of surfactants using a graph convolutional neural network (GCN) with simple molecular information input. We will also illustrate the potential of this network in molecular design by deriving new surfactant structures and validating the GCN predictions through molecular simulations.
(17.) “Low-velocity impact simulations with pkdgrav and Chrono::Parallel”
Cecily Sunday, Florian Thuillet, Naomi Murdoch, Adam Yp-Tcha, Melanie Drilleau, Patrick Michel
This work presents the results from a set of low-velocity impact simulations that were performed using the soft-sphere Discrete Element Method (DEM) and two different codes: pkdgrav and Chrono::Parallel.
(18.) “Simulating granular surfaces in reduced-gravity environments”
Cecily Sunday, Naomi Murdoch, Simon Tardivel, Stephan Schwartz, Patrick Michel
In this work, we present a series of simulations and experiments were used to validate the smooth contact model (SMC) in Chrono::Parallel. Our ultimate objective is to use Chrono::Parallel to simulate the interaction between rovers, landers, and instruments on the surfaces of asteroids and small moons.
(19.) “Balancing Wildfire Risk and Power Outages”
Noah Rhodes and Line Roald (UW-Madison), Lewis Ntaimo (Texas A&M University)
This research introduces an optimization approach to wildfire risk in electric grids that optimizes the components we disable to both reduce wildfire risk and minimize the load shedding that occurs. This is compared against a heuristic method for Public Safety Power Shutoffs.
(20.) “Stochastic Programming Approach To Resource Pooling During a Pandemic”
We focus on the problem of sharing ventilators among different states of US during a pandemic. Local shortages in a state can be met by pooling the ventilator supply from the states with their peak demand in the past or weeks ahead. We propose a novel multi-stage stochastic program formulation for the ventilator pooling problem.
(21.) “Fast Mast matrix-free elastoplastic additive manufacturing simulation”
A fast matrix-free parallelizable elastoplastic solver for predicting residual stresses in additive manufacturing is presented. The solver exhibits a 10X increase in speed on CPU when compared to ANSYS (CPU) for 1 million elements.
(22.) “GPU-Accelerated Granular Simulation with Mesh Interaction”
Ruochung Zhang, Luning Fang, Jason Zhou, Dan Negrut
Introducing Chrono::Granular, an open source, effective and efficient CUDA GPU programming framework for granular dynamics. Chrono::Granular scales linearly up to 200 million full-history frictional particles, can interact with meshes and other Chrono modules, and is highly customizable.