|By extending semi-analytical Terramechanics
methods for general three-dimensional tire and
terrain geometries and combining it with a deformable
compaction-based terrain model, general purpose
tire/terrain mobility scenarios can be simulated. A
vertical application was then created with this
framework that combines a multibody vehicle in
CHRONO::Rigid with the physics-based, 3-D deformable
terrain database of CHRONO::Terrain. Using
representative suspension hardpoints, spring/damper
rates and accurate mass/inertia information, a
representative HMMWV vehicle model was developed.
Contact patch force models were developed by
extending semi-analytical terramechanics approaches
to the general, 3-D case. Leveraging High Performance
Computing in the form of parallel CPUs and GPUs
enables real-time vehicle mobilty to be realized,
which enables operator-in-the-loop simulations.
More information available at Justin's Project Page
Contributors: Justin Madsen, Andrew Seidl, Dan Negrut - UW Madison
Prof. Paul Ayers, University of Tennessee-Knoxville
|This project investigates the simulation of
problems involving Fluid-Solid Interaction (FSI). The
term â€œFSIâ€ covers a
wide variety of problems. However, the FSI problems
considered in this project are those in which the
solid phase experiences large translation and
orientation changes. The solid phase will be
represented by a potentially very large number of
rigid bodies that interact through kinematic
bilateral constraints as well as frictional contact.
For these types of problems, a simulation algorithm
will be provided as a unified computational
framework, which leverages High Performance Computing
to enable the effective simulation of the complex
real life problems. Possible applications of this
project include: offshore floating platforms, polymer
flow, and biology applications related to the motion
of the cells in an internal flow.
Contributors: Arman Pazouki, Dan Negrut
A Parallel GPU Implementation of the Absolute Nodal Coordinate Formulation
- With a Frictional/Contact Model for the Simulation of Large Flexible Body SystemsThis contribution discusses how a flexible body formalism, specifically, the Absolute Nodal Coordinate Formulation (ANCF), is combined with a frictional/contact model using a continuous contact force model to address many-body dynamics problems; i.e., problems with hundreds of thousands of rigid and deformable bodies. Since the computational effort associated with these problems is significant, the analytical framework is implemented to leverage the computational power available on todayâ€™s commodity Graphical Processing Unit (GPU) cards. The code developed is validated against ANSYS and FEAP results. The resulting simulation capability is demonstrated in conjunction with hair simulation.
Contributors: Naresh Khude, Dan Melanz, Dan Negrut
|In an effort to support general 3-D vehicle
mobility on non-flat terrain, CHRONO::Terrain is a
deformable terrain database system that allows for
the terrain surface to be described on both macro-
and micro-scale resolutions. Inspired by previous
work that used a combination of global low-resolution
surface elements with localized high-frequency
B-Splines to add "bumpiness", it is possible to
capture slopes, hills and walls, as well as give the
driver the appearance of bumpy, non-flat off-road
terrain. A soil-compression model tracks the 3-D
stress/strain due to vehicle loads, and the terrain
surface deforms according to a visco-elastic-plastic
approach that considers effects of generalized 3-D
tire and terrain geometries.
More info available on Justin's Project Page
Contributors: Justin Madsen, Andrew Seidl, Dan Negrut - UW Madison
Prof. Paul Ayers, George Bozdech, University of Tennessee-Knoxville
Jeff Freeman, Ford Cook-MechSim Inc.
Anchoring on a Near Earth Object (NEO) in a low gravity environment
|This research investigates an approach used to
simulate and validate a simulation capability that is
able to simulate granular dynamics simulations with
millions of objects. This massively parallel granular
dynamics framework relies on ubiquitous Graphics
Processing Unit (GPU) cards that are capable of
performing trillions of floating point operations per
second and can simultaneously process hundreds of
computational threads, making them ideally suited for
the task at hand. Utilizing a variety of experimental
data, the simulation capability was validated with
existing literature. Tests included a lift/drag and
ball drop test in a bed of granular material, a
steady state vibration test, a brazil nut test and a
penetration test. This parallel simulation capability
was used to investigate two types of anchoring
systems, and inflatable anchor and a helical anchor,
when deployed on a Near Earth Object (NEO) in a low
gravity environment. In order to successfully
simulate the anchor and granular material in this
environment, the simulation capability was augmented
with a cohesion model capable of capturing the
necessary behavior and short range forces associated
with NEOs. The goal of this effort is to
experimentally validate the simulation capability and
determine which anchor design is most promising.
Contributors: Hammad Mazhar, Marco Quadrelli, Dan Negrut
Many-Body Dynamics using Multiple GPUs
|This project aims to enhance the lab's existing
massively parallel granular dynamics framework with
the ability to make use of multiple Graphics
Processing Units (GPUs) in both the same computer and
in multiple computers across a network. Nvidia's
GPUDirect technology will initially be used to enable
the sharing of particle data between GPUs. This
technology allows GPUs to communicate with each other
directly over the PCIe bus without CPU interference,
thereby reducing the amount of communications
overhead. GPUs in separate nodes will be able to
communicate over a high bandwidth, low latency
Contributors: Andrew Seidl, Hammad Mazhar, Dan Negrut
Chrono::Render A General-Purpose Rendering Capability of Large-Scale Simulation Data
As simulations grow in complexity the data extracted from the model grows in size. For engineers and scientists, it is difficult and tedious to gain meaningful insights for large data sets; hence visualization becomes critical to computer simulation since it provides a more natural means to extract the salient information of abstruse data. Additionally, visualization makes it easier to share and communicate the content of a simulation leading to wider interest and understanding of its results. Thus, with this need for visualization in mind, we have been developing a rendering pipeline called Chrono::Render, which allows for a simple means to efficiently create "movie-quality" renderings of arbitrary data. However, creating this pipeline poses several technical problems, the most apparent of which being how to handle massively complex scenes efficiently within hardware constraints as well as how to "automate" the process while retaining flexibility; we've determined that Renderman is ideal for managing these issues.
Chrono::Render is a collection of C++ binaries, a Python scripting interface, and a XML scene description which together serve to simplify the complexity of generating complex visual effects for arbitrary data sets. Chrono::Render is currently available for free download as a pre-built binary for Linux (COMING SOON). Members of the Wisconsin Applied Computing Center can use this capability remotely as a service by leveraging 320 AMD cores on which Chrono::Render is currently deployed.Project Page
Contributors: Aaron Bartholomew
Multibody Dynamics with Frictional Contact
Investigating the kinematics and dynamics of rigid bodies with frictional contact is a compute-intensive task requiring methods that can leverage commodity high performance computing resources. This project, a collaboration with Prof. Alessandro Tasora (University of Parma, Italy) and Dr. Mihai Anitescu (Argonne National Lab), investigates parallel algorithms for the simulation of multibody systems with frictional contact. A first objective of this project is to leverage parallel computers to investigate the dynamics of very large systems containing granular material such as sand. Current parallel algorithms allow the simulation of systems with over one million interacting particles. However, improved methods are required to investigate larger systems and perform accurate simulations of a Mars Rover operating on small-particle granular terrain, for example.
Firstly, this work focuses on improving the numerical methods used to solve the dynamics problem, formulated as a Cone Complementarity Problem, on the GPU. Specifically, Krylov methods will be leveraged to solve the associated optimization problem at each time-step. Additionally, parallel preconditioning is being pursued to further improve solution speed and accuracy. Secondly, this work investigates the use of heterogeneous distributed computing to analyze large systems which would not fit on a single piece of computational hardware. The Discrete Element Method will be implemented to solve large problem through domain discretization and heterogeneous computing on a cluster computer.
Contributors: Toby Heyn, Justin Madsen, Dan Negrut
Vehicle - Tire - Terrain Co-Simulation
A complete vehicle simulation involves three components: Vehicle, Powertrain Simulation and Tire-Terrain interactions. The concept of co-simulation looks at simulation of these components by using different dedicated CAE packages. Vehicle dynamics will be simulated in MSC.ADAMS/Car. Powertrain will be simulated in PSAT (Powertrain Systems Analysis Toolkit) or MSC.ADAMS/Car. For the Tire Component, nonlinear finite element model of tire will be simulated using either FTire, ABAQUS or FEAP (Finite Element Analysis Program). In addition to using existing software packages, we are collaborating with Ilinca Stanciulescu at the University of Illinois at Urbana-Champaign who is developing a fully nonlinear finite element model of a tire. Click here to see a video of Professor Stanciulescu's model tire rolling on a rigid surface. Here are a few more Simulation Videos. Video1, Video2, Video3. Here is are the clips of HMMWV on a four post test rig undergoing a pitch and roll motion.
Contributors: Makarand Datar, Dan Negrut
This project entails processing grayscale height-map images of terrain and modeling them as a set of independent particles in user-defined configurations. These terrain models can then be applied in other simulations (e.g. vehicle simulations), especially those that make use of HPC.
Contributors: Spencer O'Rourke
Understanding the Usefulness of Granular Material Simulations
|This project aims to determine to what extent
simulation can be used to model the dynamics of
granular materials. To that end, data is being
gathered from physical models, and then compared to
values from simulations created in Chrono::Engine.
Two systems are being considered in this research,
with both spherical and nonspherical particles; we
are studying the angle of repose of a pile of
granular materials and the mass flow rate of
grandular materials flowing through a slit. Two
outcomes of this effort will be providing a measure
of the accuracy of the Chrono::Engine software, and
contributing to characterizing the effects of
particle shape on the dynamics of the granular
material. This project could lead to a better
understanding of a range of fields, including terrain
(such as tire/terrain interaction); the packing of
materials such as coffee, rice, corn, or
pharmaceutical pills; earthquakes; and designs of
combines and other farm equipment.
Contributors: Rebecca Shotwell, Dan Negrut
A Stochastic Approach to Integrated Vehicle Reliability Prediction
This research addresses some aspects of an on-going multiyear project of GP Technologies in collaboration with University of Wisconsin-Madison for US Army TARDEC. The focus of this research project is to enhance the overall vehicle reliability prediction process. A combination of stochastic models for both the vehicle and operational environment are utilized to determine the range of the system dynamic response. These dynamic results are used as inputs into a finite element analysis of stresses on subsystem components. Finally, resulting stresses are used for damage modeling and life and reliability predictions. The integrated approach combines the computational stochastic mechanics predictions with available statistical experimental databases for assessing vehicle system reliability. Such an integrated reliability prediction approach represents an essential part of an intelligent virtual prototyping environment for ground vehicle design and testing.
Details on the HMMWV model used in the project can be found in Chapter 3 of Justin Madsen's MS Thesis
See the "Animations" page for examples of the vehicle simulations
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 members Justin Madsen and Martin Tupy. An accurate dynamics model of the shovel was created in MSC/ADAMS, and the corresponding motors and controls were modeled using Simulink. Through a co-simulation environment, the lab members 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
More info available on Justin's Project Page
A Framework for Uncertainty Quantification in Vehicle Dynamics pertaining to Terrain Modelling
This research area looks at a methodology for determining the statistics associated with the time evolution of a nonlinear multi-body dynamic system operated under input uncertainty. The focus is on the dynamics of ground vehicle systems in environments characterized by multiple sources of uncertainty: road topography, friction coefficient at the road/tire interface and aerodynamic force loading. Drawing on parametric maximum likelihood estimation, the methodology outlined is general and can be applied to systematically study the impact of sources of uncertainty characterized herein by random processes. The proposed framework is demonstrated through a study that characterizes the uncertainty induced in the loading of the lower control arm of an SUV type vehicle by uncertainty associated with road topography.
Quarter car vehicle simulation for durability analysis of suspension components
The purpose of this project is to assess the load histories in components that connect a wheel with various types of tires to the main body of a vehicle. Once the load histories are obtained, a durability analysis can be performed on the computer model of the component to suggest design improvements, material changes, and repairs and scheduling component replacements. This allows performance testing of a non-pneumatic tire model against a traditional pneumatic tire model. Efforts are being made to perform a co-simulation between full blown non-linear finite element tire model in ABAQUS and quarter car vehicle model in ADAMS.
Contributors: Makarand Datar, Dan Negrut
Employing Gaussian Processes to Model a Non-linear Vehicle on Road Surfaces with Random Ice Patches
Randomly distributed patches of ice arising naturally on roadways adversely effect a driver's ability to navigate a determined path; among the escalated risks in icy conditions are yaw instability (spinning out) and slippage from desired path. The anomalous and often unpredictable distribution of ice makes predictive results from traditional modeling methods inaccurate. In collaboration with Dr. Mihai Anitescu (Argonne National Lab), we employ Gaussian processes to form high fidelity, interpolative models of spatial friction coefficients from a limited data set (achievable with satellite imaging, sensors, or inter-vehicle communication). We work with a non-linear vehicle model on the ice models to a) quantify the effect of ice on a vehicle’s trajectory and b) to identify high risk speeds and turn radii on surveyed roadways. Simulation methods, first developed and verified in MATLAB, are implemented in ADAMS car and results are compared. Further investigation into this problem will develop control methods robust to the stochastic nature of the conditions. Furthermore, the inverse problem of going from trajectory to friction values will be addressed. This capability would allow for inter-vehicle communication of road friction data in the era of onboard sensing and in turn would make our methodology of data interpolation more relevant. Read More... or click here for the current version of the submitted conference paper.
Contributors: Kyle Schmitt, Justin Madsen, Dan Negrut, Mihai Anitescu
Meshless Acoustic Simulations
In this project, we are investigating the potential and limits of a meshless Lagrangian technique, called Smoothed Particle Hydrodynamics (SPH), as a method for acoustic simulations.
The most common, currently used techniques for acoustic simulations are mesh-based methods such as the Boundary Element Method (BEM), Finite Differences (FD), the Finite Element Method (FEM) or more recently, the Lattice Boltzmann Method (LBM).
Though many improvements to these methods have been made during the last few years, each method still has its weaknesses. Problems involving inhomogeneous media, moving boundaries, very high pressure fluctuations and interactions between the acoustic field and the presence of a mean flow are either particularly hard to describe or cannot be simulated at all with some of these meshbased methods.
The investigation of SPH for modeling sound propagation is carried out in order to assess its potential in relation to the limitations associated with the existing simulation methods listed above.
Contributors: Philipp Hahn, Dan Negrut
Implementation of Low Order Numerical Integration Formulas in Rigid and Flexible Multibody Dynamics
This project looks at several low order numerical integration methods: Newmark, Hilber-Hughes-Taylor (HHT), the second order BDF method of Gear , and three new stabilized numerical methods that draw on the HHT formulas and BDF method, in an effort to assess their behavior. The first objective is to briefly indicate the theoretical results available in the literature regarding the stability and convergence properties of these low order methods when applied in the context of multibody dynamics simulation (MBS). The second objective is to perform a set of numerical experiments to compare these integration formulas in terms of several metrics: (a) efficiency, (b) energy preservation, and (c) velocity/acceleration constraint drift. A set of simple mechanical systems are used to this end: a double pendulum, a slider crank with rigid bodies, a slider crank with a flexible body represented in the floating frame formulation, and a seven body mechanism. Read More...
Contributors: Naresh Khude, Toby Heyn, Dan Negrut
High-fidelity modeling and dynamic simulation of tracked vehicles using MSC ADAMS
With the continuous improvement of computer processing speed comes the opportunity to model and simulate increasingly complex mechanical systems. This investigation uses a crawler track model to produce high-fidelity simulations of a crawler propulsion system under different scenarios. A few of the scenarios currently being investigated are the behavior of the model running over obstacles, being dropped (i.e., a parachute landing), and driving over flat surfaces. A seven second simulation of the model moving across a flat surface is shown here: video 1, video 2. The support provided by Holger Haut (http://www.multibodysimulation.com) is gratefully acknowledged.
The track model used in this project consists of 45 individual tracks, each of which are constrained by revolute joints and solid to solid contact forces. There are a total of eight idlers, two sprockets attached to a drive, and a return roller. Since many of the constraints in this model are contact forces, a ten second simulation using the HHT integrator method can take upwards of four hours. This large demand for computing power makes this track model an ideal candidate for use with parallel computing, a topic currently investigated in the Simulation-Based Engineering Lab. In the near future we plan on investigating different ways of interconnecting track elements (using bushings) as well as using rubber pads on the shoes to improve ride comfort and performance.
Contributors: Justin Madsen, Dan Negrut
Investigation of new Numerical Methods for Molecular Dynamics Simulation
Molecular dynamics (MD) is an atomistic simulation technique that can be used to calculate properties of a material by measuring them as the system evolves in time. The system is evolved in time by calculating the forces on individual atoms and solving Newton's equations of motion at each time step. The numerical methods currently employed by MD simulation are only stable under short time steps. As a result, the forces between atoms must be calculated many times even for short simulations, greatly increasing the CPU time required.
We are currently investigating the usefulness of implicit methods developed in the realm of classical mechanical system simulation for use in both biological and material science applications of MD. Our results as of 3/5/2008 are summarized here in for form of a presentation. A more detailed discusion is available in both pdf and html versions.
Contributors: Nick Schafer, Dan Negrut
Multi-scale Simulation in Materials Science
In Materials Science we participated in the development of numerical methods that enable first-principles (ab-initio) computational investigation of nanostructures. Nanostructures have dimensions of the order of 10nm, with up to several thousands of atoms. The electronic structure of materials undergoes substantial qualitative changes when their dimensions are reduced to nanoscale, leading to new regimes of physical, mechanical, and chemical behavior not observed in bulk materials. The goal is to determine through simulation the properties of these nanostructures. The two research directions pursued are multi-scale methods for fine/coarse resolution and a model reduction approach based on Orbital-Free Density-Functional Method (OFDFT) for electronic structure computation. Read More...
Contributors: Toby Heyn, Dan Negrut