IPS Demify^® for Heavy Machinery and Vehicles

Soil, Rock and Material Simulation and Soil-Tool Interaction

The high variability of construction and agricultural machinery leads to high demands on the product development. Not only the interaction of the tool with the ground during excavation or plowing, but also the interaction of a harvester with crops, etc. are hardly predictable. Until now, such complex interactions are usually determined by measurement and are therefore difficult to integrate into the virtual product development.

IPS Demify^® and Demify^® for Heavy Machinery and Vehicles

IPS Demify^® is a versatile software suite with a GUI interface similar to the IPS cable simulation. IPS Demify^®’s core satisfies the newest coding standards. The code is thus light weight, easy to maintain and new features demanded by our customers can be easily added and implemented. Demify^® also has a Python Application Programming Interface (API) allowing tailored scripting methods. Demify^® enables the user to efficiently simulate large particle systems based not only on spheres of different sizes, but also on more general complex shape geometries, i.e. convex, dilated polyhedra. Demify^® runs on commercial Nvidia GPUs, but also a CPU solver is at hand, allowing the user to simulate on a standard Desktop machine with a powerful GPU, but also using cutting-edge high-performance technologies on a cluster.

Simulation of Particles

Methods of classical soil mechanics, e.g. Finite Element Method (FEM) with special material models for the nonlinear material behavior of the soil, fail in this application. A particular challenge is the separation of the material during excavation. Alternative simulation approaches, such as particle simulation can better deal with these conditions. In particular, the Discrete Element Method (DEM) is proven for simulations of this kind.

Besides the usage of the DEM in industrial projects, we also work on the development of particle methods. We recently joined forces with the Fraunhofer Chalmers Centre and develop the software suite «Demify^® for Heavy Machinery and Vehicles» together. Particularly, the modelling approaches developed at the Fraunhofer ITWM have been merged into the IPS Demify^® software suite. Our central goal is the prediction of correct reaction forces – which implies the necessity of an appropriate model parameterization – and the applicability in closed-loop scenarios.

Simulation of Interaction

We have evolved, in cooperation with Volvo Construction Equipment AB (Sweden) and Fraunhofer Chalmers Centre, a co-simulation framework in the construction equipment development context that serves to simulate and analyze the interaction of construction machines with material.

We conceptually establish a force-displacement coupling for the co-simulation setup, that is in principle not confined to a certain construction machine but can be straightforwardly applied to wheel loaders, haulers, excavators, etc. The construction machinery is modeled in multibody software which can be integrated into the coupling framework. The coupling of Demify^® to other solvers, i.e. an FEM solver has also been established.

Realtime Simulation of Soil-Tool Interaction

Recent studies on machine learning based algorithms resulted in the development of realtime capable soil-tool interaction models. The aim is to enable the interactive soil simulation at the robot-based driving and operation simulator RODOS^®. The training data can be designed specifically or generated in the interactive ride at RODOS^®. Based on these training data, so called Lookup Tables or Recurrent Neural Networks (RNNs) are trained using the full DEM simulations. Such surrogate models can then be used to predict correct force and torque feedback from soil to tool in the interactive simulation. This opens the possibility to predict the soil interaction at RODOS^® online. Details are discussed in the dissertation on »Efficient Numerical Simulation of Soil-Tool Interaction«.

More details, numerical results and a verification based on real measurements are presented in selected publications listed below.

• M. Obermayr, K. Dreßler, C. Vrettos, P. Eberhard: »Prediction of draft forces in cohesionless soil with the Discrete Element Method« Journal of Terramechanics, 48: pages 347-358, 2011.
• M. Obermayr, K. Dreßler, C. Vrettos, P. Eberhard: »A bonded-particle model for cemented sand« Computers and Geotechnics, 49:299-313, 2013.
• M. Obermayr, C. Vrettos, P. Eberhard, T. Däuwel: »A discrete element model and its experimental validation for the prediction of draft forces in cohesive soil«  Journal of Terramechanics, 53: pages 93-104, 2014.
• M. Balzer, M. Burger, T. Däuwel, T. Ekevid, S. Steidel, D. Weber: »Coupling DEM Particles to MBS Wheel Loader via Co-Simulation« Proceedings of the 4^th International Commercial Vehicle Technology Symposium, Kaiserslautern (CVT 2016), pages 479-488, 2016.
• M. Burger, K. Dreßler, T. Ekevid, S. Steidel, D. Weber: »Coupling a DEM material model to multibody construction equipment« Proceedings of the 8^th ECCOMAS Thematic Conference on Multibody Dynamics, Prague, pages 417-424, 2017.
• J. Jahnke, S. Steidel, M. Burger and B. Simeon: Efficient Particle Simulation Using a Two-Phase DEM-Lookup Approach. In: Proceedings of the 9th ECCOMAS Thematic Conference on Multibody Dynamics 2019, Computational Methods in Applied Sciences 53, pp. 425-432, 2020.
• J. Jahnke: Efficient Numerical Simulation of Soil-Tool Interaction, Fraunhofer Verlag, Dissertation, Stuttgart, 2022.
• J. Jahnke, S. Steidel, M. Burger, K. Jareteg and J. Quist: Efficient and Robust Parameter Identification for Soil modeled via the Discrete Element Method. In: Proceedings of the 7th International Commercial Vehicle Technology Symposium, Kaiserslautern, pages 52 - 63, 2022.