Skip to main content

REAL-time SIMulation for Design of Multi-physics Systems

Objective

The primary goal of this project is to optimise products and to reduce product development cost and time by design optimisation and by hardware-in-the-loop simulation for assessment, training and test automation purposes. In order to reach this goal for realistic industrial products, the project is focusing on real-time simulation of large scale, multi-physics systems (MPS), together with close integration of interactive experimentation and visualization techniques. The project is divided into:
a) technology development including object-oriented modelling of MPS, symbolic transformation of large scale models for generation of real-time code, parallel and distributed simulation to achieve real-time performance of hardware-in-the-loop systems and to enhance the performance of multi-criteria optimisation, visualization of MPS and;
b) applications in the areas: design optimisation of robots and simulation of power system transients.

Objectives:
The primary goal of this project is to optimise products and to reduce product development cost and time by design optimisation and by hardware-in-the-loop simulation for assessment, training and test automation purposes. This goal leads to the following project objectives:
1) Convenient handling of realistic industrial products, e.g. large scale (>= 100000 equations), multi-physics systems (>= 4 domains);
2) Real-time code generation;
3) Real-time simulation (sample time down to 50 micro seconds). Four Interactive experimentation and visualization. To guide the research, two applications are carried out:
a) Design optimisation of an industrial robot to increase robot speed by 10%, decrease vibrations by 30% decrease power consumption by 20%, and reduce product cost by 20%;
b) Hardware-in-the-Loop simulation of power systems transients for assessment of line fault detection.

Work description:
In the project the newly available object-oriented modeling language Modelica will be used to model in a convenient way large-scale, multi-physics systems. Based on an available Modelica translator the following work will be carried out:
1) A Modelica model is described on a high level and translated to executable code. Symbolic transformation algorithms in the translation phase are enhanced to improve the efficiency of the generated code for real-time applications. Heuristic algorithms are developed forautomatic parallelization of the evaluation of the model equations for tightly coupled multi-processor systems;
2) Simulation of the real-time code generated in (1) is carried out with real-time integration algorithms. Especially, multi-rate methods are developed to integrate subcomponents with strong varying time constants with different step-sizes;
3) The code generation of the model equations is enhanced to generate symbolic Jacobians for optimisation, based on sensitivity differential equations of the model. Design optimisation is carried out by multi-criteria / multi-model optimisation with enhanced performance by using the real-time simulation models from (1) and (2) and by parallelization of the simulation runs;
4) The previous subtasks are integrated in an experimentation environment in order to define Modelica models, simulation, design optimisation, parallelization, and to support documentation, as well as to incorporate geometric and assembly information of a CAD system into the Modelica description for combined treatment of dynamic simulation and visual effects. The research and development process is guided by two realistic industrial applications where all steps are carried out, in particular model library development, code generation, hardware-in-the-loop simulation and design optimization.

Milestones:
The project duration is 30 months. Major milestones: after 9, 18, 30 months.

Expected results: 1) A demonstrator of a real-time simulation and design environment for large scale, multi-physics systems; 2) Reports describing the details of (1) and in particular the developed algorithms for real-time simulation; 3) Modelica component libraries for robots and power systems; 4): Reports describing the application of (1) to robots and power systems using the libraries from (3).

Funding Scheme

CSC - Cost-sharing contracts

Coordinator

LINKOEPINGS UNIVERSITET
Address
Campus Valla
581 83 Linkoeping
Sweden

Participants (5)

ABB SWITZERLAND
Switzerland
Address
Brown Boveri Strasse 6
5400 Baden
DEUTSCHES ZENTRUM FUER LUFT - UND R
Germany
Address
Linder Hohe
51147 Koln
DYNASIM AB
Sweden
Address
Research Park Ideon
223 70 Lund
KUKA ROBOTER GMBH
Germany
Address
Bluecherstrasse 144
86165 Augsburg
MATHCORE AB
Sweden
Address
Tekninkringen 1A:29
583 30 Linkoeping