Even the use of simulations has increased dramatically
over the past 20 years. The models have become more accurate, the
algorithms faster and the computers are more powerful.
We have been heavily focused on simulation and optimization for many years and are of course, very pleased with Vinnova's financing of the project “Digital Twin for Geometry Assured Production” (DigiGeo). The funding makes it possible to increase the use of simulation to mirror and control proce
sses in real t
ime, such as digital twins, says Rikard Söderberg
, Professor of Product and Production Development, and continues:
- We hope to be able to use the method in early development phases to simulate and visualize the effects of geometric variation.
A digital twin is an online virtual copy of a real product or process. It can have different purposes and is therefore supplied with various real-time data. The digital twin is connected wirelessly to databases and physical devices and is predicted to be common in future production systems.
At the same time, the development of simulation models for geometry assurance has been going on for many years. It has gone from simpler simulation where details have been assumed rigid, to more advanced finite element (FEM) based simulations.
- The work of recent years has been aimed at driving simulation models for non-rigid geometry as digital twins in the future. It could mean an ability to control both the production process and product development in a more efficient way, says Professor Söderberg.
Faster algoritms demands
However, in order to fully exploit the simulation models for geometry assurance as digital twins - both in production and in early stages of development - it is necessary to develop further. The project is therefore divided into three parts:
The first part project focuses on better material models and finding faster algorithms for real-time use. This means refinement of the digital twin to handle more advanced material properties as multi-material solutions, i.e. solutions in which different materials are used in the same subassembly or the same component. Residual stresses, depending, for example welding details, also affects the geometrical deviation and variation and must therefore also be included in the simulations.
- This means more complex and demanding calculations. Therefore, we need to develop high-speed algorithms for real-time use, says Söderberg.
Visualization necessary for decisionmaking
To use the digital twin as a decision tool, you need to visualize the variation as realistic as possible. Therefore, in the second part of the project, you enhance the simulation models with Raytracing, 3D scanning / point clouds and digital environments.
Raytracing means increased realism when visualizing manufacturing variation and can be used to support decision-making in the early stages. Scanning results as point clouds are currently used to measure and verify components and products with great accuracy. Point clouds can be used to compare variations in batch details, but the method requires faster and more efficient algorithms due to the large amount of data.
Finally, there is a need for models of digital environments visualizing the impact of manufacturing variations on perceived quality.
- To effectively visualize geometric variation, models are required, as well as design and data to be combined into scenarios that then are evaluated. If we manage to automatically generate those models, we would significantly rationalize the geometry insurance process. We benefit greatly from the research on visual evaluation conducted by the department within the research area Percieved Quality, says Rikard Söderberg.
In the third and final subproject, a preliminary study of a digital information flow, a so-called “Digital Thread”, is being planned. It will predict how the future digital geometry assurance process might look.
Industrial co-operation and advanced mathematics
The DigiGeo project, which started at the beginning of the year, extends over two years and brings together academics, research centers and industry.
- We carry out this project in cooperation with Fraunhofer Chalmers Center, who has the required advanced competence in industrial mathematics for the development of algorithms, says Söderberg, and continues:
- The project would not be possible without collaboration with industry partners such as Volvo Cars, IKEA, GKN, VA Automotive and RD&T Technology.
Text: Carina Schultz
Photo: Shutterstock (large image) and Anna-Lena Lundqvist (portrait)
Project leader: Professor Rikard Söderberg, Department of Industrial and Materials Sciences
Project time: 2018-2019