A key advantage of WAAM is the ability to weld molded elements onto existing structures, for example to repair worn turbine or compressor blades. However, dimensional accuracy in WAAM processes is limited due to residual stresses and resulting deformations after production.
The high economic potential of WAAM has led to a large number of research projects in recent years. Previous research has focused primarily on the influence of process control on material properties and cooling rates. Technology-based calculation of workpiece geometry has only been carried out in a few studies to date.
In order to determine the workpiece geometry, the process behaviour must be modeled. For this purpose, SimProWAAM uses physically informed neural networks (PINNs), which take into account not only empirical test data but also known analytical relationships.
Multiscale simulations transfer microscale models to the entire workpiece, enabling precise predictions of macrogeometry. Currently, there are no approaches known that use this cross-scale modeling for WAAM process planning.
The project aims to enable the simulation-based planning of real WAAM processes. Using the MIG/MAG WAAM process as an example, a multiscale process model is being developed that will support the cost-effective production of individual parts in the future.
In addition, the project contributes to ecological sustainability goals by significantly reducing the amount of material, energy, and time required for preliminary testing. Instead of costly physical tests, many parameters and scenarios can be tested virtually. Simulations enable precise process planning that minimises material consumption and uses resources efficiently.
Contact:
For further information, please contact Nan Zhou at +49 511 762 5209 or by email at n.zhou@ifw.uni-hannover.de.
