Dynamic model based state-estimation approach for robomoulding

Robomoulding, robotized rotational moulding, is a novel polymer processing technique allowing to produce hollow complex parts in a more robust and economical way as compared to traditional rotational moulding. Polymer powder is place in the mould, which is mounted on a robotic arm, and is electrically heated and rotated, mostly using a rock-and-roll motion. The powder gets distributed along the mould and solidifies into a stress-free part. In order for the robomoulding process to break through in automotive applications, the process should be better controlled and understood. Currently, there are no commercial software tools available that allow to predict the quality of the part being formed and the influence of the temperature cycle and the movement of the mould. Over-simplified analytical first principle models are available but lack the required granularity for accurate in-process monitoring. It is known, however, that both the motion, as well as the heating play a crucial role in the final process performance and should be included in the model. This PhD project aims to further unravel the robomoulding process, by relying on a digital twin approach which combines the semi-analytical models together with novel higher order motion and thermal models, and updated through (operational) measurement data. A state-estimation framework will be developed to enable this integration and gain more insight in the operational behaviour of this process.