Utilizing Experimental Data in a Model‐Free Data‐Driven Method: The Case of Thermal Conduction

ABSTRACT The study at hand investigates the applicability of experimental data within a model‐free data‐driven solver. The heat conduction equation is facilitated and solved by the data‐driven solver without the need for a constitutive law. At first, the results of an artificially created dataset are compared to a finite element solution to assess the quality of the approach. Subsequently, in‐situ data is processed to create a dataset for the data‐driven solver. This approach is validated by comparison to experimental data. The quality of the measured data is discussed, and challenges arising in utilizing experimental data are pointed out. Furthermore, an improved measuring concept for future experimental data is detailed and tested based on finite element simulations. Another finding is the ability of the data‐driven solver to implicitly model changing material characteristics.