Digital Furnace Twin

A digital furnace twin that contains specific knowledge about a thermal process executed in the furnace system in its program code, such as the mentioned thermo-process models of the HTL or other expert knowledge, can be used for improved automatic control of the system. The necessary correlations between furnace parameters and product quality can often be determined empirically by evaluating extensive production data, also using AI procedures. This is done, for example, in the ongoing joint project HTPgeox. Generally, for the digital twin to be used for furnace control, its computational speed must be increased to the extent that it can find a good response to occurring deviations in real-time, i.e., within a few seconds. To this end, suitable model reduction methods are adapted to the specific problems at the HTL (see DiMaWert project), so that the digital process and furnace models are also capable of meeting the future requirements of a fast and precise reaction to fluctuating energy supply (keyword "demand-side management").

Unlike the goal of furnace control through digital twins, the accuracy and reliability of the calculated predictions for furnace operation is crucial in designing models for the virtual development of novel thermal process devices, rather than speed. Therefore, the HTL has been developing the required methods for several years, such as ensuring the quality of material data or experimental validation of simulation results (see DiMaWert project). With the assurance of prediction quality and the identification of particularly critical parameters through sensitivity analyses, new thermal processes and furnaces can be developed in significantly less time and with less effort because some intermediate steps that are necessary in conventional construction, such as real-built demonstrators or prototypes, can be eliminated.