On consistency of physical and DEVS models in control-targeted DTs: an industrial case study

The DT-based design and control of modern production assets requires simulation models with different viewpoints, purposes, and nature. Most notably, component-level design and control require detailed physical models, while plant-level automation and production management mostly need DEVS-type ones. Inconsistencies among the said models can have significant impacts on decision processes related e.g. to (re-)configuration, diagnostics and maintenance—in one word, on the ultimate outcome of an asset. However, given their heterogeneous nature, making physical and DEVS models consistent requires new methods and tools, in fact starting with the definition itself of what “consistent” has to mean. We here present a methodology to state and address such consistency problems, and support our statements with a case study taken from the cosmetics industry.