Toward More Realistic Simulation Models Using Object-Centric Process Mining

Abstract

Discrete-event simulation has been around for over half a century with applications in production, healthcare, logistics, transportation, etc. However, it is still challenging to create a reliable simulation model that mimics the actual process well and allows for “what-if” questions. Process mining allows for the automated discovery of stochastic process models using event data extracted from information systems. This technology is one of the key enablers for creating digital shadows and digital twins of operational processes. However, traditional process mining focuses on individual cases (e.g., an order, a patient, or a train) with events just referring to a single object (the case). Therefore, the discipline is moving to Object-Centric Process Mining (OCPM), where events can refer to any number of objects. Based on research on OCPM and the prototypes developed, now also commercial software vendors are embracing OCPM, as illustrated by Celonis Process Sphere, which allows for the discovery and analysis of object-centric process models. We believe that OCPM will help to create much more realistic simulation models. Whereas process discovery is backward-looking, with object-centric simulation models, we can also support forward-looking forms of process mining. Although such techniques still need to be developed, they provide a unique opportunity to create more realistic digital twins of organizations and their processes.

Authors 1

1. Wil M. P. van der Aalst corresponding Aachen Process and Data Science (PADS)

   RWTH Aachen University

   Affiliation as printed

   Process and Data Science (PADS) RWTH Aachen University D-52056 Aachen , Germany

   Process and Data Science (PADS) RWTH Aachen University D-52056 Aachen, Germany