Alexander Michael Kuhn

Since the early 1990s, virtual models have been used in production planning to digitally support production lines before beginning physical construction. These digital models, commonly referred to as “digital twins”, are currently primarily used for virtual commissioning. Despite being first used decades ago, digital twins are still not established in production planning on a global scale. The benefits of developing and testing a planned system in a virtual model are often not fully capitalized. While this can be partially explained by the technological challenges of designing accurate virtual models, we argue that the current processes for production planning are another important factor that hinder the widespread use of digital twins. In this paper, we summarize and analyze each step and the involved participants in a typical production planning workflow. Based on this analysis, we discuss how current practices conflict with the goal of incorporating virtual models into the established work environment. Then, we derive concrete recommendations of how a production process can be adjusted to allow easier digital twinning, showing how comparatively few changes could make virtual models significantly easier to apply.

Digital factories and smart manufacturing systems have been increasingly researched and multiple concepts were developed to cope with prevailing ever-shortening life-cycles. The ubiquitous digital twin, despite many definitions, is often praised for accurate virtual models. One key idea to improve manufacturing through such virtual models is virtual commissioning (VC), aiming at early machine code validation. VC and its virtual models are still lacking behind their real counterparts. This gap between reality and its virtual model, commonly termed reality gap, increases the complexity of creating cyber-physical systems. An especially stark contrast is visible between the idealized virtual model and a real machine encountering errors. While error simulations exist in other fields of research, a thorough investigation in VC is missing. Thus, this paper addresses the task of narrowing the reality gap in VC based on two steps. First, a comprehensive body of research of possible errors encountered in virtual commissioning is analyzed. Secondly, the feasibility of error implementation is discussed. This paper lays the foundation for narrowing the reality gap and enabling test automation and digital twin-based control.