In manufacturing, digital twins, realized as Asset Administration Shells (AAS), have emerged as a prevalent practice. These digital replicas, often utilized as structured repositories of asset-related data, facilitate interoperability across diverse systems. However, extant approaches treat the AAS as a static information model, lacking support for dynamic service integration and system adaptation. The existing body of literature has not yet thoroughly explored the potential for integrating executable behavior, particularly in the form of containerized services, into or from the AAS. This integration could serve to enable proactive functionality. In this paper, we propose a submodel-based architecture that introduces a structured service notion to the AAS, enabling services to dynamically interact with and adapt AAS instances at runtime. This concept is implemented through the extension of a submodel with behavioral definitions, resulting in a modular event-driven architecture capable of deploying containerized services based on embedded trigger conditions. The approach is illustrated through a case study on a 3-axis milling machine. Our contribution enables the AAS to serve not only as a passive digital representation but also as an active interface for executing added-value services.

Abstract Engineering digital twins is a software and systems engineering challenge for which no systematic approach exists. The Asset Administration Shell is becoming a popular foundation for digital twins in Industry 4.0 and it comes in different types that support the engineering of different kinds and parts of digital twins. We investigate how it supports realizing common requirements for digital twins. To this end, we investigate how each of the three Asset Administration Shell types can contribute to the systematic engineering of specific components of digital twins. Therefore, we analyzed popular definitions and conceptual models of digital twins and extracted requirements that at least two of them share. We compare the resulting requirements with Asset Administration Shells of different types and conclude with open challenges in the implementation of digital twins with this technology. This supports practitioners and researchers in identifying the most suitable type of Asset Administration Shell for their specific digital twin engineering needs and identifies gaps worthy of future research toward a systematic engineering of digital twins.