Interoperable collaborative manufacturing process simulation for digital twins

Authors: De Vrieze, P., Arshad, R. and Xu, L.

Journal: Computers in Industry

Publisher: Elsevier

ISSN: 0166-3615


There are various definitions and concepts of Digital Twin in use. Independently it is clear, however, that DT's play a big role in the operation and management of smart factories. In some cases, an entire production line is seen as a digital twin. Even in that case, the production line would be composed of multiple components and corresponding digital twins. When going beyond single production lines to collaborative manufacturing there is still the need to allow for ad hoc integration, again confirming the need to be able to integrate digital twins.

Integration requires interoperability. While there is a lot of work on interoperability this tends to focus on semantic interoperability of the information shared by the twins. Where digital twins are used in real-time to augment physical twins the semantic interoperability can be sufficient, but when they are used for simulation with a longer time horizon, and possibly across organisational boundaries, the simulations need to be integrated themselves.

While integration simulation starts with semantic interoperability of the information exchange (such as specified by IEC21823) this is insufficient. Looking at the information exchange between the states of various digital twins (and the influence of physical objects upon each other) we observe that generally there is a cyclic information dependency between the components. As a consequence, a sequential execution of simulators is not possible, but coordinated simulation is needed.

There are different ways to integrate multiple simulations. The first is to have a single simulator where the simulator descriptions are combined on a semantic level. The second approach is to run simulations together in real or compressed time. The third, most comprehensive approach is to use federated simulation that includes an explicit synchronisation mechanism between the simulators.

A single monolithic simulator has issues both with confidentiality, but crucially also requires a high level of compatibility between the simulation definitions and the simulator, both in semantics as well as execution model. Using real-world time (compressed or not) to synchronise is possible, but requires similar time compression and, by its nature, does not easily lend itself to time efficient simulation. Federated simulation, splitting up the simulations into coordinated slices does allow various simulation approaches to coordinated, at the cost of requiring the component simulators to expose parts of their internal workings to enable the coordination.

In this paper we will present a solution that allows digital twins to enable their use within such federated simulation, providing a building block towards context-aware, autonomous, and adaptive simulation blocks for tomorrow's Digital Factories.

Source: Manual