Abstract

Products are becoming smarter having more sensorics and more complex control-logic incorporated, allowing them to more actively interact with the environment. Simulating such smart products becomes a challenge as the simulation models becomes more complex as also the whole interaction with the environment and the controller logic needs to be represented in the simulation model. With this added complexity, also the number of load cases and scenarios will increase as more and more interaction patterns and behaviors of the model need to be simulated. For autonomous system this can become an extremely huge number, actually getting to a point, where the identification (such as finding all the edge cases9of all possible situation is in-itself an simulation exercise. As companies adopt a more agile development process for such smart products, there is a growing demand to test and validated the development continuously (CI/CD; Continuous Integration, Continuous Delivery and Continuous Deployment)), especially the control-logic and a high frequencies which demands continuous testing and validation of the product throughout the development. To handle the sheer amount of simulations, all the different models of components and environment, the scenarios and load cases and the need to conduct simulation more automatically with every change in the design (particularly of the control-logic) makes it almost mandatory to have an simulation data and process management (SPDM) solution. In this paper we will outline all the main architectural challenges to support this of such an SPDM system. We will cover: • Management of the different model component constituting a simulation model. These models will be generated by different groups, which adds significant challenges in terms of collaboration, version and revisions management and automated integration to the final simulations model. • Providing an execution environment, where multiple (potentially 100’s to 1000’s) instances of complex co-simulation can be performed. • Having an result management environment which then aggregated the results and their KPIs into a form allowing dashboarding, analysis tools and AI/ML methods to easily process the data.