Statusvortrag im Rahmen des Promotionsver­fahrens von M. Sc. Volodymyr Prokopets

Title: „Variability Management for Dynamic Digital Twin Simulations”

Abstract:
Digital Twin (DT) engineering currently faces a "complexity wall" we refer to as the n-twin problem. It can be described as the challenge of managing multiple heterogeneous instances of a common asset class (e.g., vehicles in a road network), where each twin must undergo independent, state-preserving runtime adaptations based on its unique physical context. Currently, maintaining these individualized models remains an ad-hoc and unscalable process.

This work presents the Mosaic Digital Twin (MDT) architecture, a formal approach to managing n-twin DTs leveraging software product line engineering principles. Central to this research is the Waffle Workbench, a framework that introduces a constraint-based staged configuration process. To overcome the limitations of static templates, Waffle utilizes a domain-specific language backed by an SMT-based derivation engine to derive valid DT instances. By organizing feature dependencies into a directed acyclic graph, we enable a novel incremental reconfiguration mechanism that allows the system to identify "affected islands" of constraints, triggering localized SMT-solving to update twin structures in real-time (e.g., swapping thermo-mechanical tire-road interaction simulations based on current road structure and environmental conditions) without the overhead of full-model recomputation.

To validate the architectural feasibility, we demonstrate the current implementation's integration with the FIWARE ecosystem, bridging the gap between high-level variability models and low-level NGSI-LD context entities. The talk concludes with initial results from a highway infrastructure case study, illustrating how the MDT framework resolves the trade-off between high-fidelity individualization and system-wide scalability.

Betreuer: Prof. Uwe Aßmann
Fachreferent: Prof. Dirk Habich