Design and Evaluation of Vertical Scalability Strategies for Deploying Large-scale Co-simulations

Multi-Energy Systems (MES) represent a paradigm in which various energy systems, such as buildings, power grids, and heating networks, are integrated to operate in a cohesive manner. These systems provide a significant opportunity to improve technical, economic, and environmental outcomes. However, the complexity of MES poses challenges in accurately capturing their interdisciplinary interactions using standalone simulations, which often do not adequately model their interconnected nature. Co-simulation frameworks have emerged as a promising solution to this challenge, enabling the coordination of multiple simulators within a single scenario. However, computationally intensive simulators can often hinder scalability, especially as the complexity of the simulations increases. This paper proposes a methodology to apply vertical scalability techniques to simulators and optimise each single-node performance in co-simulation frameworks. Using COESI, a Mosaik-based co-simulation platform, it was possible to systematically test these techniques by increasing the number of model entities a simulator class must manage in a complex MES scenario. This study highlights selective parallelisation as a key strategy for optimising computational efficiency in MES co-simulation, offering insights to scale complex energy systems effectively. The results demonstrate significant performance gains, particularly for CPU-intensive tasks with stateless simulators, such as executing intensive numerical computations.