Decentralized renewable energy (RE) generation and consumption through microgrids (MG), combined with short- and long-term storages and demand flexibility, present a promising approach for mitigating grid stress and reducing emissions in the industrial sector. This study aims to develop a simulation model framework that balances economic viability with the transition towards a sustainable industry, particularly regarding RE and its storage. The framework focuses on three key research areas: MG system sizing, optimal energy allocation, and demand flexibility. This framework is used for modeling multi-agent simulation models, which serve as the basis for metaheuristic-based optimization and enable a comprehensive approach to overall optimization. In an evaluation, a simulation model was established through a practical demonstration in a research factory environment, using historical data for MG system sizing and optimal energy allocation. The production was then scheduled based on RE availability. The results of the metaheuristic optimization demonstrate that the proposed methodology reduces costs by 18% and emissions by 3% compared to conventional, non-multi-agent optimization approaches.