Hybrid photovoltaic–battery energy storage system (PV-BESS) power plants are emerging as strategic grid-friendly assets, combining variable renewable generation with short-term flexibility. This case study focuses on a full year of operation of two French hybrid sites—8 MWp PV plus 3.8 MW 1h BESS and 18 MWp plus 7.5 MW 1h BESS—each subject to grid-connection contracts limiting active-power export to 50% of total capacity.

Goal: Maximise photovoltaic production revenues while complying with strict export caps and network operators grid codes, and extending battery lifetime.

Results: Over the 12-month period, battery economic performance ranged from 75% to 90%, of a perfect forecast scenario. The most remarkable result is that the hybrid configuration allows retains over 92% of the gross margin generated by a configuration of standalones PV and BESS, while the injection max power capacity contracted with the electricity network operator in the hybrid configuration is 50% of the injection max power capacity contracted in equivalent standalone configuration.

Methods: The optimization is based on hierarchical, multi-scale At Level 2, the Unit Commitment process is structured as a two-level decision-making framework. The first level relies on a day-ahead dispatch strategy based on dynamic programming. The second level involves a mixed-integer linear programming (MILP) optimisation, which co-optimises energy arbitrage, ancillary service provision, and degradation cost minimisation, under both price and meteorological uncertainty. These two-steps decision processes are fed by a probabilistic set of weather scenarios. The second step of the Level 2 optimization process can be repeated during intraday (Level 1). Level 0 governs real-time set point tracking through decentralised control logic implemented on edge-level hardware, ensuring compliance with unit commitment schedule, power-quality and fault-resilience requirements.

The hybrid PV-BESS system model combines two complementary components. First, a machine-learning surrogate model captures photovoltaic performance as a function of solar irradiance and geometrical positioning. Second, the physics-based dynamic programming and MILP formulations account for the site's intrinsic operational constraints. Battery degradation is evaluated through a laboratory-calibrated electrochemical ageing model, enabling degradation-aware optimisation across the entire control stack. To address the sensitivity of dispatch decisions to forecast uncertainty, the framework uses a dynamically updated distribution of solar forecasts rather than single deterministic inputs.

Relevance of the subject: This study presents a field-validated, multi-objective optimisation framework for PV-battery systems under distribution grid constraints, integrating economic, technical, and environmental objectives. The architecture accounts for climatic variability, supports co-optimisation, uncertainty management, and degradation-aware control, and aligns with evolving network operators’ grid codes. Its open-source, standard-compliant design ensures replicability and scalability across constrained distribution networks

In conclusion, this study demonstrates the strategic role of hybrid systems in mitigating variable renewable generation (VER) and in providing ancillary services, reinforcing their role as a novel asset category in the evolution of power system architectures.