Scheduling of aggregated energy systems using a distributed augmented Lagrangian based alternating direction inexact Newton optimization algorithm

This work investigates the suitability and applicability of the derivative-free version of the distributed augmented Lagrangian based alternating direction inexact Newton (AL-ADIN) optimization algorithm for distributed weekly scheduling in aggregated energy systems. A hierarchical setup is considered, in which an aggregator acts as a system-level coordinator with a peak-shaving objective, while prosumers seek to minimize their time-of-use electricity costs. The prosumers are modeled with both flexible and inflexible electrical and thermal components, including photovoltaic units, battery storage systems, heat pumps, and thermal energy storage. The proposed ALADIN algorithm is applied to a focused simulation test case featuring prosumer portfolios of varying sizes. In this context, the impact of three key algorithmic parameters on ALADIN's performance is systematically analyzed. Using the right combination of parameters, the results show that the algorithm converges quickly, producing weekly schedules and objective values identical to those of a centralized reference benchmark optimization. However, the coupling quadratic problem of the ALADIN algorithm emerges as a critical step, with computing times being up to 33.6 times longer than those required for solving individual prosumer subproblems. This limitation reveals challenges in scalability of the algorithm for large-scale system setups.