Generic Energy Flow Model for application in optimized combined operation of a Battery Energy Storage System and an Electrolyzer

The research project “ZET Reallabor Wunsiedel” collects data from different power plants at Wunsiedel energy park. The main goal of the project is to find a suitable strategy for the combined economical operation of an electrolyzer and a battery energy storage system (BESS). This optimization is achieved by a dynamic programming algorithm according to Bellman. Dynamic models with low complexity and an efficient management of working memory are a necessity for optimization algorithms with high computational effort. Therefore, a generic energy flow model has been developed within the framework of this project.

This model utilizes only information about the input power and emulates the behavior of the BESS with an approach based on control technology modelling methods and efficiency maps. Losses originating from conversion, energy management, standby, dimensioning, and control losses are considered by this approach. The parameters need to be fitted once for establishing the model. Afterwards simulations can be run, and the system efficiency is adjusted by interpolation of efficiency maps. Those efficiency maps are derived from pulse tests and characterize overpotential losses during operation. Consequently, the efficiency values need to be determined for different power inputs/outputs and different states of energie (SOE).

A new measurement procedure was developed to derive those values and the measurements were carried out for NCA cells. The model is upscaled to match the dimensions and characteristics of the system installed in Wunsiedel. Combined with a suitable model of the electrolyzer on site, dynamic programming according to Bellman is used to assess the optimal operational strategy for the BESS and the electrolyzer. The optimization is carried out aiming at a maximum economic gain. The real day ahead prices of the German energy market are used as input parameters with an optimization period one week using a predefined amount of hydrogen that must be produced by the end of the week. The battery can compensate the dynamics of the energy prizes and ensure constant operation of the electrolyzer even during periods with high energy prices where a stand-alone operation of the electrolyzer would not be cost efficient. Additionally, an aging model for both models is implemented to account for the economical disadvantages of stack degradation and quantifying the investment in form of equivalent operational hours (EOH). In case of the battery, degradation is quantified by full equivalent cycles (FEC).

With this optimization algorithm the electricity procurement costs per ton H₂ can be reduced by 15 % while aging in form of EOHs is reduced by 13 % due to the combined operation of both systems.