Towards a sustainable society, achieving carbon neutrality is globally required, and the expansion of renewable energy sources (RES) and the proliferation of battery electric vehicles (BEVs) are expected to accelerate further. However, RES pose challenges due to their instability and significant fluctuations in power output caused by weather conditions. In this context, it is effective to establish microgrids (MGs) at a regional level and ensure power supply stability by utilizing storage batteries to balance demand and RES supply within the grid.

On the other hand, installing storage batteries in small-scale kW-class MGs is generally expensive. To overcome this issue, previous research by the authors focused on a rural MG utilizing reused lithium-ion batteries (LIBs) retired from BEVs. The biogas (BG) output setpoint and battery capacity were investigated, and a balancing scheme that regulates BG power according to the battery state of charge (SOC) was proposed. The continuous year-long operation was evaluated via Model-in-the-Loop Simulation (MILS), in which all components, including BG, batteries, and loads, were modeled. However, a challenge remains in that MILS cannot fully reproduce actual battery behavior due to the characteristic differences between simulation models and real hardware.

Therefore, in this study, to more accurately validate the performance of the proposed balancing method, a test circuit was constructed using a real-time simulator and actual reused batteries. HILS was conducted continuously for 24 hours, and the results were compared with those of the MILS. The power consumption of a dairy farm consisting of milking machines, ventilation fans, and air conditioning units was assumed as the load, and the power demand was categorized into the intermediate and summer seasons. The rated output of the BG power plant was set at 25 kW, assuming a nearly constant output within the range of 50% to 100% of its rating. The capacity of the reused LIB used in this study was 64.35 kWh.

The HILS test circuit consists of a real-time simulator, a power supply unit, actual reused LIBs, and measuring transducers. Among the MG components, the BG power plant, loads, and the utility grid were represented by MATLAB/Simulink models running on the OP4510 real-time simulator manufactured by OPAL-RT. The charge/discharge current commands for LIB, calculated based on its SOC, were transmitted to the Imperix power supply via UDP communication. Based on these commands, the power supply unit exchanged charge/discharge currents with the actual LIB through a bidirectional DC/DC converter. For the demand data, data from April 28, a representative day of the intermediate season, was employed. Additionally, the initial SOC of LIB was set at 50%, and the simulation time step was 100 µs.

The results of comparing MILS and HILS confirmed that there are certain differences in the SOC of LIB, and the extent of these differences was quantitatively clarified. This comparison revealed that factors specific to real hardware, such as current sensor noise and transmission delays, contribute to SOC estimation errors and minute fluctuations in the power exchanged with the utility grid. The full paper will design the BG power plant output setpoint and battery capacity based on summer demand, compare the results with those for intermediate season demand, and report the findings.