Implicit Incorporation of Heuristics in MPC-Based Control of a Hydrogen Plant

TL;DR

This work tackles efficient operation of a hydrogen-plant EMS under renewable variability by embedding complex fueling heuristics into a convex MPC framework using an allocator that validates MPC plans against a high-fidelity heuristic model. The allocator adds constraints to implicitly enforce tank-pressure and fueling requirements while preserving tractability, enabling a seven-day horizon with $N_{pred}=35$ and a nonuniform sampling scheme $\boldsymbol{T}_s$. Through simulations with real Offenbach facility data, the proposed MPC approach achieves 100% fueling success and lower electricity costs compared with two rule-based benchmarks, while maintaining competitive hydrogen production costs and lower CO$_2$ emissions. This approach demonstrates practical potential for microgrid hydrogen management, offering a scalable method to reduce peak charges and emissions and to improve fueling reliability; future work includes hardware validation and digital-twin testing, with explicit reference to the horizon $7$ days and the decision variables $\boldsymbol{u}$ and $\boldsymbol{u}_{aux}$ in the OCP.

Abstract

The replacement of fossil fuels in combination with an increasing share of renewable energy sources leads to an increased focus on decentralized microgrids. One option is the local production of green hydrogen in combination with fuel cell vehicles (FCVs). In this paper, we develop a control strategy based on Model Predictive Control (MPC) for an energy management system (EMS) of a hydrogen plant, which is currently under installation in Offenbach, Germany. The plant includes an electrolyzer, a compressor, a low pressure storage tank, and six medium pressure storage tanks with complex heuristic physical coupling during the filling and extraction of hydrogen. Since these heuristics are too complex to be incorporated into the optimal control problem (OCP) explicitly, we propose a novel approach to do so implicitly. First, the MPC is executed without considering them. Then, the so-called allocator uses a heuristic model (of arbitrary complexity) to verify whether the MPC's plan is valid. If not, it introduces additional constraints to the MPC's OCP to implicitly respect the tanks' pressure levels. The MPC is executed again and the new plan is applied to the plant. Simulation results with real-world measurement data of the facility's energy management and realistic fueling scenarios show its advantages over rule-based control.

Figures (3)

• Figure 1: The electrolyzer directly fills the tank of max. $11\,\mathrm{kg}\xspace / 30\,\mathrm{bar}\xspace$ capacity. A compressor is then used to transfer the hydrogen to 6 individual tanks with max. $43.33\,\mathrm{kg}\xspace / 450\,\mathrm{bar}\xspace$ capacity each, which are organized in 2 sections with 3 tanks each. The compressor can also be used to shift hydrogen between these sections (pressure recovery). The (dispenser) is connected to the tanks and used to refuel with $350\,\mathrm{bar}\xspace$.
• Figure 2: The allocator uses a second, heuristic simulation model, which models all 6 tanks individually. Using the input variables determined by the solving \ref{['eq:opt_prob_mpc_without_allocator']}, it checks whether all planned car refueling events would be successful, i.e. if at least one tank with enough pressure is present. If not, it sets the parameters for the constraints \ref{['eq:tpr']}--\ref{['eq:mp_lim_alloc']} and the runs a second time. Afterwards, the first input variables of the updated plan are sent to the plant. In this work, we use the same heuristic simulation model as in the allocator as the plant. Finally, the state is updated and the entire process repeated for the next time step.
• Figure 3: Electricity costs per $\mathrm{kg}\xspace$ hydrogen, fueling success rate and PV self consumption in each month of the simulated year 2021 for the proposed MPC approach, the RBC Excess controller and the RBC Peak controller.