Modeling of hydrogen production system for photovoltaic power generation and capacity optimization of energy storage system

Daohong Wei, Huawei Li, Yan Ren, Xianhe Yao, Long Wang, Kunyong JinFrontiers in Energy Research 2022 · DOI

Abstract

Hydrogen production using solar energy is a critical pathway for sustainable energy. However, the inherent intermittency and randomness of solar power reduce hydrogen production efficiency. This study establishes a mathematical model for a photovoltaic (PV) hydrogen production system and optimizes its energy storage capacity. Using Matlab/Simulink, the system is modeled and analyzed. We propose a storage capacity configuration algorithm based on low-pass filtering principles and optimal time constant selection. A case study of a 30 MW PV plant demonstrates that the algorithm effectively reduces power variation. Results show that decreasing the cut-off frequency increases storage capacity while significantly improving the smoothing effect, reducing 1-hour maximum power variation from 4.31 MW to as low as 0.04 MW.

Introduction: The Intermittency Challenge

Green hydrogen production via solar energy faces a fundamental hurdle: intermittency. Solar irradiance is inherently random and volatile; feeding raw, fluctuating power directly into an electrolyzer compromises hydrogen production efficiency and can strain the electrical grid. To mitigate this, Energy Storage Systems (ESS), such as batteries, are integrated to act as buffers.

The challenge lies in capacity optimization—balancing the need for sufficient power smoothing against the economic costs of over-sized storage. This paper addresses the challenge of finding the optimal storage capacity required to satisfy grid-connection standards while maintaining a steady power supply for hydrogen production.

System Modeling and Methodology

The researchers developed a multi-stage simulation framework using Matlab/Simulink and Python. The system architecture, as shown in Figure 1, utilizes a direct DC coupling design to minimize energy conversion losses and simplify the overall architecture.

Figure 1: System structure diagram showing the integration of PV array, DC bus, PEM electrolyzer, and energy storage.

PV and Electrolyzer Modeling

The PV model employs an equivalent circuit that accounts for temperature and irradiance corrections. The non-linear relationship between voltage and power output is critical for accurate simulation, as illustrated by the U-P characteristic curves at different temperatures (Figure 2). As temperature increases, the peak power point shifts, requiring robust control strategies.

Figure 2: U-P characteristic curve of the PV module at 25°C and 45°C.

The system also includes a Proton Exchange Membrane (PEM) electrolyzer model based on U-I characteristic equations and an ideal pressure model for hydrogen storage and compression.

The Smoothing Algorithm

The core innovation is a Low-Pass Filter (LPF) strategy. By treating volatile PV output as a high-frequency signal, the LPF removes the "noise" (sudden drops or spikes). The filter's time constant ($\tau$) is the primary lever for optimization. The authors optimized $\tau$ based on the Chinese National Standard (GB/T 19964-2012), which mandates that a PV plant's power variation must not exceed 10% of its installed capacity per minute.

Key Results and Discussion

The simulation utilized a real-world dataset consisting of 8,760 hours of irradiance and temperature data for a 30 MW installed capacity PV plant. The results validated the effectiveness of the LPF approach:

  • Volatility Reduction: Before smoothing, the 1-hour maximum power variation was 4.31 MW. By adjusting the filter's cut-off frequency, this variation was reduced to as low as 0.04 MW—a 99% reduction.
  • Optimization Trade-offs: As the time constant increased, the power curve became smoother, but the required energy storage capacity grew proportionally. For the 30 MW plant, required storage ranged from 6.84 MWh to 7.11 MWh to meet different smoothing targets.
  • Regulatory Compliance: The study found that a time constant greater than 750 seconds was necessary to satisfy the 10%-per-minute grid variation limit for this specific site.

Conclusion

This research provides a robust framework for sizing energy storage in PV-hydrogen systems. By applying low-pass filtering, developers can precisely determine the minimum storage capacity needed to ensure grid stability and efficient hydrogen production. While the current model focuses on technical optimization, future work will incorporate economic factors, such as the Levelized Cost of Hydrogen (LCOH), and consider battery degradation to balance system reliability with capital investment.

Summary updated

Abstract

Hydrogen production using solar energy is an important way to obtain hydrogen energy. However, the inherent intermittent and random characteristics of solar energy reduce the efficiency of hydrogen production. Therefore, it is necessary to add an energy storage system to the photovoltaic power hydrogen production system. This paper establishes a model of a photovoltaic power generation hydrogen system and optimizes the capacity configuration. Firstly, the mathematical model is modeled and analyzed, and the system is modeled using Matlab/Simulink; secondly, the principle of optimal configuration of energy storage capacity is analyzed to determine the optimization strategy, we propose the storage capacity configuration algorithm based on the low-pass filtering principle, and optimal time constant selection; finally, a case study is conducted, whose photovoltaic installed capacity of 30 MW, verifying the effectiveness of the proposed algorithm, analyzing the relationship between energy storage capacity and smoothing effect. The results show that as the cut-off frequency decreases, the energy storage capacity increases and the smoothing effect is more obvious. The proposed algorithm can effectively reduce the 1 h maximum power variation of PV power generation. In which the maximum power variation of PV generation 1 h before smoothing is 4.31 MW. We set four different sets of time constants, the maximum power variation of PV generation 1 h after smoothing is reduced to 0.751, 0.389, 0.078, and 0.04 MW, respectively.
BibTeX
@inproceedings{wei2022modeling,
  author = {Daohong Wei and Huawei Li and Yan Ren and Xianhe Yao and Long Wang and Kunyong Jin},
  title = {{Modeling of hydrogen production system for photovoltaic power generation and capacity optimization of energy storage system}},
  booktitle = {Frontiers in Energy Research},
  year = {2022},
  doi = {10.3389/fenrg.2022.1004277},
}
Figure