The rise in hydrogen production powered by renewable energy is driving the field toward the adoption of systems comprising multiple alkaline water electrolyzers. These setups present various operational modes: independent operation and multielectrolyzer parallelization, each with distinct advantages and challenges. This study introduces an innovative configuration that incorporates a mutual lye mixer among electrolyzers, establishing a weakly coupled system that combines the advantages of two modes. This approach enables efficient heat utilization for faster hotstartup and maintains heat conservation postlye interconnection, while preserving the option for independent operation after decoupling. A specialized thermal exchange model is developed for this topology, according to the dynamics of the lye mixer. The study further details startup procedures and proposes optimized control strategies tailored to this structural design. Waste heat from the caustic fully heats up the multiple electrolyzers connected to the lye mixing system, enabling a rapid hot start to enhance the system’s ability to track renewable energy. A control strategy is established to reduce heat loss and increase startup speed, and the optimal valve openings of the diverter valve and the manifold valve are determined. Simulation results indicate a considerable enhancement in operational efficiency, marked by an 18.28% improvement in startup speed and a 6.11% reduction in startup energy consumption in multielectrolyzer cluster systems, particularly when the systems are synchronized with photovoltaic energy sources. The findings represent a significant stride toward efficient and sustainable hydrogen production, offering a promising path for largescale integration of renewable energy.
In March 2022, the National Development and Reform Commission, along with the National Energy Administration of the People’s Republic of China, issued the “Medium and LongTerm Plan for the Development of the Hydrogen Energy Industry (2021–2035).”
This plan delineates the strategic development of the hydrogen energy industry at a national level, setting a target to reach a renewable hydrogen production capacity of 100,000 to 200,000 tons annually by 2025 in China. By the first nine months of 2023, China had 57 green hydrogen projects operational, under construction, or awaiting approval [
According to the current advancements in water electrolyzers, alkaline electrolysis stands out as the primary method for largescale hydrogen production [
Typically, a multielectrolyzer cluster operates in two modes: standalone and integrated. The standalone mode is advantageous for its simplicity in equipment manufacturing and installation. The focus on cluster control strategies is to enhance operational performance within this structure. Various studies have contributed to this area: Shen et al. introduced a rotational control strategy to equalize electrolyzer usage, thereby extending lifespan and safety [
Compared with the standalone mode, the integrated operation mode of multielectrolyzer clusters is increasingly preferred in the industry [
Both the strongly coupled and completely independent forms have their advantages and disadvantages. This paper presents a novel weakly coupled operation mode that requires minimal equipment modification. The transition between independent and coupled operations can be achieved through valve switching, enabling full utilization of energy–matter interaction while retaining the ability for decoupled and independent operation, thus combining the advantages of both forms. Moreover, it does not affect the pressure and liquid level of the separator equipped with the electrolyzer, ensuring efficient separator operation. A thermal model specific to this topology is established, and optimized startup procedures and control strategies are proposed. The system modeling and optimization solution for a threestack system are completed, and the effects of topology and control strategy are verified.
The proposed weak coupling in the multielectrolyzer cluster is achieved through the integration of a lye mixer connected to each electrolyzer. The addition of a lye mixer does not modify the original lye circulation system within individual electrolyzers, as illustrated in
The core of the lye mixing system is the mixer, which contains lye. Its volume needs to be designed according to the installed capacity and the amount of lye circulation; the larger the volume, the more lye reserves can be available for sharing. However, this also increases the demand for heat within the system.
The temperature of the stack (
The formula for
The mass flow rate of lye, denoted by
With the assumption of no heat dissipation in the pipeline, the temperature at the outlet of the cooler is approximately
The
It is assumed that there is no heat loss in the pipeline from the separator outlet to the confluence valve. Thus,
The flow rate of lye at the confluence value outlet should be equal to that of the separator outlet to maintain a constant volume of internal circulating lye within the stack. Therefore, the opening degree of the confluence valve varies with changes in the divider valve.
The temperature of the lye in the mixer,
The formula for calculating
The formula for calculating
The formula for calculating
The temperature of the lye at the outlet of the separator,
With the assumption that heat dissipation from the pipeline between the stack outlet and the separator is negligible and that the temperature of the lye at the inlet of the separator is still
The dissipation term
The formula for calculating
The formula for calculating
As water supplementation typically occurs in the separator on the hydrogen side,
The temperature of the desalination water is represented by
Water consumption primarily results from water electrolysis and the carrying away of water with the gas. This can be expressed by multiplying the consumption of water electrolysis by a certain coefficient ξ. The separator typically has upper and lower water level limits, and water supplementation is not continuous but initiated when the lower water level is reached. For simplicity, it is assumed that water supplementation occurs simultaneously with water consumption. Therefore, the mass flow rate of desalinated water is equivalent to the water consumption:
According to the given formulas, the temperature changes in major components can be calculated, as illustrated in
First, a stack with small capacity is chosen for the first startup to facilitate rapid heat generation and temperature rise. The flow directed to the mixer from stack
Assessing whether the operating pressure difference between stack
The heat quantity obtained from the mixer for stack
The power of the operating stacks is adjusted according to temperature needs. If stack
The initiation process for subsequent stacks depends on upstream power supply changes. Opening degree “k” and stack power “p” are used as control variables for optimal initiation sequencing.
To reduce the temperature of stacks at the upper operational limit, the circulation rate is increased to maximize the mixer lye’s cooling effect. If the temperature surpasses this limit, the electrolyzer’s cooling system is activated. Operational adjustments or shutdown of running stacks are executed based on demand. For complete process cessation, relevant valves are closed, and the stacks are disconnected from the mixer. In a hot standby state, the lye circulation rate and valve opening are adjusted to ensure the stack temperature remains within a predetermined range.
Efficiently managing heat consumption and enhancing the startup speed are pivotal for the optimal operation of a multielectrolyzer cluster. However, efforts to minimize both startup energy consumption and the cumulative startup time can present conflicting objectives. To balance these goals, a specific function is defined:
Here,
System constraints include the following:
At the upper temperature limit, control is exerted through heat exchange with the mixer and adjustment of the cooling water flow rate. In this state, k = 1 implies that the divider valve is fully open. If the temperature cannot be reduced below the limit,
These constraints prevent unfeasible solutions, such as sudden power reduction to zero in operational stacks or excessive heat sharing with other stacks, which are not permitted in practical scenarios.
Controlled variables in the simulation include valve openings and stack power:
Optimizing valve openings and power influences temperature variations as follows:
Here,
This subsection examines the operational characteristics of an alkaline electrolyzer under a constant voltage startup mode, as depicted in
The permissible current is restricted by the stack’s temperature. Exceeding this limit can lead to localized overheating and uneven thermal expansion, potentially resulting in thermal stress or damage to critical components. For instance, beyond a temperature threshold of 65°C, the recommended current is capped at 2000 A, which is the rated current of the electrolyzer.
Furthermore, the stack’s resistance shows a strong correlation with temperature. As the stack temperature approaches the optimal level, a decrease in cell voltage is observed, leading to lower energy consumption [
This subsection outlines the setup for a simulation scenario in a hydrogen production facility predominantly powered by PV sources, as illustrated in
The objective of this case study is to effectively start up a multielectrolyzer cluster in coordination with PV power dynamics. To achieve this, the parameter
The criterion for the successful startup of the multielectrolyzer cluster is defined as the point where all stacks reach the optimal operational temperature of 90°C. This scenario aims to demonstrate the effectiveness of the proposed weakly coupled structure and control strategy in maximizing the use of available PV power while ensuring optimal operational conditions for each electrolyzer stack. Details about the parameters used in this simulation case are presented in
Parameter  Unit  Value 

J/K  1.288 

m  1.05  
/  0.8  
J/kg/K  3.1 

kg/s  2.84  
J/K  4.5 

K/W  0.04  
J/K  1.8 

K/W  0.24  
J/kg/K  4.2 

K  293  
K  293  
/  1.2 
The simulation’s initial conditions are set as follows:
All variable temperatures are initially set to
The simulation outcomes, showcasing temperature and power variations, are depicted in
1) Initialization: Both PV power input and stack
2) Early Heating Phase (0–64 min): Stack
3) Connection of Stack
4) Preheating of Stack
5) Activation of Stack
6) Joint Heating Phase (101–129 min): Stacks
7) Connection of Stack
8) Ramping Up Power (129–141 min): Stacks
9) Activation of Stack
10) Maximum Temperature of Stack
11) Power Increase in Stack
12) Optimal Temperature Achievement (168 min): At minute 168, stacks
Each stack is preheated by the lye mixer, and the three electrolyzers begin heating up from their respective preheated temperatures (a = 20°C, b = 33.5°C, c = 41.89°C) to ensure that when the PV power increases rapidly, each electrolyzer can apply sufficient power within the temperature constraints.
For comparative analysis, the simulation was conducted using the topology of independent operation of multiple electrolyzers and the strategy of successive startup for each electrolyzer [
Economic analysis and exergetic efficiency are important system performance indicators [
In terms of economy, transforming a 3stack system into a weakly coupled topology requires adding three divider valves, three confluence valves, one mixer, and several pipelines, with an estimated material purchase and installation cost of $4800. Following installation, the fast startup advantage of the strategy outlined in this paper reduces the time for all three electrolyzers to reach the optimal operating temperature to 168 min, compared with the 199 min for the traditional sequential start–stop method. Within the 199min timeframe after the PV system begins outputting, the effective hydrogen production capacity increases by 86.53 Nm^{3}. This increase results from two factors: efficient power consumption due to effective PV tracking, and earlier closure of venting valves in stacks b and c to achieve effective hydrogen production. At a selling price of $0.3 per Nm^{3}, the cost of the lye mixing system will be recovered after 186 days, with future hydrogen production yielding additional income.
Hydrogen production through multielectrolyzer clusters is central to largescale renewable energy hydrogen production, and the topology design and control of clusters are crucial. However, the traditional standalone operation does not fully utilize waste heat from previously started stacks, and sharing the separator increases challenges in safety and stability control. To address this, the paper undertakes the following initiatives:
1) Innovative Weakly Coupled Topology: A novel weakly coupled multielectrolyzer topology that facilitates heat exchange among stacks via a lye mixer is introduced. The operational temperature of each stack is adjustable through the manipulation of the divider and confluence valves. Notably, this topology enables decoupled operation by closing these valves, thus isolating any malfunctioning stack without disrupting the overall functionality of the cluster.
2) Enhanced Startup Process: The proposed topology is optimized for rapid startup, achieved through strategic preheating and effective heat conservation. This approach significantly reduces total energy consumption and shortens the cumulative startup time for the multielectrolyzer cluster.
3) Optimized Control Strategy and Thermal Model: An advanced control strategy, complemented by a thermal model that incorporates the dynamics of the lye mixer, is developed for the startup process. The validity and effectiveness of this model are demonstrated through simulations based on typical daily PV output power curves.
Despite the contributions of this paper, there are a few areas that need further investigation:
1) Optimization of Objective Function Weighting: Future research could focus on identifying the ideal weighting configuration for the objective function to accommodate diverse upstream power supplies effectively.
2) Adaptation to Diverse Operational Scenarios: Further research could focus on developing optimization strategies tailored to electrolyzer stacks with different manufacturers or operational characteristics, such as varying startup sequences and mixer capacities.
Heat generated by a stack over a unit time interval (
Heat required to heat the electrolyte, W
Heat dissipation from the stack, W
Thermal capacity of the stack, JK^{−1}
Voltage of the stack, V
Current of the stack, A
Number of electrolytic cells, 1
Thermal neutralization voltage, V
Ambient temperature, K
Boltzmann constant, 1.38 × 10^{−23}m^{2} kg s^{−2}K^{−1}
Blackness, 1
Stack diameter, m
Natural convection heat transfer coefficient, 1
Mass flow rate of lye, kgs^{−1}
Specific heat capacity of lye, JK^{−1}kg^{−1}
Temperature of the lye input to the stack, K
Heat brought by the electrolyte flowing through the cooler, W
Heat taken away by the cooler, W
Temperature of the mixed lye, K
Temperature of the separator outlet, K
Temperature of the lye flowing out of the mixer, K
Flow rate of lye into the mixer, kgs^{−1}
Flow rate of lye along the internal circulation, kgs^{−1}
Opening degree of the divider valve, 1
Heat brought in by the lye from each stack entering the mixer, W
Heat dissipated by the mixer, W
Heat taken away by the lye that flows out of the mixer to the stack, W
Thermal resistance of mixer, KW^{−1}
Heat brought in by the lye entering the separator, W
Heat dissipated by the separator, W
Heat carried away by the lye leaving the separator, W
Heat carried away by the gas leaving the separator, W
Heat required to heat up the supplied desalinated water, W
Mass flow rate of H_{2}, kgs^{−1}
Mass flow rate of O_{2}, kgs^{−1}
Temperature of the desalination water, K
Specific heat capacity of the desalination water, JK^{−1}kg^{−1}
Mass flow rate of water, kgs^{−1}
The consumption of water electrolysis with a certain coefficient, 1
Energy consumption when an electrolyzer is independently started, W
Duration when an electrolyzer is independently started, min
Weighting factor of the startup time in the optimization target, 1
Power of stack
Power supply by a photovoltaic (PV) source, W
The first nonzero power point in the PV data, W
The authors acknowledge the reviewers for providing valuable comments and helpful suggestions to improve the manuscript. Special thanks to the manufacturer for providing the experimental electrolyzer, and to the operators of the Narisong hydrogen production test platform for their assistance during experimentation.
This work was supported by the Key Technology Research and Application Demonstration Project for LargeScale MultiScenario Water Electrolysis Hydrogen Production (CTGTC/2023LQ06).
The authors contributed to the paper as follows: study conception and design: M. Chen, H. Xu; data collection: J. Jia, M. Ji, H. Xu, W. Wang, D. Li; analysis and interpretation of results: B. Zhang, L. Han, Z. Yu, J. Jia, M. Chen; drafting of the manuscript: M. Chen, J. Jun, B. Zhang, D. Li, W. Wang, Z. Yu. All authors reviewed the results and approved the final version of the manuscript.
The data supporting the findings of this study are available from the corresponding author upon reasonable request.
Not applicable.
The authors declare that they have no conflicts of interest to report regarding the present study.