Control process of an installation containing several electrolyzers fluidically connected in parallel

Description

1 / 23 PROCESS FOR CONTROLLING AN INSTALLATION CONTAINING SEVERAL ELECTROLYZERS FLUIDICALLY CONNECTED IN PARALLEL

[001] The present invention relates to the field of production of dihydrogen and dioxygen. BACKGROUND OF THE INVENTION

[002] The fight against global warming has forced authorities and industrialists to rethink the energy supply of our societies, with a view to replacing fossil fuels with energies that emit less carbon dioxide, particularly in the transport sector for the motorization of vehicles, but also for the manufacture of fertilizers and steel, not forgetting energy storage and the general decarbonization of industrial processes that use fossil fuels. The need to reduce the production of greenhouse gases and use renewable energy is now well known. Dihydrogen is an alternative to hydrocarbons because it is an easily storable energy vector, unlike electricity, and its oxidation releases very significant energy (285 kJ / mol).

[003] There are several processes for producing dihydrogen. The most advantageous consists of electrolyzing the water molecule because it is a high-yield reaction that does not produce CO2 directly, unlike the widely used processes for reforming methane, coal, and hydrocarbons.

[004] Three main types of electrolyzers for water electrolysis are well known from the prior art: Petition 870250095120, dated 10 / 17 / 2025, page 8 / 53 2 / 23 - Alkaline electrolyzers (AWEs), which are characterized by the use of a liquid electrolyte that allows the transfer of hydroxyl ions (OH-) from the cathode to the anode, - high-temperature electrolyzers, whose electrolyte is a ceramic; and - membrane electrolyzers (PEMs), whose electrolyte is a proton-conducting ion-exchange membrane.

[005] In all three cases, the system must be fed with very pure water (in the case of alkaline electrolyzers, by feeding an electrolytic solution of sodium hydroxide (NaOH) or potassium hydroxide (KOH)). For the sake of brevity, the remainder of the description will refer to an alkaline electrolyzer, but it is understood that the present invention also applies to a membrane electrolyzer (for example, a proton exchange membrane).

[006] According to the well-known process of the prior art, an electrolytic solution (commonly designated by the English term lye) is brought to a set of electrolytic cells (known as an electrolyzer stack) through a specific inlet. The electrolytic solution passes through the electrolyzer stack. Water is decomposed into gaseous hydrogen molecules H2 at the cathode and oxygen O2 at the anode. Generally, a diaphragm separates the anode from the cathode so that, under normal conditions, hydrogen and oxygen do not mix. The installation comprises an outlet for dihydrogen and the electrolyte circulating on the cathode side (catholyte) and an outlet for dioxygen and the electrolyte circulating Petition 870250095120, dated 10 / 17 / 2025, page 9 / 53 3 / 23 on the anode (anolyte) side. In other words, there are two distinct streams, so there is a gas-liquid separator dedicated to separating dihydrogen from the catholyte, and a gas-liquid separator for separating dioxygen from the anolyte. The liquid outputs from the two gas-liquid separators are then mixed before feeding back into the electrolyzer stack. In both streams, at the outlet of the electrolyzer stack, the liquid phase (lye) is loaded with gas bubbles. At the outlet of the gas-liquid separator, only a few gas bubbles remain in the lye evacuated through the lower orifice of the gas-liquid separator dedicated to the liquid phase, while most of the gas phase is extracted from the gas-liquid separator through the upper orifice of the gas-liquid separator. For various reasons, it is important to separate the gas from the lye. Firstly, the more the gas is separated from the electrolyte, the greater the gas production, which contributes to the good electrochemical efficiency of the process.Next, the H2 / O2 mixture is highly explosive. If the separation is not performed correctly, a significant amount of gas, commonly referred to as "residual gas," is carried over into the liquid outlet of the gas-liquid separator. During subsequent circulation in the electrolyzer stack (the electrolyte rotates in a closed circuit), some of this gas passes into the other compartment and therefore to the wrong side.

[007] It is understood that the interest, both economic and ecological, of using hydrogen in an energy process depends largely on the performance of the hydrogen production device.

[008] Ideally, it is necessary that: Petition 870250095120, dated 10 / 17 / 2025, page 10 / 53 4 / 23 - the manufacturing and operating costs of the production device should be as low as possible; - the impact on natural resources for the manufacture and operation of the production device is limited; - the device produces few or no polluting emissions during its operation; - The device should be simple, efficient, reliable, and relatively compact...

[009] Furthermore, the energy efficiency of the installation is not the only issue when considering operating costs. Indeed, objectives - Flexibility / reactivity to adapt to rapid changes in hydrogen production demand; Maintainability and efficiency are also crucial for these installations, which often comprise several electrolyzers operating at the same production site and / or centrally controlled. Finally, the ability to maximize electrical energy consumption for a given dihydrogen production can, paradoxically, prove advantageous in certain circumstances, particularly in terms of regulating the electricity grid (local or otherwise). OBJECT OF THE INVENTION

[010] The invention aims, in particular, to improve all or part of the energy aspects of an installation containing several electrolyzers. SUMMARY OF THE INVENTION

[011] For this purpose, a process for optimizing the control is provided for, in accordance with the invention. Petition 870250095120, dated 10 / 17 / 2025, page 11 / 53 5 / 23 an installation containing several electrolyzers in parallel (referred to as MIEL-S hereafter).

[012] The essential feature of the invention process is that control is performed to homogenize the individual loads of the electrolyzers in operation, while aiming to minimize the specific consumption of the installation for a predetermined total production load of dihydrogen or for the availability of electrical energy of the installation.

[013] It is under these conditions that the installation can achieve optimal performance, simultaneously promoting overall responsiveness and flexibility, and also taking into account the availability constraints of the electrolyzers (referred to as ELY in the following document) with regard to, for example, the installation's maintenance schedule. The invention also allows the exploration of a strategy to maximize the electrical power consumed by MIEL-S for a given dihydrogen production, in order to maximize the regulation range of the absorbed electrical power within the framework of the electricity grid regulation service (local or otherwise).

[014] Other features and advantages of the invention will become apparent from reading the following description of particular, non-limiting embodiments of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[015] Let MIEL-S be a set of N electrolyzers that are a priori identical (Multi Identical Electrolyzer System) and that can operate a number n of electrolyzers, where n ranges from 0 to N. Petition 870250095120, dated 10 / 17 / 2025, p. 12 / 53 6 / 23

[016] Consider a HOLOS(n) strategy, which consists of distributing the load homogeneously within the n electrolyzers in operation (Homogenous Load Operation Strategy).

[017] Reference will be made to the attached drawings, in which: [Fig. 1] Figure 1 is a characteristic curve of the individual specific consumption of one of the multiple identical electrolytes of MIEL-S; [Fig. 2] Figure 2 shows an N HOLOS curve network. Each curve in the graph is associated with a number n, where n can be equal to a value between 1 and N electrolyzers operating within the MIEL-S system, which includes N electrolyzers in total. The n electrolyzers in operation follow a HOLOS strategy, and the other Nn electrolyzers in the MIEL-S system are stopped (or on standby). Therefore, the curve network shows the specific consumption (kWh / Nm3) of the MIEL-S as a function of the system load and the number n of electrolyzers in operation, considering that the operating strategy is always of the HOLOS type regardless of the number n. A HOLOS type strategy will always be based on this curve network. An operating strategy that does not follow this curve network is not of the HOLOS type; [Fig. 3] Figure 3 is analogous to Figure 2 and shows the intersections of the curves with each other. These intersection points are called HOLOS points. There are 2*N intersection points. These HOLOS points are numbered from 1 to 2*N in order of occurrence during an increase in the load of the MIEL-S system; Petition 870250095120, dated 10 / 17 / 2025, page 13 / 53 7 / 23 [Fig. 4] Figure 4 is analogous to Figure 2 and shows the best efficiency curve (BEL = Best Efficiency Line) of a MIEL-S; [Fig. 5] Figure 5 is a graph showing the position of the 2*N “HOLOS points” on the individual specific consumption characteristic of multiple identical MIEL-S electrolyzers; [Fig. 6] Figure 6 is a graph showing the evolution of the charge of the electrolyzers within MIEL-S during the occurrence of 2*N HOLOS points; [Fig. 7] Figure 7 is a graph showing the relative load relationship at consecutive HOLOS points during a monotonic ramp of the MIEL-S; [Fig. 8] Figure 8 is analogous to Figure 2 and shows the best efficiency curve (BEL) for electrolyzers considered “invisible” in groups of four in terms of load; [Fig. 9] Figure 9 is analogous to Figure 4 and shows the worst efficiency line (WEL) curve, which does not follow a HOLOS-type strategy. Its interest, in the context of a grid regulation service (local or otherwise), is to identify the highest electrical energy threshold that MIEL-S can consume for a given production of dihydrogen; [Fig. 10] Figure 10 is analogous to Figure 9 and shows the worst-case efficiency curve (HOLOS type WEL) within the HOLOS strategy framework; [Fig. 11] Figure 11 is analogous to Figure 9 and shows the curve that minimizes the number of electrolyzers in operation (T LEO L = The Less Electrolyser in Operation Line) for a given MIEL-S load and aiming to maintain a HOLOS type strategy; Petition 870250095120, dated 10 / 17 / 2025, page 14 / 53 8 / 23 [Fig. 12] Figure 12 is analogous to Figure 9 and shows the curve that maximizes the number of electrolyzers in operation (T MEO L) for a given MIEL-S load and aims to maintain a HOLOS type strategy; [Fig. 13] Figure 13 shows the number of electrolyzers in operation as a function of the MIEL-S load, and this is for the different HOLOS strategies mentioned above; [Fig. 14] Figure 14 shows an example of the intended hydrogen production curve of the plant as a function of time and plant load; [Fig. 15] Figure 15 is analogous to Figure 14 and also shows the response curve in terms of the number of electrolyzers in operation when the strategy followed is of the “BEL” type; [Fig. 16] Figure 16 is analogous to Figure 14 and also shows the response curve in terms of specific consumption when the strategy followed is of the “BEL” type; [Fig. 17] Figure 17 is similar to Figure 14 and also shows the response curve in terms of power consumed by the installation when the strategy followed is of the “BEL” type.

[018] It should be noted that the values ​​are provided for informational purposes only, to give orders of magnitude and also depend on the configurations of the equipment that make up the electrolyzers.

[019] This also applies to time scales, which are also mentioned for illustrative purposes. DETAILED DESCRIPTION OF THE INVENTION

[020] By way of example, the multi-electrolyzer installation or system (MIEL-S or “Multi Identical Petition 870250095120, dated 10 / 17 / 2025, page 15 / 53 9 / 23 The Electrolyzer System consists of N=24 identical electrolyzers (ELY) of 1000 Nm3 / h fluidically connected in parallel, each being considered totally independent in terms of production load. The invention is evidently applicable to any other nominal capacity of the electrolyzers (normally from 1 Nm3 / ha to 4000 Nm3 / h, in particular between 100 Nm3 / h and 2000 Nm3 / h, in particular from 500 Nm3 / h to 1500 Nm3 / h).

[021] Optimizing the control of a multi-electrolyzer installation is not trivial and needs to be conceptually built step by step.

[022] The first part of this description sets out a methodology for establishing different operational strategies for a MIEL-S, answering, for example, the following questions: • How to distribute the production load inside a MIEL-S? • At what production load should an ELY be stopped or started inside a MIEL-S? • The answer to these first two questions changes depending on the objective being pursued, such as: to minimize the energy consumption required for the production of dihydrogen; to maximize reactivity in terms of changes in production load; To increase the maintainability of the system; or to minimize the degradation of the system's energy performance?

[023] This methodology also allows the operator of a MIEL-S (or its designer) to know the Petition 870250095120, dated 10 / 17 / 2025, page 16 / 53 10 / 23 Installation properties (specific consumption as a function of load, start-up / stop time, degradation rate as a function of operating demands, etc.) to arbitrate between different operational strategies and identify the one that best suits your operational context.

[024] This first part is based on an idealized approach to reality, taking into account only the theoretical behavior of the installation at the beginning of its useful life (BOL) to define a strategy for optimizing its control.

[025] The second part of this description aims to complete this approach with “non-ideal” phenomena such as, for example, the impact of performance degradation over time on control optimization.

[026] The initial hypothesis is that the demand in terms of hydrogen production load is perfectly known in a time horizon T and that the problem to be solved consists of minimizing the specific electrical consumption of the installation (specific consumption Cs [kwh / Nm3]) taking into account this production load in that time horizon T. 1. Idealized approach a. Critical points of an electrolyzer efficiency curve (see figure 1)

[027] Three characteristic points should be observed on the specific consumption curve, Cs, of the ELY: - MCR or maximum continuous rate (maximum continuous flow). In the example in Figure 1, consider 100% load (equivalent to a nominal production of, for example, 1000 Nm3 / h) for a Cs of 4.66 kWh / Nm3; Petition 870250095120, dated 10 / 17 / 2025, page 17 / 53 11 / 23 - BEP or best efficiency point. Let's say 52% load for a Cs of 4.56 kWh / Nm3 in the example of Figure 1. The specific consumption curve effectively passes through a minimum (BEP) that results from the antagonistic internal phenomena in the electrolyzer cell when its current density varies; - MSOL or "minimum stable operating point". In Figure 1, the minimum stable operating point is at 27% hydrogen production load. b. Best Efficiency Line (BEL) - Optimization of the operation of a MIEL-S installation

[028] Given the concavity of the curvature (positive second derivative) of the Cs curve in Figure 1 above, a system operating with several ELYs and having to meet a given load will always minimize its specific consumption, homogenizing as much as possible the individual production load of each ELY in the system. In the installation example described here, it will be considered that the twenty-four ELYs each operate at 70% of their load; the installation will then be at 70% of its total capacity (or load). The specific consumption of the installation would then be (see Figure 2, scale on the left) approximately 4.59 kWh / Nm3. One could imagine varying the individual load of one of the twenty-four ELYs downwards and keeping the total load of the installation unchanged, varying another of the twenty-four ELYs in the same proportions. The result would be worse in terms of the specific consumption of the installation.

[029] This defines a first optimization strategy by homogenizing the operational load, called Petition 870250095120, dated 10 / 17 / 2025, page 18 / 53 12 / 23 HOLOS stands for Homogeneous Load Operation Strategy, which simplifies the concept by assuming that all electrolyzers always have the same specific consumption curve.

[030] This strategy is based on the principle that, for a given number of operating ELYs, out of the twenty-four in the installation (MIEL-S), these will minimize the installation's electrical consumption for a given total production load, all being at the same individual production load. Furthermore, in Figure 2, twenty-four specific consumption curves are considered. They are all of the HOLOS type. The first one mentioned above relates to the case where all twenty-four ELYs are in operation. The other twenty-three curves correspond to 23, 22, 21, ..., 3, 2, 1 operating ELYs, thus following a HOLOS strategy (i.e., the individual loads of the operating electrolyzers are identical to each other).

[031] The next step is to define when to stop or start an ELY when the load of the MIEL-S plant varies. This is the object of Figure 3 which shows the intersections of the HOLOS (24) with HOLOS (23) curves; HOLOS (23) with HOLOS (22); ...; HOLOS (2) with HOLOS (1). The production loads at the aforementioned intersections (HOLOS points) are actually percentages of the plant's (MIEL-S) production load to which it is theoretically appropriate to vary the number of ELYs in operation to minimize the specific consumption of the system.

[032] This optimization does not take into account, at this stage, the penalty that the MIEL-S installation suffers in terms of future performance degradation, or performance variations within the same batch due to tolerances of Petition 870250095120, dated 10 / 17 / 2025, p. 19 / 53 13 / 23 manufacturing. This "less than ideal" aspect will be addressed in the dedicated section of this description.

[033] From there, it is possible to trace the BEL envelope curve of the MIEL-S. This is shown in Figure 4 and constitutes a lower limit of the specific consumption of the installation (MIEL-S). In other words, the BEL envelope curve defines the lower limit of the set of specific consumption curves. Although this line appears to describe a plateau (particularly between 25% and 50% load in the example considered here), it actually passes through twenty-four minima, all equal and corresponding to the BEP defined above.

[034] The individual loads of the ELYs in the installation when the number of ELYs in operation changes at the installation level are represented in Figure 5. The chronology of the numbers shown in the graph corresponds to an acceleration of the MIEL-S installation starting with one ELY operating at MSOL (point 1). This ELY increases in load, following the Cs curve until point 2, when a second ELY must be started to minimize the specific consumption Cs, continuing to increase the installation load (MIEL-S). By doing this, the individual load of the two ELYs is positioned “instantaneously” at point 3. The individual loads thus continue to evolve towards points 4, 5, 6, 7, ..., 46, 47. These points are referred to as “HOLOS points” hereafter. For simplicity, the start-up / load increase times of each additional electrolyzer are neglected here: this is a static optimization in steady state.

[035] The graph in figure 6 shows the evolution of the “HOLOS points” located “to the left and right” of the BEP. Petition 870250095120, dated 10 / 17 / 2025, page 20 / 53 14 / 23

[036] The individual load relationship of consecutive HOLOS points (immediately before and immediately after the change in the number of operational ELYs) follows a logical sequence illustrated in the graph in figure 7.

[037] If the MIEL-S installation were to comprise ELYs protected four by four in terms of load (for example considering six ELY4000 instead of twenty-four ELY1000), the BEL would evolve as shown in the graph in figure 8 (consider the red line and not the black line). c. Worst Efficiency Line (WEL) of a MIEL-S installation

[038] In Figure 9, the worst efficiency line WEL (in red) was plotted along with the best efficiency line BEL (in black) to show the maximum potential for improvement allowed when trying to approach the best efficiency line BEL. Also, the difference between these two curves, for a given dihydrogen production, is the image of the lever for varying electrical power consumed by MIEL-S. This concept is useful in the context of a power grid regulation service, whether local or not.

[039] WEL corresponds to a two-phase implementation.

[040] Considering a monotonic load reduction and starting from an initial situation where MIEL_S is at full load (all ELYs at 100% load), during phase 1, the load reduction is organized as follows: a number i (i ranging from N-1 to 0) of ELYs at 100%, a number equal to Ni-1 of ELYs at MSOL, and one ELY adjusting the installation load with its own load. When this ELY reaches MSOL, i becomes i-1 and one of the ELYs operating at 100% of the Petition 870250095120, dated 10 / 17 / 2025, page 21 / 53 15 / 23 charge begins to regulate, on its own, the charge of MIEL-S. Phase 1 leads to a final situation where all ELYs N are in MSOL.

[041] Phase 2 therefore begins with an initial situation that is the final situation of phase 1, i.e., all N ELYs in MSOL. During phase 2, the continuation of load reduction is managed as follows: a number j (j ranging from N-2 to 0) of ELYs in MSOL, a number equal to Nj-1 of ELYs are stopped (or put on standby) and one ELY adjusts the installation load with its own load. When this ELY reaches MSOL, it is switched off (or put on standby) and j becomes j-1. Phase 2 results in a final situation where all N ELYs are stopped (or put on standby).

[042] The graph in Figure 10 shows a purposefully degraded alternative to the Best Efficiency Line: the Worst Efficiency Line - HOLOS. Although this alternative still considers a HOLOS approach, it ignores the HOLOS points defined above. In effect, considering, for example, the case of a monotonic load decrease and starting from the case of twenty-four ELYs at 100% load, the strategy followed consists of stopping the ELYs as quickly as possible, one by one, thus leaving the other ELYs operating with a relatively high individual load and this permanently.After stopping a certain number K of ELYs (K equal to fifteen in the defined numerical example) and decreasing the load of the remaining NK (NK equal to nine in the described numerical example) ELYs in operation, a second phase of this strategy consists, when the installation load is low enough to cross the HOLOS curve (N), in restarting the set of N (N equal to twenty-four) ELYs to make them all operate with a very low individual load, where the specific consumptions. Petition 870250095120, dated 10 / 17 / 2025, p. 22 / 53 Individual 16 / 23 units are less good. Following the illustration, a one-by-one shutdown of the ELYs is also considered until only one ELY is operational and descends to its MSOL.

[043] The use of WEL and WEL HOLOS strategies results in a growing relative lack of optimization as the installation load decreases. d. HOLOS line of the minimum number of electrolyzers in operation (T LEO L = The Less Electrolyzer in Operation Line) of a MIEL-S installation

[044] This strategy, illustrated in Figure 11, consists of stopping an ELY as quickly as possible, once the total production load of the MIEL-S plant allows it, while also being part of a HOLOS strategy. Therefore, the number of electrolyzers simultaneously operating is reduced to the minimum number necessary to supply the total dihydrogen production load: the result is that the individual load of the operating electrolyzers is relatively high, even when the total production load is relatively low. The main optimization objective concerns the number of electrolyzers, and specific consumption is a secondary objective.

[045] This strategy is preferably used during facility maintenance operations that require stopping as many ELYs as possible or in cases where a rapid reduction in the total production load is necessary. e. HOLOS line of the maximum number of electrolyzers in operation (T MEO L = The More Electrolyzer in Operation Line) of a MIEL-S installation

[046] This strategy, illustrated in figure 12, consists of keeping as many ELYs as possible in operation. Petition 870250095120, dated 10 / 17 / 2025, page 23 / 53 17 / 23 during a monotonic reduction in the MIEL-S load, simultaneously following a HOLOS strategy. When the load falls below the load limit corresponding to N ELY in MSOL, an ELY is stopped (or placed on hold) and the MIEL-S then evolves to HOLOS (N-1). The continuation of the load reduction is managed identically, successively transitioning from HOLOS (N-1) to HOLOS (N- (1+1)) (1 going from 0 to N-2). When the charge leads to In the MSOL system at HOLOS-1, the last ELY is, in turn, shut down (or put on standby). As a result of this strategy, the individual load on the operating electrolyzers is relatively low when the total production load is low.

[047] This strategy can be used throughout the installation's lifespan, for example to mitigate performance degradation associated with the individual load rate of the electrolyzers. Furthermore, this strategy offers the widest operating range without starts and stops (between 27% and 100%). This start / stop range can be called "SSL (Start & Stop Less)". This SSL range offers, on the one hand, the minimization of performance degradation due to start / stop occurrences and, on the other hand, the maximization of system reactivity (thanks to the absence of a time-consuming ELY start / stop procedure in the MIEL-S) in the face of load variations that would have led, through other operating strategies, to ELY starts / stops, increasing performance degradation and decreasing the reactivity of the entire MIEL-S. In addition, this strategy simplifies the control of the installation and its operation.It should be noted that it is possible to generalize the notion of T MEO L as indicated below. The case that has just been described considers the initial situation “HOLOS (N) (here “HOLOS. Petition 870250095120, dated 10 / 17 / 2025, p. 24 / 53 18 / 23 (24)). The resulting SSL range can be precisely called SSL (N) (here SSL (24)). And then we emphasize that this strategy can be specifically denoted by T_MEO_L(N). For a given MIEL-S, it is quite conceivable that during a given period the load will vary many times, mainly in a range P not included in the SSL(N) range. It will then be entirely possible to optimize the choice of another T MEO L (m) (with m < N) which would lead to better inscribing the range P in the SSL (m) thus generated by this choice of operational strategy. In other words, the electrolyzers are controlled to make them operate at their optimum operating point a maximum number m of electrolyzers less than the total number N of electrolyzers, the maximum number m corresponding to an installation load that varies in an operating range below the maximum operating range of the installation involving the total number of electrolyzers. f. Load Management of a MIEL-S Plant

[048] The plant load can be managed according to a first approach based on dihydrogen production or a second approach based on the availability of electrical energy to power the plant. The description in this document assumes the first approach. Indeed, from the graph in Figure 1, the ordinate is expressed as a function of the hydrogen production load. The set of graphs and operational strategies described below are therefore expressed as a function of the hydrogen production load. To consider the second approach, simply reproduce the same reasoning used for the first approach, starting from an alternative to the graph in Figure 1, expressing the evolution of specific consumption as a function of the Petition 870250095120, dated 10 / 17 / 2025, page 25 / 53 19 / 23 electrical load” instead of “the evolution of specific consumption as a function of hydrogen production load”. The graphs look substantially similar and lead to the same types of reasoning and analysis.

[049] For this reason, figure 13 illustrates the first approach and represents the production of dihydrogen by the MIEL-S plant as a function of the number of ELYs in operation and according to different usable strategies, namely, BEL, T LEO L, TMEO L(N), HOLOS WEL.

[050] Figure 14 shows the required production load of dihydrogen (thin solid line) according to the first approach (different profile shapes were chosen to illustrate the adaptation of the plant's production to these different profiles).

[051] Figure 15 shows this same required production load of dihydrogen (thin dashed line) superimposed on the load produced in response to the implementation of the BEL strategy (thick red solid line) and the number of ELYs in operation (thick blue dashed line).

[052] Figure 16 shows this same required production load of dihydrogen (thin dashed line) superimposed on the load produced in response to the implementation of the BEL strategy (thick solid line) and the specific consumption Cs of this strategy (thin blue solid line).

[053] Figure 17 shows the required production charge of dihydrogen (thin dashed line) over which the charge produced in response to was superimposed. Petition 870250095120, dated 10 / 17 / 2025, page 26 / 53 20 / 23 implementation of the BEL strategy (thick solid line) and the power of the MIEL-S installation (thin blue solid line). 2. Consideration of non-ideal phenomena a. List of “non-ideal” phenomena that influence the optimization of load management in the MIEL-S installation: - Limitation of the operating range (% of dihydrogen production load) associated with certain plant components (including those internal to the ELYs in the case of grouping several ELYs); - Performance degradation associated with ELY stops and starts; - Performance degradation associated with the operating hours of the ELYs; - Start-up and shutdown times of ELYs, as well as energy losses during transients; - Performance imbalance between the ELYs in the electrolyzer stack (the electrolyzers are theoretically all identical, but in reality their performances are different); - Dynamic need to absorb or reduce the load on the MIEL-S installation based on the forecast of production demand; - The total or partial unavailability of one or more ELYs; - Emergency stop occurrence. b. Consideration of “non-ideal” phenomena that influence the optimization of load management in the MIEL-S installation.

[054] In general, the answer to the question “at what installation load should ELYs be stopped or started within a multi-ELY system (MIEL-S)?”, having in Petition 870250095120, dated 10 / 17 / 2025, page 27 / 53 Accounting for these "non-ideal" phenomena, the 21 / 23 factor will be a function of the forecast of future load and the remaining lifespan of the installation...

[055] Indeed, bearing in mind, for example, that each shutdown of an ELY implies a deterioration in its future performance, it is necessary, during operation and in the face of an increase in the demand for hydrogen production: - Should we start an additional ELY to reduce the average load on the ELYs and thus decrease the installation's energy consumption in the short term, while simultaneously being informed that this load, which has been increasing until now, is about to decrease and that the ELY that has just been started will have to be stopped soon? - On the contrary, should we tolerate a lack of optimization in the short term in order to preserve the installation's performance in the long term?

[056] The answer to this question will not necessarily be the same depending on when we ask it during the life of the installation. This brings into play the concept of LCOH and its continuous recalculation based on short- and long-term load forecasts.

[057] Thus, the optimization of an installation's performance can be achieved for, separately, or wholly or in part in combination: - the individual homogenization of ELYs; - the operation of a maximum ELY at its optimal operating point; - the dynamic determination of the load of each individual ELY according to its particular specific efficiency (linked to its own aging, its own capacity, etc.); Petition 870250095120, dated 10 / 17 / 2025, page 28 / 53 22 / 23 - adapting the individual capabilities of the installation to fine-tune the load or production capacity and optimize the efficiency of the installation.

[058] It is clear that the invention is not limited to the embodiment described, but encompasses any variant that falls within the scope of the invention, as defined by the claims.

[059] In particular, numerical values ​​are given for example purposes only.

[060] The specific consumption Cs and dihydrogen production values ​​are valid for electrolyzers at the maximum limit of their operation, taking into account energy losses (electrolyzer cell, rectifier, transformer, gas-liquid separator, purifier, pump, etc.) and dihydrogen losses in gas-liquid separation and purification units.

[061] The standard cubic meter is considered to be at 0 °C and 1 atm.

[062] The number of electrolyzers may be less than or greater than twenty-four.

[063] The actual operating parameters are determined by monitoring the installation or by a statistical analysis of the existing installations.

[064] Each electrolyzer can be fitted with its own energy conversion device (rectifier). Alternatively, at least two batteries can be connected electrically in series (voltage addition).

[065] The process may include the step of defining a worst-efficiency mode in which the installation load is adjusted by adjusting the load of only one of the electrolyzers, Petition 870250095120, dated 10 / 17 / 2025, page 29 / 53 23 / 23 with the other electrolyzers at their maximum load, or at their minimum operating point, or stopped; and the step of controlling the installation between an optimal mode corresponding to the homogenization of the individual loads of the electrolyzers in operation and the worst efficiency mode.

[066] It should be noted that the Cs versus load curve is preferably representative of the entire installation (including, therefore, the Balance Of Plant”) rather than representative only of the electrolyzer stack (and some of its loss mechanisms), since the end user is interested in the overall consumption (i.e., including the BOP). Petition 870250095120, dated 10 / 17 / 2025, page 30 / 53

Claims

1 / 3 CLAIMS 1. A process for controlling an installation containing several electrolyzers fluidically connected in parallel, characterized by comprising the step of controlling the electrolyzers in order to homogenize the individual loads of the electrolyzers in operation, so as to minimize a specific electrical consumption of the installation for a predetermined total production load or for an electrical power available to supply the installation.

2. Process, according to claim 1, characterized by comprising the step of determining specific consumption curves, each corresponding to a number n of electrolyzers in operation, the number n varying from 1 to N, which is the total number of electrolyzers in the installation.

3. Process, according to claim 2, characterized by comprising the step of determining an envelope curve representing a lower limit of the set of specific consumption curves.

4. Process, according to claim 3, characterized in that the number of electrolyzers to be kept simultaneously in operation is determined as a function of the envelope curve.

5. Process, according to claim 4, characterized by comprising the step of reducing the number of electrolyzers simultaneously in operation to a minimum number necessary to supply the total production load.

6. Process, according to claim 4, characterized by comprising the step of increasing the number of electrolyzers simultaneously in operation to a maximum number necessary to supply the total production load. Petition 870250095120, dated 10 / 17 / 2025, p. 31 / 53 2 / 3 7. A process, according to claim 1, characterized by first determining a theoretical mode of control of the installation based on the state of the installation at the beginning of its useful life, and then modifying the theoretical mode of control according to at least one real operating parameter of the installation.

8. Process, according to claim 7, characterized in that the actual operating parameter belongs to the following group of parameters: - limitation of an operating range of at least one component of the installation; - performance degradation associated with the start-up and stop cycle of the electrolyzers; - performance degradation associated with the operating time of the electrolyzers; - start-up and stop time of the electrolyzers and energy losses during the transient phase between the start-up and stop of the electrolyzers; - performance imbalance between electrolyzers; - dynamic need for absorption or reduction of the installation load based on the forecast of production demand; - total or partial unavailability of at least one electrolyzer; - occurrence of an emergency shutdown of all or part of the installation.

9. Process, according to any one of claims 7 or 8, characterized in that the actual operating parameter is determined by monitoring the installation or by statistical analysis of existing installations. Petition 870250095120, dated 10 / 17 / 2025, p. 32 / 53 3 / 3 10. Process, according to claim 1, characterized by comprising the step of controlling the electrolyzers to operate a maximum of the electrolyzers at their ideal operating point.

11. Process, according to claim 1, characterized by comprising the step of controlling the electrolyzers to operate at their ideal operating point a maximum number of electrolyzers less than the total number of electrolyzers, wherein the maximum number corresponds to an installation load varying in an operating range less than a maximum operating range of the installation involving the total number of electrolyzers.

12. Process, according to claim 1, characterized by comprising the step of dynamically determining the individual load of each electrolyzer as a function of its particular specific efficiency.

13. A process, according to any of the preceding claims, characterized by comprising the step of defining a worst-efficiency mode in which the installation load is adjusted by adjusting the load of only one of the electrolyzers, with the other electrolyzers at their maximum load, at their minimum operating point, or stopped.

14. Process, according to claim 13, characterized by comprising the step of controlling the installation between an optimal mode corresponding to the homogenization of the individual loads of the electrolyzers in operation and the worst efficiency mode. Petition 870250095120, dated 10 / 17 / 2025, pp. 33 / 53

Images