Electrolysis temperature control

Abstract

A method of controlling the temperature of an inlet fluid of at least one electrolyser cell stack. The at least one electrolyser cell stack comprising at least one fluid inlet for the inlet fluid and at least one fluid outlet for an outlet fluid, the method comprising: predicting, based on a model, an outlet fluid temperature a desired operating condition; and setting an inlet fluid temperature setpoint based on the predicted outlet fluid temperature; the method further comprising: receiving a measurement of the temperature of the outlet fluid; determining an delta between the predicted outlet fluid temperature and the measurement of the outlet fluid temperature; and inputting a correcting parameter into the model based on said determined delta.

Description

[0001] Electrolysis temperature control

[0002] This invention relates to the field of electrolysis, specifically to methods for controlling the temperature of an inlet fluid in electrolyser cell stacks.

[0003] An electrolyser may comprise one or more stacks of electrolyser cells - commonly known as electrolyser cells or regenerative fuel cells. The electrolyser is used to split a source fluid into its constituent parts and to output those constituent parts in the off gases from the electrolyser. For that purpose, it requires a source of electricity for supplying an electric current and voltage through / across the at least one stack of electrolyser cells. As an example, electrolysers can be used to produce hydrogen and oxygen from water, or carbon monoxide and oxygen from carbon dioxide, each by way of electrolysis.

[0004] The collection, storage or distribution of the produced oxygen is important as it can be utilised in industry and medical applications, amongst many other uses. Carbon monoxide is also useful for numerous chemical processes, and thus too will be collected, stored or distributed for downstream or later use. The collection, storage or distribution of the produced hydrogen is perhaps most important as it is a fuel that can help in the race for decarbonisation and for helping to achieve net zero targets. For example, the produced hydrogen can be utilised as a fuel for combustion, or as a fuel for a use in a fuel cell system for achieving an electrolytic reaction in the fuel cell to recombine the hydrogen with oxygen, with resultant electrical and heat outputs. The hydrogen can also have many other industrial and chemical uses as a reducing agent.

[0005] Given the various uses of the off-gases, the off-gas demands from an operator of the electrolyser can change. For example, lower rates of production of the oxygen, hydrogen or carbon monoxide might be required by a downstream service, or increases may be required at peak times. Similarly, if the electrolyser is being powered by an intermittent fuel supply, such as solar, wind or hydroelectric power, it is beneficial to change the output of the electrolyser to match the input, this is termed Toad following’.

[0006] The operating temperature of an electrolyser affects the efficiency and rate of production of off gasses. This is particularly the case for intermediate or high temperature solid oxide electrolysis cells (SOEC) which operate at temperatures in excess of 400°C. As such, a change in desired production rate often necessitates a change in operating temperature.

[0007] The present invention relates to how to quickly and efficiently change operating temperature to enable efficient switching between operating conditions.

[0008] Statements of invention

[0009] According to a first aspect of the invention there is provided a method of controlling the temperature of an inlet fluid of at least one electrolyser cell stack. The electrolyser cell stack comprises at least one fluid inlet for the inlet fluid and at least one fluid outlet for an outlet fluid. The fluid outlet being in fluidic communication with the fluid inlet via a chemically active region in which a reduction / oxidation reaction occurs. The general structure of an electrolysis stack is shown in Figure 1 and described below.

[0010] The control method comprises a prediction where an outlet fluid temperature for a desired operatingcondition is predicted based on a model; and settingan inletfluid temperature setpoint based on the predicted outlet fluid temperature. The inlet fluid temperature set point may apply to the oxidant (air) inlet, fuel (steam, CO2) inlet, or both oxidant and fuel inlets.

[0011] The operating condition may correspond to a desired production rate, which is in turn defined by a certain current level. The operating condition may be defined by the current available, for example, from an associated renewable energy source. The temperature associated with the operating condition may be defined by a ‘thermoneutral’ point where the temperature of the outlet is substantially the same as that of the inlet. However, an offset (rendering the electrolyser endo- or exothermic) may be desirable in certain circumstances.

[0012] The method further comprises an extended state observer used to improve the model. This component comprises receiving a measurement of the temperature of the outlet fluid; determining a delta (also called an error) between the temperature of the outlet fluid predicted by the model and the measurement of the outlet fluid temperature. The method comprises inputting a correcting parameter, optionally in the form of a power balance correction, based on the determined delta into the model as a model parameter. The measured outlet fluid temperature may be the oxidant (air) outlet, product (hydrogen, CO) mixed with unutilised fuel outlet, or both oxidant and product outlets.

[0013] In summary, the method consists of a model-based prediction, where the model is updated by ‘extended state observer’ which monitors the temperature of the fluid outlet in response to the change in inlet temperature set point. This enables continual improvement to the model during transient conditions.. It should be appreciated that the desired steady state may be thermoneutral, or the desired steady state may be at a given temperature offset.

[0014] Such a method enables faster switching between operating conditions compared to simply setting the inlet to match a measured outlet. This is because the method proposed uses the model to predict where the desired steady state operating condition (e.g. thermoneutral) will be, speeding up significantly the transition between operating points.

[0015] As a model alone would lead to steady state error, an integral contribution is needed. The extended state observer provides this integral effect via the power balance correction. This method is more efficient than a proportional, integral, derivative (PID) control which can over or undershoot. The extended state observer will only integrate the error between the measured behaviour and the expected behaviour, reducing significantly the risk of over and undershooting due to the integral component. Configuring a PID controller in an optimum way is challenging for electrolyser stacks / arrays as each stack / array differs due to manufacturing tolerances, environmental factors, and as they age.

[0016] This method may run continuously, and can act to maintain steady state (or pseudo steady state) conditions, for example as the electrolyser stack ages.

[0017] Optionally, the method further comprises monitoring the correcting parameter and determining a stack condition based on the value of the correcting parameter. In such a way, by monitoring the accuracy of the model, it is possible to determine whether the stack is performing as predicted. Performance deviating from that predicted by the model may indicate damage, degradation or another problem which may necessitate a change to operating conditions (or preemptive maintenance).

[0018] According to another aspect, there is provided a method of determining a stack condition of at least one electrochemical cell stack (which may be a fuel cell or electrolyser), the electrochemical cell stack comprising at least one fluid inlet for the inlet fluid and at least one fluid outlet for an outlet fluid, the method comprising: predicting, based on a model, an outlet fluid temperature for a desired operating condition; and setting an inlet fluid temperature setpoint based on the predicted outlet fluid temperature.

[0019] The method further comprises: receiving a measurement of the temperature of the outlet fluid; determining a delta between the predicted outlet fluid temperature and the measurement of the outlet fluid temperature; determining a correcting parameter for the model based on said determined delta; monitoringthe correcting parameter, and determininga stack condition based on the value of the correcting parameter.

[0020] Optionally, the stack condition depends on the rate of change of the correcting parameter. For example, if the correcting parameter increases over time, this may be indicative of degradation rather than model inaccuracy.

[0021] Optionally, the stack condition is at least one of, a degradation status, a damage status, a predicted lifetime, thermal management status, leak status, and a future failure state. An operator can be alerted to a determined condition and modify the operation of the stack accordingly.

[0022] Optionally, the determined error input into the model as a model parameter is in the form of a power balance correction.

[0023] Optionally, the method further comprises receiving a series of measurements of the temperature of the outlet fluid, and determining a covariance between the series of measurements of the outlet fluid temperature and a corresponding series of predicted outlet fluid temperature, wherein the correcting parameter input into the model is dependent on the determined covariance. In such a way, the impact of noisy measurements can be reduced.

[0024] Optionally, the correcting parameter input into the model is dependent on the determined covariance by a applying a covariance dependent weight to the delta between the predicted outlet fluid temperature and the measurement of the outlet fluid temperature.

[0025] Optionally, the series of measurements of the temperature of the outlet fluid is taken over a time window of between 1 second and 1 minute, preferably 10 seconds. This balances the need to accurately determine covariance and the frequency of updating the correcting parameter.

[0026] Optionally, the correcting parameter corresponds to a measure of heat loss. Heat loss is a characteristic of leaks, damage, or other system-related issues.

[0027] Optionally, the correcting parameter corresponds to a stack resistance correction. Increased resistance is more likely to indicate stack damage.

[0028] Optionally, the method comprises adjusting the temperature of the inlet fluid with a component proportional to an error between a desired current and a measured current passing through the at least one electrolyser cell stack. The desired current may correspond to a desired production rate, or an available current (for example from a renewable energy source). This provides an increase / decrease in inlet temperature setpoint to speed up the transition and is particularly effective for large changes in current.

[0029] Advantageously, the method is used in a method of controlling an electrolyser system, wherein the electrolyser system further comprises a power control system for changing the electrolytic conversion rate within the at least one electrolyser cell stack. The method further comprising: receiving a desired a stack current; and controlling a stack voltage based on the received desired current and / or measured current.

[0030] Optionally, controlling the stack voltage comprises: i) changing a voltage across the at least one electrolyser cell stack; ii) allowing a fluid outlet temperature to change in response to said change in voltage; and iii) allowing the voltage to revert towards its initial value (i.e. the voltage level prior to the change) so as to normalise the electrolyser cell stack, preferably at a substantially isothermal operation, with the at least one electrolyser cell stack at a changed stack temperature (compared to the temperature prior to the voltage change), and thus a changed stack resistance and thus a changed stack current (corresponding to the a desired current level) and a changed electrolytic conversion rate.

[0031] Such a method leverages endo- or exothermic conditions to move the stack to a new operating condition.

[0032] It should be appreciated that the method described herein can be applied to a single electrolyser stack, or a multi-stack ‘array’. When controlling the inlet temperature for an array of electrolyser cell stacks, the inlet and outlet fluid temperatures may be of a combined inlet and a combined outlet of the array of electrolyser stacks.

[0033] Alternatively, the inlet and outlet fluid temperatures are an average inlet temperature and an average outlet temperature across the array of electrolyser cell stacks. In both arrangements the stacks are treated as one large stack which leads to simplified control.

[0034] Optionally, the method further comprises adjusting a voltage for a subset of the stacks in the array of electrolyser stacks in dependence on an outlet temperature measurement specific to the subset.

[0035] Stacks in an array mayvarydependingon age, design, manufacturing variability etc. and not every stack would be atthe same operating condition (e.g. thermoneutral) atthe same current and inlet temperature. Monitoring the outlet temperature for a specific stack (or subset of stacks) can identify such differences, and modifying the voltage can compensate for these differences whilst still maintaining a homogeneous inlet temperature, which may be difficult or impossible to control at an intra-array level.

[0036] Optionally, the voltage adjustment is related to a temperature delta between the average outlet fluid temperature of the array a nd the outlet fluid temperature for the subset of stacks. Preferably, this relation is a proportional relation. This allows control of the power supplied to subsets of stacks based on their deviation from an average stack.

[0037] Optionally, a subset of stacks is group of stacks sharing a power connection and / or one or more temperature sensors. Such a subset may be indistinguishable from an individual stack from a control perspective. If a stack has an independent power and one or more individual temperature sensors for the outlet fluid, it can be individually controlled.

[0038] Optionally, the method is iterated when a new outlet fluid temperature measurement is received. The outlet fluid temperature may be monitored continuously, or at specified intervals.

[0039] Optionally, the desired operating condition is one where the inlet fluid temperature substantially matches the outlet fluid temperature at a given current level, this is termed ‘galvanostatic thermoneutral’ operation. Optionally, the desired operating condition is dependent on an available power level. For example, if the electrolyser is being powered by a renewable energy source, the power level may fluctuate, and the controller consequently modifies the operating condition of the electrolyser to maximise production given the available power. As disclosed herein, this can be achieved by maintaining substantially thermoneutral operation at varying production rates, which is enabled by controlling a change in operating temperature.

[0040] Optionally, the or each electrolyser cell stack is a solid oxide electrolyser cell (SOEC) stack. Inlet fluid temperature control is particularly impactful for SOEC due to the strong temperature- dependencyon efficiency and degradation.

[0041] There are many possible forms of SOEC, using different electrochemically active electrolyte chemistries. For example, three well known electrolyte materials are yttria stabilized zirconia (YSZ), scandia stabilized zirconia (ScSZ) and gadolinium doped ceria (GDC or CGO).

[0042] The present invention is highly suitable for use with solid oxide electrolyser cells (or SOECs) operating at temperatures above 400 degrees C (generally known as intermediate temperature SOECs), or instead high temperature SOECs if above 700 degrees C. (Sometimes the cut-off between intermediate temperature and high temperature SOECs is 750 degrees C).

[0043] In some embodiments, the electrolyser cell system instead comprises a high temperature electrolyser cell with an operational stack temperature between 750 degrees C and 1100 degrees C.

[0044] Optionally, the or each electrolyser cell stack is used to electrolyse water in the form of steam.

[0045] According to an aspect of the invention there is provided a controller configured to perform the methods discussed herein.

[0046] According to an aspect of the invention there is provided a computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method according to the aspects above.

[0047] According to an aspect of the invention there is provided a non-transitory computer-readable medium with instructions stored thereon, that when executed by a processor, perform (or cause the processor to perform) the steps of the method of according to the aspects above.

[0048] Particular and preferred aspects of the invention are set out in the accompanying independent claims. Combinations of features from the dependent claims may be combined with features of the independent claims as desired and appropriate and not merely as explicitly set out in the claims. The term "comprising" as used herein to specify the inclusion of components also includes examples in which no further components are present.

[0049] The present invention will now be described, purely by way of example, with reference to the accompanying drawings, in which:

[0050] Figure 1 schematically shows a typical electrolyser cell, multiples of which may be stacked in a stack;

[0051] Figure 2 shows a schematic of an example of an electrolyser system with a single electrolyser cell stack and external stream flow paths for air electrode and fuel electrode sides of the stack with heat exchangers and heaters for input fluid temperature control; Figure 3 is a schematic diagram showing an embodiment of the present invention, showing the elements used for controlling an electrolyser system;

[0052] Figure 4 is a schematic diagram showing a further embodiment of the present invention, showing the elements used for controlling an electrolyser system;

[0053] Figure 5 shows plots of current, voltage and temperature when controlling an electrolyser system using the controller of Figure 4; and

[0054] Figure 6 shows a modification to the controller of Figures 3 or 4 when controlling multiple electrolyser cell stacks.

[0055] Referring first to Figure 1 , the basic structure and operation of a typical electrolyser cell 11 within an electrolyser 10 of an electrolyser system 20 is shown by reference to one fuel / electrolyser cell 11 of an electrolyser cell stack 12 - hereinafter a stack. It should be noted that other ancillary components related to the electrolyser cell 11 are included in an electrolyser system 20. These usually include heat exchangers, heaters, valves and sensors.

[0056] Figure 2 schematically shows an example of an electrolyser system 20, similar to that known from WO2023 / 012456, the entire contents of which are incorporated herein by way of reference. This shows example ancillary components, such as heat exchangers, heaters and sensors. These will be discussed in more detail below.

[0057] Figure 2 also shows that the electrolyser system 20 comprises a stack 12 which comprises a stack of electrolyser cells 11 , hereinafter referred to as cells. As shown, this example has five such cells, although generally a stack will have tens or even hundreds of cells - usually all in parallel, although some may use cells in series or combinations of the two. In some electrolysers, multiple stacks 12 may be provided, electrically connected in series or in parallel (referred to as a ‘stack array’ or ‘array’).

[0058] Referring now back to Figure 1 , the electrolyser cell 11 comprises an air electrode (anode for electrolysis) 33, a fuel electrode (cathode for electrolysis) 34 and an electrolyte 35. Such a structure for an electrolyser cell 11 is well known in the art. In this example, the electrolysis of water will be discussed, although other gases can similarly be electrolysed to split them into component parts, such as carbon dioxide into oxygen and carbon monoxide.

[0059] Water - here in the form of steam 43 from a water source - which may be a steam source if less internal heating of the water is desired in the electrolyser system 20 - is passed over the fuel electrode 34 via inlet 41 and hot air 42 is passed over the air electrode 33 via inlet 40. To power the electrolyser cell, an electric current is applied across the electrolyser cell 11 via electric terminals / connections 36, 37 at the air and fuel sides of the electrolyser cell 11 . These terminals may be positioned adjacent to one-another on one side or end of the stack 12 of cells 11 , for example by having stacked cells in parallel and extending one terminal to the other end of the stack 12 using a bus bar, as known in the art. Via the terminals 36, 37, a voltage can be applied across the stack 12 - and thus current through the cells 11 . The application of electrical power through an electrolyser 10 in this manner is well known in the art.

[0060] As a consequence of the electrical current, an electrolytic reaction occurs across the electrolyte 35, with oxygen ions passing across the electrolyte 35 from the fuel electrode 34 to the air electrode 33, and some of the steam breaking down into hydrogen on the fuel electrode side of the electrolyser cell 11 and oxygen is produced at the air electrode side.

[0061] The oxygen can be extracted via an air flow or sweep flow provided by the hot air 42, thus venting it out of an off-gas outlet 38 on the air electrode side of the electrolyser cell 11 . That output is generally oxygen enriched air (with the oxygen enriching the hot gas flow). The hydrogen can instead be extracted and vented out of another off-gas outlet 39 on the fuel electrode side of the electrolyser cell 11 . This off gas typically will be mixed with the remaining steam, as the splitting of the steam into oxygen and hydrogen is usually only in respect of a proportion of the supplied steam. The hydrogen is thus vented as ‘wet’ hydrogen on the fuel electrode side. Thus, the steam exiting the fuel electrode side is hydrogen enriched, and the air exiting the air electrode side is oxygen enriched.

[0062] Those off-gases will usually be at a similar temperature to the operational temperature of the electrolyser cell 11 . However, the specific delta from the input temperature will depend upon the amount of electrical power supplied to the electrolyser, and the internal resistance of the cells.

[0063] Such operational characteristics of electrolyser cells, including SOECs, are well known in the art.

[0064] Due to the high operating temperature of the electrolyser cell 11 (i.e. above 100 degrees C for a steam electrolyser, and in the case of a solid oxide electrolyser cell (SOEC) it is usually in excess of 400 degrees C), the heat of the off-gases from the off-gas outlets 38, 39 is able to be usefully used by the electrolyser system 20, rather than being wasted, for example to provide at least some of the heat for the steam generation on the input side of the stack, and likewise for heating the hot air entering the electrolyser, in advance of the inlets 40, 41 .

[0065] As shown in Figure 2, this may be achieved via heat exchangers 22, 50, 62. Heaters 36, 52, 58 can also be provided where required to increase the temperature of the fluid through flow lines 54, 56, 18, 24, 28 between components of the system 20, when needed. Bypasses 90 for the heat exchangers 22, 50, 62 might also be provided to reduce heating of the inflowing fluid, such as the one shown for heat exchanger 50 in Figure 2. In some embodiments, a heat exchanger can even be used to cool the fluid using a colder fluid source or fan.

[0066] For controlling the temperatures and flow of fluid around the electrolyser system 20, sensors 64, 66, 68, 70, 72, 74, 76 may be provided alongthe flow lines 54, 56, 18, 24, 28. These can also allow a control of flow rates of the respective fluids through the heat exchangers 22, 50, 62 or through the heaters 36, 52, 58 or bypasses 90 - for example for controlling the flow rate of the air 42, or the water 16, or the off-gases - herein from stack outlets 46, 48, which are connected to the outlets 38, 39 of the cells 11 . Flow rates and or temperatures of the input fluids at the two stack inlets 14, 44, which are connected to the inlets 40, 41 of the cells 11 , can thus be controlled as necessary, as discussed below.

[0067] A flow line or flow path represents a flow that passes through one or more pipe or line from one position to another. Therefore, it defines a fluid connection or fluid communication between points.

[0068] In Figure 2 there is a sweep gas supply flow path 18. It is a sweep gas supply pipe or line 18 and provides fluid communication from a sweep gas supply 16 (usually air) to an air electrode side stack inlet 14.

[0069] The stack 12 also comprises an air electrode side stack outlet 46, wherein an air electrode side exhaust product (off gas -usually oxygen enriched air) is expelled for utilization in anyone or more of various possible purposes - usually outside of the system 20, although its heat can be usefully utilized in the system 20, as discussed below.

[0070] Also entering the electrolyser system 20 is an external fluid stream 26. The external fluid stream 26 is a stream that is produced separate from the process of the electrolyser system 10. The external fluid stream 26 can be a hot flow of fluid or gas - e.g. an external heat source.

[0071] Whilst it is expected that the external fluid stream 26 will be an exhaust gas from another process and may be of low-grade heat (e.g. around 200°C), different sources can be utilised to provide the external fluid stream 26 with higher or lower grades of heat.

[0072] The externalfluid stream 26 forms an external stream flow path 28. The external stream flow path 28 passes through a first heat exchanger 62. The first heat exchanger 62 is also arranged across the sweep gas supply flow path 18 between the sweep gas supply 16 and the air electrode inlet 14. Therefore, the external stream flow path 28 and the sweep gas supply flow path 18 both pass through the first heat exchanger 62 and exchange heat in the first heat exchanger 62.

[0073] It is also shown in Figure 2 that an air electrode outlet flow path 24 from the air electrode side stack outlet 46 passes through a second heat exchanger 22 before passingout of the first exhaust 88 (e.g. for collection for subsequent use elsewhere). This second heat exchanger 22 is also arranged across the sweep gas supply flow path 18. The second heat exchanger 22 in this embodiment is arranged across the sweep gas supply flow path 18 between the first heat exchanger 62 and the air electrode inlet 14. Therefore, the air electrode outlet flow path 24 and the sweep gas supply flow path 18 exchange heat in the second heat exchanger 32.

[0074] During operation of the system 10, the exhaust gas (also known as off-gas) may contain highgrade heat (e.g. around 550°C for an SOEC electrolyser system). Therefore, the heat from the exhaust gas can be used to heat the sweep gas supply flow path 18 in the second heat exchanger 32. This is beneficial to provide sweep gas at the air electrode inlet 14 of high temperature to increase efficiency of the electrolysis reaction in the stack 12.

[0075] A first heater 36 is provided in the sweep gas supply flow path 18. In this embodiment this is positioned between the second heat exchanger 32 and the air electrode inlet 14. The first heater 36 is used for heating the sweep gas supply 16 in the sweep gas supply flow path 18 when needed. The first heater 36 can be an electric heater or a combustion heater as required.

[0076] The first heater 34 is also referred to as a trim heater as it provides small amounts of heat for fine- tuning the temperature at the air electrode inlet 14 to ensure a consistent and efficient electrolyser reaction in the stack 12.

[0077] A bypass flow path 30 is shown connectingto the sweep gas supply flow path 18. The bypass flow path 30 is connected at its first end 78 at a point on the sweep gas supply flow path 18 between the sweep gas supply 16 and the first heat exchanger 62. Therefore, the first end 78 is upstream of the first heat exchanger 62. The bypass flow path 30 is connected at its second end 80 at a point between the second heat exchanger 22 and the first heater 62. Therefore, the second end is downstream of the second heat exchanger 22.

[0078] The bypass flow path 36 bypasses the first and second heat exchanger 30, 32. This can allow for the sweep gas supply flow path 18 to avoid any heat exchange with the heat exchangers. This can be beneficial where the heat energy in the external stream flow path 28 and / or the air electrode outlet flow path 24 is of a level that is not required for the sweep gas supply flow path 18. Such situations can include start-up where the flow paths might be cold, or during cool down, where transfer of heat to the sweep gas supply flow path 18 is no longer required and the sweep gas supply 16 is being used to cool the system 10.

[0079] The bypass flow path 36 is operable by a control valve 82 or the like. The control valve can be a mechanical or electrical control valve set to a certain setting to open or close, or can be manually or automatically operated by a controller, which may include a processor and a memory that can be programmed to operate the control valve in certain situations as required.

[0080] A similar control valve may be provided for the bypass 90 for the heat exchanger 50, likewise connected to a or the controller.

[0081] In a preferred embodiment the controller is a single controller for the electrolyser system and thus also connected to a power, current orvoltage control system 108, as shown in Figure 3.

[0082] Returning to Figure 2, the control valve may be connected to, for example, one or more sensors 66, 72, 74, 76 provided on one or more of the flow lines between the components of the system to allow control of the flows, and thus the fluid temperatures. The controller is then connected to these sensors and potentially to the heaters and further flow valves or bypasses.

[0083] Whilst the above are presented as shown in Figure 2, it is highlighted that certain variations can be made. In particular, the presence of the first heat exchanger 62, the second heat exchanger 22, the first heater 36 and the bypass flow path 30 can be varied as required for the function of the system 10. For instance, a second heat exchanger 22 may not be required in some situations, such as where the external fluid stream 26 is high grade heat. Additional heaters can also be supplied, such as the illustrated third heater 58 on the air electrode outlet flow path 24, which can also be controlled to operate when required.

[0084] Referring next to a fuel electrode side of the system 20, a water supply 84 is provided and is connected via a water or steam supply flow path 56 to a fuel electrode side stack inlet 44 of the electrolysis stack 12. The fuel electrode side stack inlet 44 connects to the steam inlet 41 on the fuel electrode side of each electrolyser cell 11 of the electrolysis stack 12. The water supply flow path 56 provides a flow, pipe or line for the water - or steam - to pass into the electrolyser 10.

[0085] The electrolysis stack 12 also comprises a fuel electrode outlet 48 that is the outlet from the fuel electrode of the electrolysis stack 12, and is connected to the steam outlet 39 of each electrolyser cell 11 . The fuel electrode outlet 48 is connected to a second exhaust 86 by a fuel electrode outlet flow path 54. Wet hydrogen is the exhaust gas on this side of the electrolyser 10, and it can be collected for downstream processing and use.

[0086] A third heat exchanger 50 is provided in the water supply flow path 56 between the water supply 42 and the fuel electrode inlet 44. This has the bypass 90 provided for it. The fuel electrode outlet flow path 54 is also connected across the third heat exchanger 50 between the fuel electrode outlet 48 and the second exhaust 86 of the fuel electrode outlet flow path 54. Therefore, heat is exchanged between the fuel electrode outlet flow path 54 and the water supply flow path 56, although the amount of transfer might be controlled by the bypass 90.

[0087] The product, e.g. wet hydrogen, produced in the stack 12 and output at the fuel electrode outlet 48 can be of high heat energy (e.g. around 550° for an SOEC system). Therefore, this heat energy can be used to heat the water supply flow path 56. This can be beneficial as it is preferred that steam is supplied forthe electrolyser reaction. Therefore, heat energy can be transferred from the fuel electrode outlet flow path 54 to the water supply flow path 56 in the third heat exchanger 50. This heat exchanger 50 can be referred to as a recuperating heat exchanger. This can also be beneficial as the product, e.g., wet hydrogen, at the second exhaust 86 is preferred to be lower in heat energy for exporting to other processes and storage. Therefore, the transfer of heat from the product is beneficial.

[0088] A second heater 52 is provided on the water supply flow path 56 for heating the flow path. The second heater 52 is positioned between the third heat exchanger 50 and the fuel electrode inlet 44. Therefore, the second heater 52 is downstream of the water supply 84. The second heater 52 is similar to the first heater 36 in that it can be an electric heater or a combustion heater as required. The second heater 52 provides additional heating to the water supply flow path 56 as required. Sensors 64, 68, 70 and a control system can again be provided for this purpose.

[0089] In situations where there is high heat transfer at the third heat exchanger 50, the amount of heat energy supplied at the second heater 52 can be low. Whereas where there is low heat transfer at the third heat exchanger 50, e.g. during start-up, the heat energy supplied at the second heater 52 can be high. This can be of benefit during start-up where the system 10 is notyet hot.

[0090] The second heater 52 is also referred to as a trim heater for the same reasons as the first heater 34 as it provides small amounts of heat for fine-tuning the temperature at the fuel electrode inlet 44 to ensure a consistent and efficient electrolyser reaction in the stack 12, although both can be used for start-up heating as well.

[0091] As discussed above, therefore, it is possible to control the input temperature of the fluids entering the electrolyser 10, and in particular, the electrolyser cell stack 12 and thus the electrolyser cells 11 therein.

[0092] Figure 3 shows a diagram of an electrolyser system controller 300 and the associated inputs and outputs. The system controller 300 comprises power controller 302 and an inlet fluid temperature controller 304.

[0093] The power controller 302 controls a power supply 306 to the stack or stack array. This portion receives a current demand 310 and provides the voltage set point to the power supply to deliver the desired current to the stack (which is related to the rate of production from the electrolyser system).

[0094] The power controller 302 receives a current demand and a current measurement 318 at feedback unit 314. This unit determines the magnitude and direction of the change in current required. This is fed into a linear (proportional / integral (PI)) controller 316 which calculates the voltage set point for the power supply to meet the desired current level. The mechanism for changing the current level is discussed in more detail below.

[0095] The inlet fluid temperature controller 304 controls a heater controller 308 by outputting an inlet fluid temperature setpoint. The heater controller 308 may control one or more of the trim heaters, heat exchangers or fluid paths as described in Figure 2, or it could send a control signal to a separate system, for example a steam generator.

[0096] The inlet fluid temperature controller 304 a predictive model 328 and an extended state observer 330. Inputs comprise the current demand 310, a desired inlet vs outlet fluid temperature delta 320, stack boundary conditions 322 and a stack temperature measurement 324. Stack boundary conditions 322 allow for the predictive model to determine the impact of a changed inlet temperature set point. In other words, these are the parameters which are inputs into the model which may vary due to operating conditions. These may include parameters such as airflow rate, fuel (steam) flow rate, product (H2) rate, voltage measurements, and fuel utilization. They may also include information about the stack such as age, model number, cell design etc.

[0097] The predictive component enables a faster transition (compared to a traditional feedback loop) to an appropriate inlet fluid temperature setpoint as knowledge of the end state is leveraged. The extended state observer component improves the control by modifying the prediction model based on an observed outlet fluid temperature. In such a way, the controller is not solely reliant on an accurate model, nor solely reliant on reacting to feedback.

[0098] The predictive model 328 receives a delta inlet fluid temperature setpoint 320 and stack boundary conditions 332. Knowledge of the boundary conditions of the stack from stack data 322 enables a look-up table (or similar) to determine an operating temperature for a given current demand. Such a model may be based on finite element modelling of the physical properties of an electrolyser cell stack, and / or be derived from empirical test data. The optimum operating temperature is often a ‘thermoneutral’ temperature, where the input and output temperatures are substantially equal. The delta inlet fluid temperature setpoint 322 is used by the model to produce an output to modify the inlet fluid temperature setpoint signal at point 332.

[0099] The extended state observer 330 is present to correct for errors or uncertainties in the model when predicting steady state conditions. This is achieved by observing the behaviour of the stack (or array) and determining an error term to input into the predictive model as a model parameter. This error term may be in the form of a power balance correction which modifies the predictive model.

[0100] For example, if the outlet fluid temperature is different to the outlet fluid temperature calculated by the extended state observer (based on the current correction term), the correction term is updated. This process is continuously performed so that eventually the prediction model becomes correct at the current operating point. In such a way, the extended state observer fulfils the role of an integral action in a PID controller, but this method is more efficient than a PID controllerwhich can over or undershoot. The extended state observer will only integrate the error between the measured behaviour and the expected behaviour, reducing significantly the risk of over and undershooting due to the integral component.

[0101] In the example shown in Figure 3, this is achieved by the extended state observer 330 receiving an input from an outlet fluid temperature sensor 324. This is shown as a thermistor, but could be any appropriate temperature sensor. There may be several sensors for redundancy or for taking an average. The temperature may correspond to an internal stack temperature, which may be approximated by the temperature of an outlet fluid close to the stack. The extended state observer 330 receives (or has the ability to retrieve) stack boundary condition measurements 322, for example the current inlet fluid temperature or flow rate.

[0102] It should be appreciated that Figure 3 shows the proportional controller 326, predictive model 328 and extended state observer 330 as separate components, they may however be implemented as a single component. For example, as an application specific integrated circuit (ASIC) or field programmable gate array (FPGA) performing all functions of the inlet fluid temperature controller 304. Such a single component could also include the power controller 302. Temperature delta data 320 and stack boundary conditions 322 could be stored in an associated memory unit such as solid-state storage, or hard disk drive, or could be accessible via a wired or wireless connection to an external data storage system. Figure 4 shows a variation of the controller 300 of Figure 3, but with the addition of a component proportional to the error between the desired current and the measured current, used to give an increase / decrease in inlet temperature setpoint to speed up the transition.

[0103] A proportional controller 326 receives the error between the current demand 310 and the current measurement 318 and outputs a delta inlet fluid temperature setpoint signal which is proportional to the current error - an increased current demand results in a positive delta inlet fluid temperature setpoint and a decreased current demand results in a negative delta inlet fluid temperature setpoint. This delta inlet fluid temperature setpoint is added to the output of the predictive model 328 at summing point 322.

[0104] The current measurement 318 also feeds into a feedback unit 314, thus the signal into the proportional integral controller 316 incorporates a component proportional to the difference between the measured current and the desired current level. This results in an inlet fluid temperature setpoint which initially overshoots the eventual steady state temperature, speeding up the transition.

[0105] It should be appreciated that limits may be placed on maximum & minimum inlet fluid temperature setpoints to maintain safe operation, these limits override the output of the controller.

[0106] Figure 5 shows a series of plots showing the temperature response of an electrolyser undergoing a series of transitions between current levels controlled by a controller as described above with reference to Figure 4. It should be noted that these charts show control at thermoneutral conditions, namely that the inlet fluid temperature is controlled to substantially match the outlet fluid temperature, however controlling for an offset would use a similar approach.

[0107] Electrolysers operate most efficiently when operated at thermoneutral temperature, which has an associated thermoneutral voltage. This is approximately 1 .28V per cell for water electrolysis. The amount of product (e.g. Hydrogen) produced is a function of the current through the cell; to maintain thermoneutral operation for differing production rates, the temperature changes. This change in temperature changes the area specific resistance of the cells, thus changing the current level for thermoneutral operation due to the application of Ohm’s law.

[0108] During system operation, there will be cases where the operating current and the corresponding production rate needs to be changed. This may be because the system is harvesting available power to generate Hydrogen, or because the amount of Hydrogen demanded by a downstream consumer has changed. In this event it is an advantage if the transition can be made quickly, rather than being constrained to match the rate at which the system operating temperature can be changed.

[0109] Transitioning between current setpoints can occur without deviating from the thermoneutral voltage, rather the change solely being brought about by modifying the inlet fluid temperature setpoint as described above. However, controlling the voltage to deviate away from thermoneutral conditions can be utilised during load point changes to allow rapid changes in operating current and to bring the system back to a stable thermoneutral operating point at the new power level. The combination of deviating from the thermoneutral voltage and utilizing the inlet fluid temperature control leads to significantly faster transition times, as well as reducing the reliance on trim heaters etc. which reduce the system efficiency. Figure 5a shows the case where the system moves to a lower power operating point with a lower Hydrogen production rate. The current demand from the power supplier is be reduced using a relatively fast ramp rate at time ti.

[0110] Initially the reduction in current is achieved by reducing the voltage applied to the stack(s), and this moves the stack from a thermoneutral to an endothermic operating mode before thermoneutral operation is restored at the new operating point.

[0111] The inlet temperature setpoint also drops substantially at point ti.from the initial thermoneutral temperature Toto temperature Ti which near, but below, the eventual changed thermoneutral temperature T2. This difference is in part an effect of the discrepancy between the model and the actual stack and is eventually corrected by the extended state observer correction.

[0112] The inlet fluid temperature follows the inlet fluid temperature setpoint, but with a lag due to thermodynamics of the system, and system design.

[0113] The reducing cell temperature after time ti increases the electrical resistance of the cells and so the voltage starts to rise again in order to maintain the desired current set-point.

[0114] The voltage rise back towards thermoneutral voltage progressively reduces the level of endotherm associated with the rate of Hydrogen production.

[0115] As the outlet temperature reduces, the error between the inlet fluid temperature setpoint and the outlet fluid temperature reduces and the inlet temperature setpoint moves asymptotically to the new thermoneutral temperature T2.

[0116] Figure 5b shows a transition to a higher current operating point (with an increased Hydrogen production rate). Initially the higher current is achieved by increasing the voltage applied to the stack(s), and this moves the stack from a thermoneutral to an exothermic operating mode before thermoneutral operation is restored at the new operating point.

[0117] The system is operating at thermoneutral in steady state at point t0, followed by a request to change the current set point at time ti causing voltage to ramp up to point Vi where the voltage has saturated to the maximum allowed level. As the cell voltage is now above the thermoneutral voltage Vo, the stack is in an exothermic operating mode and so the outlet temperature rises.

[0118] It should be noted that the current lags the desired current level until the stack temperature increase sufficiently reduces the resistance of the cell to enable desired current level.

[0119] The inlet fluid temperature setpoint increases from Toto Ti, which is above new thermoneutral temperature T2. This overshoot is caused in part by the proportional component related to the difference between desired current and the measured current. Such an overshoot allows the current to ramp up quicker by rapidly reducing the resistance of the cell. It should be noted that the inlet fluid temperature setpoint is shown to flatline, this is due to a restriction on the maximum inlet fluid temperature (related to safe operation, or physical constraints of the system). It is also noted that the inlet fluid temperature lags the inlet fluid temperature setpoint due to thermodynamics of the system, and system design - for example the power of trim heaters.

[0120] As the error between the outlet fluid and the inlet fluid temperature setpoint decreases, the inlet fluid temperature setpoint decreases towards the revised thermoneutral temperature T2.

[0121] Once the current is at the desired level, the voltage reduces back towards thermoneutral voltage Vo; this progressively reduces the level of exotherm. The plot shows the inlet fluid temperature setpoint then undershooting T2, this is due to the current temporarily exceeding the desired current, demonstrating the controller reacting to changing conditions.

[0122] Once the desired current is met, the inlet fluid temperature setpoint slightly overshoots T2before approaching the T2asymptotically.

[0123] An increasing load step can cause a rapid rise in outlet temperature, and so the size of the load step that can be made may be limited by the maximum stack operating temperature limit. This means that large steps in load may need to increase current in an initial step to the point where the maximum stack outlet temperature is reached and then increase current more gradually as the bulk stack temperature rises.

[0124] Furthermore, any increasing load step must also respect the maximum stack voltage limit, because subjecting the cells to excessive voltage, even for a limited time, can be detrimental to stack durability. The maximum voltage is not a constant value, but rather a function of operating temperature, higher operating temperatures acting to reduce the tolerance of the stack to high voltages.

[0125] The maximum load point for thermoneutral operation will therefore be defined by the load at which the uniform stack temperature and the thermoneutral voltage coincide with the safe operating envelope of the stack.

[0126] One particularly advantageous implementation is to load follow a renewable energy source such as solar and / or wind power. Such sources have fast changing components (e.g. clouds passing overhead or gusts of wind) and also components which change over a longer timescale (e.g. the diurnal cycle, or a stormy period). This control method can be used to adapt to the changing current demand (essentially, current available) to maximise the usage of the renewable energy resource.

[0127] It should be noted that inlet fluid temperature setpoints may briefly go outside of the normal operating window, but there would still be maximum and minimum temperature setpoints to maintain safe operation.

[0128] Multi-stack operation

[0129] The following section considers the scenario where multiple stacks are present.

[0130] One strategy is to treat the plurality of stacks as a single stack. In such a strategy, the properties of each stack and the input and output fluids are considered identical. The inlet fluid temperature setpoint is the same for all stacks. It is convenient to deliver the same inlet fluid to all stacks in an array, supplying a different temperature inlet fluid may require a complicated system design and potentially introduce efficiency losses.

[0131] The outlet fluid temperature measurement represents an average of the various stack outlet fluid temperatures. In such a control strategy, the controller and method of control described above could be used.

[0132] However, under such a control strategy, stacks which deviate from the average would not be operating optimally. There is scope for optimisation as it is possible to control the power to stacks (or groups of stacks) within an array and to introduce additional temperature sensors without significantly changing system design. Figure 6 shows additions to the controller 300 of Figures 3 or 5 when a controlling an array of stacks to account for variations between stacks. The primary difference is the addition of linear controllers 714-1 , 714-2 for individual stacks (or group of stacks). These controllers 714-1 , 714-2 modify the power for that stack (or group of stacks) by an amount related to the difference between the stack outlet fluid temperature and the outlet fluid temperature across all stacks. Each controller has an associated temperature sensor 724-1 , 724-2 for this reason. A stack array outlet fluid temperature sensor 724-a may also be provided; however, this temperature could be determined (or approximated) by averaging the outlet fluid temperatures of each stack.

[0133] As described above with reference to Figure 3, the array power controller 702-a receives a desired current level 710 and current measurement 718. The array power controller 702-a, similarly to the power controller 302 for the single stack case, outputs an average desired voltage for the stack array in order to maintain the desired current level 710.

[0134] Each stack power controller 714-1 , 714-2 then adds a voltage delta (which could be positive or negative) to this average voltage to maintain the outlet fluid temperature of each stack (or group of stacks) at the same temperature as the stack array outlet fluid temperature sensor 724-a and outputs this value to the power supply 706 which provides the determined voltage to each stack (or group of stacks).

[0135] The total current demand is satisfied due to the array power controller 702-a changingthe average voltage to meet desired total current level 710, but each stack (or group of stacks) would receive a different split of the current depending on their condition. Older, more degraded, stacks have a higher area specific resistance (ASR) at thermoneutral conditions, meaning that to maintain a thermoneutral voltage the current is lower than a stack with a lower ASR. This method, at steady state, thus prioritises stacks in a better condition by preferentially passing more current to the stacks with a lower thermoneutral operating temperature. This control system ensures the stacks operate at the same temperature (due to the common inlet fluid temperature setpoint), as such the voltage and current for a given stack (or group of stack) depends on the ASR of that stack (or group of stacks).

[0136] During transitions in current demand, the voltage supplied to each stack deviates from the thermoneutral voltage (as described with reference to Figures 4 and 6), however, stack control rules are imposed so that the maximum or minimum stack voltages are not breached during a transition.

[0137] In such a way, individual stacks (or groups of stacks) are supplied with a different power level which depend on theirthermodynamic properties. Such properties change over life, primarily due to degradation, and as such this method allows for an array to use different age stacks, stacks which do not age consistently (e.g. due to manufacturing variability), or even stacks of different design. This is controlled by monitoring outlet fluid temperature and not by any a priori knowledge of the properties of the stack.

[0138] Kalman filter

[0139] In one example, the approach of predicting and correcting can be provided by a Kalman filter, A Kalman filter operates by applying a predictive model of the underlying system (which could be non-linear) and uses measurements to correct the model. A Kalman filter may be considered a special case of a recursive least squares fitter.

[0140] A Kalman filter essentially has two phases, a ‘prediction’ phase where the state of the system is predicted, based on a model of the system. An ‘update’ phase then takes measurements to correct the prediction. However, measurements are ‘noisy’ and as such may not be entirely reliable. The magnitude of the update to the predicted state is dependent on the reliability of the measurements. In one example, the reliability of the measurements is defined by its covariance with the predicted state. The lower the covariance, the more reliable the measurement is deemed, and as such is given a greater weight in a correcting factor. A low covariance does not necessarily mean the prediction is accurate. For example, if the prediction is consistently off by a certain delta, and if measurements are perfect, the covariance would be zero. In such an example, the correcting factor would represent the delta between prediction and measurement. Where the covariance is non-zero, the magnitude of the correcting factor is smaller.

[0141] An overarching gain (Kalman gain) may then be applied to tune the overall impact of the correcting factor.

[0142] Such filters are typically used in guidance, navigation, or tracking of moving objects where a location is predicted by physical laws of motion and then updated by (noisy) measurements. Pursuant to the present disclosure, it has surprisingly been found that Kalman filters can be used for controlling the temperature of an inlet fluid of an electrolyser. This is for at least the following reasons:

[0143] • There is a delay between control action and system reaction which can be factored in when calculating the covariance by using a series of measurements and their corresponding prediction. Depending on the design of the system the delay from control action and outlet temperature change may be between 1 minute and 60 minutes, in one example approximately 15 minutes.

[0144] • Kalman filters (especially extended Kalman filters) are also able to approximate nonlinear systems

[0145] • The system model may not incorporate all variables, especially for a generic controller which is able to control a variety of different electrolysers or fuel cells, the correction applied by a Kalman filter is well suited to account for missing variables in models.

[0146] Generally, the temperature of an inlet fluid of at least one electrolyser cell stack can be controlled by predicting, based on a model, an outlet fluid temperature for a desired operating condition. An inlet fluid temperature setpoint is set based on the predicted outlet fluid temperature. A series of measurements of the outlet fluid temperature is then received over a specified measurement window. A correcting parameter is then input into the model based on the difference between the measurement and the prediction. The magnitude (weight) of the correcting parameter is dependent on the reliability of the measurements, which in one example is defined as the covariance of the measurements and the prediction within a measurement window.

[0147] The measurement window may be between 1 second and 1 minute, preferably 10 seconds. A longer window means a more accurate covariance measurement, but means a less frequent change to the correcting parameter.

[0148] The following gives a specific example of using an extended Kalman filter to control the temperature of an electrolyser:

[0149] The voltage of a solid oxide electrolyser stack can be modelled by following equation: stack ()CV + stack_corr ' stack + ^(oss

[0150] Equation 1 where Vocvis the open circuit voltage for the present operating conditions,

[0151] Viossis the voltage losses (e.g. leakage) for the present operating conditions, and

[0152] Rstack_corr isacorrection to account for the discrepancy between the measured and predicted stack voltage.

[0153] So, if the model is perfect, Rstack_corr=0 •

[0154] The temperature differential equation can be modelled by the following equation: dTstack> lstack■ Vstack+ f (operating conditions) + Q dtstack ' -Pstack

[0155] Equation 2 where f(operating conditions) is a function of the operating conditions (e.g. whether the stack is exothermic, thermoneutral, or endothermic). The operating conditions include input and output flow rates, and their temperatures, and accounts for energy balance of what’s going in and what’s coming out of the stack. Understanding of the underlying (expected) chemical reaction is embodied in this function;

[0156] <2 is the heat loss to the environment;dTstackjs the derivative of the stack temperature with respect to time; dtmstack isthe mass of the stack; andcPstack isthe specific heat capacity of the stack.

[0157] Equations (1) and (2) are combined to form a nonlinear state-space model of the stack temperature dynamics. The parameters Rstack_corr and Qaretreated as unknown and timevarying, influenced by factors such as stack degradation and ambient conditions. These parameters are incorporated into the state vector of the EKF.

[0158] State Vector Definition

[0159] The EKF estimates the following state vector:

[0160] T1stack x(t) = Rstack_corr

[0161] Q

[0162] Equation 3

[0163] EKF Algorithm Overview

[0164] The EKF operates in discrete time steps and comprises two main phases:

[0165] Prediction Phase

[0166] The state vector is propagated forward using the nonlinear model derived from Equations (1) and (2). The system model is linearized around the current estimate using the Jacobian matrix of partial derivatives.

[0167] Update Phase Sensor measurements (e.g., stack temperature and voltage) are used to correct the predicted state. The measurement model is also linearized using its Jacobian. The Kalman gain is computed to optimally combine the predicted state and the measurements, accounting for model and sensor noise.

[0168] In such a way a Kalman Filter, and in particular an Extended Kalman Filter (EKF) can be used to correct the model due to the absence of these unknown stack parameters, and as such improve predictions for model-based control strategies and diagnostics.

[0169] An EKF is better suited to controlling non-linear systems compared to a Kalman Filter which assumes linearity. However, an EKF is computationally more complex and may be more sensitive to tuning parameters.

[0170] Other applications

[0171] As well as controlling the temperature of an inlet fluid of at least one electrolyser cell stack, the methods described herein can also have diagnostic use for any electrochemical cell stack (i.e. fuel cell or electrolyser).

[0172] If the predicted behaviour of the stack deviates from the actual, this is a sign that the stack is not performing as it should, for example due to degradation or damage. This difference can be detected by monitoring the value (i.e. magnitude and sign) of the correcting factor over time to determine a stack condition. If the correcting factor is in the same direction and getting larger, this is a sign that the stack is degrading. If the correcting factor exceeds a given threshold, the controller can recommend a change in operating conditions or shut down to prevent further damage. The rate of change of degradation can also be used to predict a future failure state, and / or a stack lifetime. is a specific parameter which is linked to stack degradation and possible failures in the near future.

[0173] Similarly, one can use Q (heat loss) to detect leaks and similar system issues. If this increases overtime, it may also be a sign of stack degradation or damage.

[0174] Increased resistance is more likely to indicate stack damage (high degradation). Increased heat loss could be due to an issue with the thermal measurement or a leak (air or fuel) in the system.

[0175] On the other hand, if the correcting factor, and specifically Rstack_corr, varies suddenly (or periodically), this may be a sign of a sudden (or periodic) change to inlet conditions.

[0176] Similar considerations can be applied to fuel cell (e.g. SOFC) stack technology as a diagnostic tool.

[0177] In fuel cell mode, the differential equation shown in Equation 1 above is similar, but l*V has a negative sign:

[0178] Equation 4

[0179] The function of the operating conditions g(operating conditions) is different because different gas species and different chemical reactions are present in fuel cell mode compared to electrolysis mode. This negative sign is also present in the corresponding equation to Equation 2 above: stack QCV Rstack_corr ' stack + ^loss

[0180] Equation 5

[0181] The sign of Rstack_corr ’ Istack is negative because the current is extracted from the stack in fuel cell mode instead of supplied in electrolysis mode.

[0182] Rstack_corr and Q can be estimated using the same process as described above and used to identify a stack condition. A stack condition includes stack damage, degradation, fluid leakage (air or fuel), electrical short status, lifetime prediction, or thermal management status. The stack condition can extend to the status of surrounding systems which impact the stack, such as leaks or thermal management issues.

[0183] The operation of the electrochemical cell stack may be modified in dependence on the determined stack status. For example, moving to a safer operating mode if damage is detected. Similarly, an alert to an operator may be issued, alerting an operator to conduct further diagnostics, or to manually change the operating conditions. The present invention has therefore been described above purely by way of example with reference to the accompanying drawings. Modifications in detail may be made to the invention within the scope of the claims as appended hereto.

Claims

Claims1 . A method of controlling the temperature of an inlet fluid of at least one electrolyser cell stack, the at least one electrolyser cell stack comprising at least one fluid inlet for the inlet fluid and at least one fluid outlet for an outlet fluid, the method comprising: predicting, based on a model, an outlet fluid temperature for a desired operating condition; and setting an inlet fluid temperature setpoint based on the predicted outlet fluid temperature; the method further comprising: receiving a measurement of the temperature of the outlet fluid; determining a delta between the predicted outlet fluid temperature and the measurement of the outlet fluid temperature; and inputting a correcting parameter into the model based on said determined delta.

2. The method of claim 1 further comprising monitoring the correcting parameter, and determining an electrolyser stack condition based on the value of the correcting parameter.

3. A method of determining a stack condition of at least one electrochemical cell stack, the electrochemical cell stack comprising at least one fluid inlet for the inlet fluid and at least one fluid outlet for an outlet fluid, the method comprising: predicting, based on a model, an outlet fluid temperature for a desired operating condition; and setting an inlet fluid temperature setpoint based on the predicted outlet fluid temperature; the method further comprising: receiving a measurement of the temperature of the outlet fluid; determining a delta between the predicted outlet fluid temperature and the measurement of the outlet fluid temperature; determining a correcting parameter for the model based on said determined delta; monitoring the correcting parameter, and determining a stack condition based on the value of the correcting parameter.

4. The method of claim 2 or 3 wherein the stack condition depends on the rate of change of the correcting parameter.

5. The method of any one of claims 2 to 4 wherein the stack condition is at least one of, a degradation status, a damage status, a predicted lifetime, thermal management status, leak status, and a future failure state.

6. The method of any preceding claim wherein the determined correcting parameter inputted into the model is in the form of a power balance correction.

7. The method of any preceding claim comprising:receiving a series of measurements of the temperature of the outlet fluid, and determining a covariance between the series of measurements of the outlet fluid temperature and a corresponding series of predicted outlet fluid temperature, wherein the correcting parameter input into the model is dependent on the determined covariance.

8. The method of claim 7 wherein the correcting parameter input into the model is dependent on the determined covariance by a applying a covariance dependent weight to the delta between the predicted outlet fluid temperature and the measurement of the outlet fluid temperature.

9. The method of claim 7 or 8 wherein the series of measurements of the temperature of the outlet fluid is taken over a time window of between 1 second and 1 minute, preferably 10 seconds.

10. The method of any preceding claim wherein the correcting parameter corresponds to a measure of heat loss.11 . The method of any preceding claim wherein the correcting parameter corresponds to a stack resistance correction.

12. The method of any preceding claim further comprising adjusting the inlet fluid temperature setpoint with a component proportional to an error between a desired current and a measured current passing through the at least one electrolyser cell stack.

13. The method of any preceding claim being used in a method of controlling an electrolyser system, wherein the electrolyser system further comprises a power control system for changing the electrolytic conversion rate within the at least one electrolyser cell stack, the method further comprising: receiving a desired a stack current; and controlling a stack voltage based on the received desired current and / or measured current.

14. The method of claim 13 wherein controlling the stack voltage comprises i) changing a voltage across the at least one electrolyser cell stack; ii) allowing a fluid outlet temperature to change in response to said change in voltage; and iii) allowing the voltage to revert towards its initial value so as to normalise the electrolyser cell stack with the at least one electrolyser cell stack at a changed stack outlet temperature, and thus a changed stack resistance and thus a changed stack current and a changed electrolytic conversion rate.

15. The method of any preceding claim wherein controlling the temperature is for an array of electrolyser cell stacks, wherein the measured outlet fluid temperature is of a combined outlet fluid of the array of electrolyser stacks.

16. The method of any of claims 1 to 14 wherein controlling the temperature is for an array of electrolyser cell stacks, wherein the measured outlet fluid temperature is an average outlet fluid temperature across the array of electrolyser cell stacks.

17. The method of claim 15 or 16 comprising adjusting a voltage supplied to a subset of the stacks in the array of electrolyser cell stacks in dependence on an outlet temperature measurement specific to the subset of stacks in the array.

18. The method of claim 17 wherein the voltage adjustment is related to a temperature delta between the average outlet fluid temperature of the array and the outlet fluid temperature for the subset of stacks in the array.

19. The method of claim 17 or 18 wherein the subset of stacks comprises a group of stacks sharing a power connection and / or one or more temperature sensors.

20. The method of any preceding claim further comprising iterating the method when a new temperature measurement is received.

21. The method of any preceding claim wherein the desired operating condition is one where the inlet fluid temperature substantially matches the outlet fluid temperature at a given current level.

22. The method of any of claims 1 to 20 wherein the desired operating condition is one where the inlet fluid temperature is offset from the outlet fluid temperature by a predetermined amount at a given current level.

23. The method of any preceding claim wherein the desired operating condition is dependent on an available power level.

24. The method of any preceding claim, wherein the or each electrolyser cell stack is a solid oxide electrolyser cell stack.

25. The method of any preceding claim, wherein the or each electrolyser cell stack is used to electrolyse water in the form of steam.

26. A controller for controlling the temperature of an inlet fluid of at least one electrolyser cell stack, the at least one electrolyser cell stack comprising at least one fluid inlet for the inlet fluid and at least one fluid outlet for an outlet fluid, the controller comprising: means for predicting, based on a model, an outlet fluid temperature for a desired operating condition; and an output unit for setting an inlet fluid temperature setpoint based on the predicted outlet fluid temperature; the controller further comprising: a receiver for receiving a measurement of the temperature of the outlet fluid; means for determining a delta between the predicted outlet fluid temperature and the measurement of the outlet fluid temperature; andmeans for inputting a correcting parameter into the model based on said determined delta.

27. Acomputer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of any of claims 1 to 25.

28. A non-transitory computer-readable medium with instructions stored thereon, that when executed by a processor, perform (orcause the processor to perform) the steps of the method of any of claims 1 to 25.

Images