Control scheme for a modular multilevel converter

Abstract

There is disclosed herein a method for controlling a grid-connected modular multilevel converter, MMC (100), wherein the MMC (100) comprises a plurality of phase arms (104a-1, 104b-1, …, 104c-1), and each phase arm 5 (104a-1, 104b-1, …, 104c-1) comprises a plurality of series-connected converter cells (106) each having respective DC units (108). The method comprises determining a dissimilarity in loading between DC units (108) in a first phase arm (104a-1) and, in response to determining the dissimilarity, determining a first phase arm power to be delivered by the first phase arm 10 (104a-1). The method further comprises determining a cell power to be delivered each cell (106) in the first phase arm (104a-1) by distributing the first phase arm power among the cells (106) in the first phase arm (104a-1), and controlling the cells (106) in the first phase arm (104a-1) to deliver their respective determined cell power. There is further disclosed herein an MMC 15 converter (102-1) configured to carry out such a method, and a system (100) comprising a plurality of such MMC converters (102-1, 102-2, …, 102-N).

Description

CONTROL SCHEME FOR A MODULAR MULTILEVEL CONVERTERTechnical Field

[0001] The present disclosure relates to modular multilevel converters, MMCs, and control schemes therefor. More particularly, the present disclosure relates to an improved control scheme for an MMC that has imbalanced DC loads between cells, and an MMC having a controller that implements such an improved control scheme.Background

[0002] A Modular Multilevel Converter (MMC) is a type of power electronic converter used for medium- and high-voltage direct current conversion. MMCs typically comprise several hundreds of series-connected submodules, also referred to as ‘converter cells’ or simply ‘cells’, each cell contributing a portion of the converter’s overall output. This topology creates multiple output voltage levels, providing improved efficiency and voltage control in high- power electrical systems. Each phase (or ‘leg’) of an MMC, of which there are typically three, may comprise two arms of series-connected cells.

[0003] The cells may comprise two or more insulated-gate bipolar transistors (IGBTs), e.g., arranged in a half-bridge or full-bridge arrangement, and a DC unit, typically a capacitor. This capacitor stores the DC energy and contributes to the multilevel voltage waveform generation. The IGBTs in the cell control the charging and discharging of this capacitor.

[0004] MMCs are being explored for use in hydrogen electrolyzer installations. The hydrogen produced by electrolyzer stations using renewable energy holds great potential for decarbonizing various sectors of global energy system including long-haul transportation, power generation, and steel / ammonia production etc. Moreover, electrolyzer stations can also act as versatile power-to-gas energy storage devices that can help to buffer and facilitate large-scale integration of intermittent renewable energy. Hydrogen electrolyzer units can serve as DC units in place of capacitors in MMC cells.

[0005] Electrolyzer stations also have the capability to serve as flexible responsive load that can support the AC grid during transients. On equippingother stored energy sources such as batteries, fuel cells, etc., as DC units in the MMC, the electrolyzer station can also behave as a responsive source, thereby becoming an important asset to enhance the security and stability of power grid.

[0006] However, with different DC units used in an MMC, an issue of imbalanced DC loads may arise, leading to a desire for improved control schemes for an MMC.Summary

[0007] Aspects of the present disclosure are directed at addressing at least one or more of the above-mentioned issues, with a view to enabling an improved electrolyzer station based on an MMC converter topology. It is realized as a part of the present disclosure that a chain-link MMC unit is not only susceptible to distort a DC unit’s operation but may also fail to provide grid services when its converter cells operate at different load conditions.

[0008] Such differences in the converter cell loading are rather common, especially in the context of an electrolyzer station, and can occur either when converter cells have DC units of different characteristics such as unequal aging conditions or when converter cells and / or DC units are lost / bypassed due to fault. Thus, aspects of the present disclosure provide an improved control scheme that enables the chain-link MMC unit to operate its series- connected converter cells independently at different load conditions without compromising on the grid support feature and grid code requirements.

[0009] More particularly, according to an aspect of the present disclosure, there is provided a method for controlling a grid-connected MMC, wherein the MMC comprises a plurality of phase arms, and each phase arm comprises a plurality of series-connected converter cells each having respective DC units.

[0010] The method first comprises determining a dissimilarity in loading between DC units in a first phase arm. The dissimilarity may be detected using any suitable means, such as using DC voltage and / or current measurements for one or more of the DC units. Preferably, each DC unit liable to give rise to a dissimilarity will have its DC voltage and current monitored. As used herein, the ‘loading’ of a DC unit may refer to thecapability of the DC unit to provide and / or consume DC power. For example, an MMC used in an electrolyzer station may comprise both electrolyzers and batteries as DC units, such that the electrolyzers are dissimilar to the batteries in their inability to provide DC power.

[0011] In response to determining the dissimilarity, the method further comprises determining a first phase arm power to be delivered by the first phase arm, where the first phase arm power may be a constraint on an overall apparent power, an active power, or a reactive power. For example, the first phase arm power to be delivered by the first phase arm may be based on an overall phase apparent power to be delivered for a phase on a plant level, which may be shared among the MMC converter and one or more other MMC converters. This overall control constraint may in turn be based on grid and / or DC unit requirements, such as, for example, a request from the grid control for a particular amount of reactive power, or the like.

[0012] The method then comprises determining a cell power to be delivered each cell in the first phase arm by distributing the first phase arm power among the cells in the first phase arm, and controlling the cells in the first phase arm to deliver their respective determined cell power. By contrast to conventional MMC control schemes, the contribution of each cell of the phase arm is controlled individually, thereby allowing an individualized accounting for differences in DC units that may have triggered the determination of the dissimilarity in loading. Preferably, the cell power to be delivered a cell is based on information about the DC unit corresponding to said cell, e.g., the type of DC unit, the availability, bypass status, state of health, state of charge, or the like.

[0013] Since the cells (e.g., the AC terminals thereof) are connected in series, all converter cells in a phase arm will operate at same current irrespective of their individual power loadings. As a result, independent operation of converter cells in a phase arm is limited when each converter cell operates at different power loading. Such differences in power loading among converter cells are quite common and can occur either when converter cell is lost / bypassed due to fault in DC unit or cell itself or when converter cells haveDC units with different aging conditions, operating points, and manufacturer types.

[0014] Controlling a cell to deliver a determined cell power may comprise adjusting the modulation index of the cell, where the modulation is preferably capped at a value of 1 , as discussed in more detail below. The modulation index of a converter cell is the ratio of its input AC voltage peak over its DC voltage. The input AC voltage of a converter cell is dependent on the phase arm voltage and the ratio of its individual power loading to the total phase power. Thus, the modulation index of the converter cell is affected with its loading condition. An increased modulation index may be associated with an increased cell apparent power, and vice versa.

[0015] According to a first example, determining the first phase arm power to be delivered by the first phase arm comprises determining a maximum deliverable active power for the first phase arm and setting a first phase arm active power to said maximum deliverable active power. Optionally, according to this first example, the method further comprises determining a phase arm power for each of the other phase arms by equating their respective phase arm active powers to the first phase arm active power.

[0016] Different converter cell loading can cause power unbalances between the phase arms of a unit. A chain-link MMC converter in a wye (Y) configuration with floating neutral point may have a constraint that mandates summation of its phase arm currents to be zero. As a result, the phase arms of an MMC converter with dissimilar active powers may operate only if a commensurate imbalance in reactive power is introduced between the phase arms. However, such an approach may be undesirable, as all phases of the unit at the PCC side may not operate at unity power factor (UPF), thereby violating the grid code requirements.

[0017] The zero neutral current constraint can be relaxed either by connecting the neutral point back to source or by adopting a delta configuration, for example. Even though these configurations can ensure UPF at PCC during unbalanced active power, the unit may draw unbalanced currents from the PCC which could compromise the grid code requirements.

[0018] Hence, the balancing of active power between phase arms advantageously enables satisfaction of grid code requirements and an attainment of UPF at PCC.

[0019] The MMC controls scheme may comprise determining that one or more cells in the first phase arm are overmodulated, and reducing a cell active power or a cell reactive power for the one or more overmodulated cells to thereby reduce the apparent power of the one or more overmodulated cells. As discussed above, the cell apparent power may be associated with the modulation index for the cell. Hence, by reducing the active or reactive power for an overmodulated cell, the apparent power can be reduced, thereby reducing the modulation index.

[0020] In an event of overmodulation (which may be expressed as, for example, m > 1), the input AC voltage of the converter cell may get saturated, thereby introducing undesirable saturated peaks in its phase arm voltage. These saturated peaks may not only distort the phase arm current but may also deviate the phase arm current away from the desired reference waveform. As a result, converter cells in the phase arm may fail to deliver the desired (reference) power to its respective DC units and the distorted phase arm currents drawn by the unit can violate the grid code requirements at the point of common coupling (PCC), i.e. , the connection to the grid.

[0021] Determining that one or more cells are overmodulated comprises determining that one or more cells have a modulation index exceeding 1 (one), above which these undesired effects may begin to affect the output waveform. The cell reactive power for the overmodulated cells may therefore be reduced to thereby return the modulation index to one (or less). It may be preferred, as a default, to operate cells with a modulation index of below 1 , such as 0.7 > m > 0.9, to allow some headroom for minor increases in the modulation index during operation of the MMC, without the modulation index exceeding 1 .

[0022] Reduction in cell active power can also reduce the apparent power. However, it may be preferred that active power of cell remains unchanged such that that the production from DC unit connected to cell remains unaffected. For example, considering cell connected to an electrolyzer, areduction in active power of cell may result in an undesirable drop in hydrogen production. Therefore, reducing the cell reactive power is preferred to reducing the cell active power.

[0023] According to one example, reducing a cell reactive power for the one or more overmodulated cells comprises determining a first phase arm reactive power, and redistributing cell reactive power from the one or more overmodulated cells to other cells in the first phase arm to thereby maintain the first phase arm reactive power.

[0024] Such an approach may limit the modulation index of converter cell particularly in situations where there is a higher phase arm reactive power.

[0025] If total reactive power requirement for the phase arm is lower, then such technique may be less effective in limiting the modulation index.

[0026] Hence, according to another example, which may be implemented instead of or in addition to the previous example, the total reactive power of the phase arm is altered to change its total apparent power and phase arm voltage.

[0027] Specifically, according to this other example, reducing a cell reactive power for the one or more overmodulated cells comprises operating the MMC in an inductive mode to thereby reduce the first phase arm reactive power.

[0028] This approach may provide a wider range of flexibility to handle converter cells with different loading characteristics, especially in scenarios where an entire converter cell (and DC unit) is bypassed from the phase arm due to fault. The extent of flexibility of this algorithm may depend on the number of converter cells lost per phase arm and the number of healthy MMC converters connected to the PCC.

[0029] According to the aforementioned aspects of the present disclosure, the operation of a (chain-link) MMC converter is enhanced by extending its ability to operate converter cells independently, especially in scenarios where converter cells are lost / bypassed due to fault in DC unit or cell itself or converter cells have DC units with different aging conditions, operating points, and manufacturer types.

[0030] Moreover, the proposed control scheme can ensure that the plant remains in compliance with grid code such as UPF operation, balanced phasecurrents, even while operating the converter cells with dissimilar loading, especially during complete / partial loss of converter cells due to fault.

[0031] Furthermore, the proposed control scheme can operate asymmetric MMC converter having, for example, over 70% of converter cell failures in a phase arm without violating grid-code requirements. As a result, the MMC converter can stay connected and operate remaining healthy converter cells with DC units, which increases the availability of the MMC converter without modifying or introducing additional hardware.

[0032] According to a further aspect of the present disclosure, there is provided an MMC converter, comprising a plurality of phase arms (e.g., three), each phase arm comprising a plurality of series-connected converter cells, and each converter cell having an associated DC unit.

[0033] Viewed from one perspective, and contrasting with conventional MMC control architecture, the MMC disclosed herein has a more granular control. Specifically, the MMC comprises a converter-level controller configured to control the power output of each of the plurality of phase arms, and a plurality of cell-level controllers communicatively coupled to the converter-level controller and configured to control the power output of respective cells.

[0034] The converter-level controller and the plurality of cell-level controllers are configured to perform the control method substantially as described above. For example, the cell-level controllers may be responsible for monitoring respective converter cells for dissimilarities in DC loading, and reporting such dissimilarities to the converter-level controller. The converterlevel controller may then be responsible for distributing the phase arm power among the cells in a phase arm such that the cells are controlled on an individualized basis, e.g., via their cell-level controllers. Put another way, the converter-level controller may be configured to monitor, via the cell-level controllers, the DC units in each phase arm for dissimilarities in loading.

[0035] The DC units may comprise one or more of a capacitor, a battery, a fuel cell, and an electrolyzer, preferably an electrolyzer, and at least one DC unit may be of a different type to at least one other DC unit. For example, in a hydrolyser station context, at least one DC unit is a DC source (e.g., a battery), and at least one other DC unit is a DC load (e.g., a hydrolyser unit).As discussed above, the control method disclosed herein may be especially well suited to managing dissimilarities in DC loading, which may arise due to the DC units themselves, associated with converter cells, being different types.

[0036] The MMC converter may further comprise one or more additional converter cells without associated DC units. Converter cells without DC units can absorb / supply reactive power, thereby altering the loading factor of the converter cells with DC units in the same phase arm.

[0037] Additionally or alternatively, the converter may further comprise converter cells with bi-directional DC units. Converter cells with bi-directional DC units can enable the synthesis and injection of an AC voltage vector of desired phase angle, thereby facilitating an easy control over the loading factor and phase arm voltage without having to alter the reactive power as performed by the proposed control scheme. Such additional hardware installment can be particularly beneficial for chain-link MMC units with unidirectional DC units such as electrolyzers.

[0038] In a further aspect of the present disclosure, a plurality of MMC converters, substantially as described above, are provided in a system such as a power plant. In such a system, a system-level controller is configured to control a power output of each MMC converter.

[0039] Hence, three levels of control may be defined: plant-level control, converter-level control, and cell-level control. The plant-level control may determine the apparent (i.e. , active, and reactive) phase power of the plant by considering the DC units’ operating conditions and / or requirements at grid. The plant-level control may also divide the pre-determined apparent phase power among the plurality of MMC converters based on, for example, their respective status and converter cell availability.

[0040] During normal operation, i.e., without dissimilarities in converter cell loading, the converter-level control may distribute a preset apparent phase power among its converter cells. If dissimilarities in converter cell loading are detected, the control scheme performs necessary modifications to apparent phase power before distributing it among the converter cells.

[0041] Finally, the cell-level control translates the apparent power of a converter cell to, for example, a modulation index, which may be further converted to switching signals using a modulator. These switching signals drive the converter cells in the unit such that its respective DC units operate effectively without distorting the phase arm currents or violating the grid code requirements, as discussed above.Brief Description of the Drawings

[0042] 0ne or more embodiments will be described, by way of example only, and with reference to the following figures, in which:

[0043] Figure 1 schematically shows an example power plant having a plurality of MMC converters;

[0044] Figure 2 illustrates a method for controlling a grid-connected MMC, according to aspects of the present disclosure;

[0045] Figures 3A and 3B schematically illustrate a first example implementation of the present disclosure;

[0046] Figure 4 illustrates the method exemplified in figures 3A and 3B;

[0047] Figures 5A, 5B, and 5C schematically illustrate a second example implementation of the present disclosure;

[0048] Figure 6 illustrates the method exemplified in figures 5A to 5C;

[0049] Figures 7A and 7B schematically illustrate a third example implementation of the present disclosure;

[0050] Figure 8 illustrates the method exemplified in figures 7A and 7B; and

[0051] Figure 9 illustrates, in flow chart form, an example control scheme for an MMC combining multiple example implementations of the present disclosure.Detailed Description

[0052] The present disclosure is described in the following by way of a number of illustrative examples. It will be appreciated that these examples are provided for illustration and explanation only and are not intended to be limiting on the scope of the disclosure.

[0053] Furthermore, although the examples may be presented in the form of individual embodiments, it will be recognized that the present disclosure also covers combinations of the embodiments described herein.

[0054] Figure 1 schematically shows a power plant system 100 (also referred to as simply ‘plant 100’). The plant 100 comprises a plurality of modular multilevel converters (MMCs) 102-1 , 102-2, ... , 102-N. Although three MMCs 102-1 , 102-2, 102-N are illustrated, it will be appreciated that, in alternative examples, the plant 100 may comprise only one MMC, two MMCs, or more than three MMCs, depending on the requirements for the plant 100.

[0055] Each MMC 102-1 , 102-2, ... , 102-N comprises three phase arms 104a, 104b, 104c, with a reference numeral suffix corresponding to the MMC the phase arms are comprised in. Each phase arm 104a, 104b, 104c is connected to a respective phase line 114a, 114b, 114c by a respective transformer 110 and reactor 112. The phase lines 114a, 114b, 114c in turn connect to a point of common coupling (PCC) 116 which acts as an interconnect between the plant 100 and a power grid (not shown).

[0056] The chain-link MMC 102-1 is formed by connecting input AC terminals of multiple converter cells 106 in series. Each of these converter cells 106 can be realized either by a full-bridge (FB) AC / DC converter cell configuration or by a combination of a FB cell cascaded to an isolated or non-isolated DC / DC converter configuration. The output DC terminals of these converter cells are integrated to either one DC unit 108 or, in other examples, multiple DC units connected in series and / or parallel.

[0057] The converter cell 106 configuration with a DC / DC converter introduces additional flexibility to FB cell operation by making its modulation index independent of the DC unit’s 108 operating point. A large-scale plant 100 with multiple DC units 108 can be formed by connecting the plurality of chain-link MMCs 102-1 , 102-2, ... , 102-N in parallel to the PCC 116 through the transformers 110.

[0058] Each phase arm 104a-1 , 104a-2, 104a-3 comprises a plurality of series-connected converter cells 106, which may also be referred to as simply ‘cells 106’, where each cell 106 has an associated DC unit 108. The DC units 108 may comprise one or more of a capacitor, a battery, a fuel cell, and anelectrolyzer, preferably an electrolyzer. At least one DC unit 108 may be of a different type to at least one other DC unit 108. For example, some of the DC units 108 may be electrolyzers and some of the DC units 108 may be capacitors.

[0059] It will be appreciated that, although the MMCs 102-1 , 102-2, ... , 102-N are shown as being similarly configured, in other examples, each MMC may have a different configuration.

[0060] Although not illustrated, the plant 100 may comprise three levels of controllers: a plant-level controller, converter-level controllers, and cell-level controllers.

[0061] A method 100 for controlling an MMC, such as the MMC 102-1 shown in figure 1 , is illustrated in figure 2, and may be carried out in a computer- implemented manner by a converter-level controller. Viewed from a general perspective, the method 200 comprises determining 210 a dissimilarity in loading between DC units in a first phase arm.

[0062] In response to determining the dissimilarity, the method 200 further comprises determining 220 a first phase arm power to be delivered by the first phase arm, determining 230 a cell power to be delivered each cell in the first phase arm by distributing the first phase arm power among the cells in the first phase arm, and controlling 240 the cells in the first phase arm to deliver their respective determined cell power.

[0063] The step of determining 230 a cell power and controlling 240 the cells may be carried out by receiving and sending messages to and from cell-level controllers, respectively. The particular configuration and interconnection of the controllers is not discussed in detail herein but can be readily realized by those skilled in the art to achieve the functionality described herein.

[0064] If, for example, the method 200 is carried out in the plant 100 shown in figure 1 , the method 200 may advantageously enhance the operation of MMCs 102-1 , 102-2, ... , 102-N by extending their ability to operate their converter cells 106 independently in scenarios where converter cells 106 are lost / bypassed due to, for example, a fault in their DC unit 108 or the cell 106 itself, or where converter cells 106 have DC units 108 with different aging conditions, operating points, or manufacturer types.

[0065] The method 200 may further ensure the plant 100 remains in compliance with grid code such as UPF operation at the PCC 116 and balanced currents between phase lines 114a, 114b, 114c, even while operating the converter cells 106 with dissimilar loading, especially during complete / partial loss of converter cells 106 due to fault.

[0066] The method 200 also enables operation of an ‘asymmetric’ MMC, e.g., in a case where the MMC 102-1 has over 70% of converter cell 106 failures in its phase arm 104a-1 , without violating grid-code requirements. As a result, the MMC 102-1 can stay connected to the PCC 116 and operate remaining healthy converter cells 106 with DC units 108, which increases the MMC’s 102-1 availability without introducing additional hardware.

[0067] Three example implementations of the method 200 are described below. A first example implementation is discussed in connection with figures 3A-B and 4, a second example implementation is discussed in connection with figures 5A-C and 6, and a third example implementation is discussed in connection with figures 7A-B and 8. An overall control scheme, e.g., for a hydrolyser plant, is presented in figure 9, combining each of these example implementations together.

[0068] Figures 3A and 3B illustrate a first example implementation of the presently disclosed method. Figure 3A shows an MMC having three phase arms 304a, 304b, 304c at the point of a fault occurring in two converter cells 306 in a first phase arm 304a, where the fault also affects the cells’ 306 associated DC units 308. Figure 3B shows the same MMC, after the fault has caused the converter cells 306 and their DC units 308 to be bypassed, as indicated by the shading of these cells 306 and their corresponding DC units 308. The numbers shown in cells 306 correspond to the modulation index, m, adopted by the cell 306.

[0069] Figure 4 illustrates a method 400 being undertaken by a converterlevel controller in response to the failure shown in figure 3A, to thereby transition the MMC to the state shown in figure 3B.

[0070] First, in step 410, the loading dissimilarity is detected, e.g., by receiving an indication from one or more cell-level controllers that monitor the DCvoltage and / or current of the DC units 308. In this case, the dissimilarity has been caused by the failure.

[0071] In response to determining the dissimilarity in the first phase arm 304a, in step 420, the controller identifies the minimum phase arm active power among the phase arms 304a, 304b, 304c. It will be understood that the minimum phase arm active power may be that delivered by the first phase arm 304a, as a result of the failure therein. The minimum phase arm active power may thus correspond to the maximum deliverable active power for the first phase arm 304a, given the number of remaining healthy cells 306 and DC units 308.

[0072] In determining the maximum deliverable active power for the first phase arm 304a, in this example, a maximum modulation index of 1.0 is applied as a constraint.

[0073] The controller then, in step 430, set the first phase arm active power to said maximum deliverable active power. This step 430 may comprise controlling each of the remaining converter cells 306 to adopt a modulation index of 1 .0, e.g., via instructing the individual cell-level controllers of each remaining healthy cell 306 in the first phase arm 304a.

[0074] In step 440, the converter-level controller sets the second phase arm active power and the third phase arm active power as equal to the first phase arm active power. Thus, it is seen that the method 400 results in varying the active power of each available converter cell 306 individually to achieve a balanced three-phase active power. In doing so, the currents at PCC are balanced and operate at UPF.

[0075] In this example, it can be seen that the maximum deliverable active power for the first phase arm 304a after the fault is less than that before the fault. Hence, the cells 306 in the second and third phase arms 304b, 304c are controlled to reduce their modulation index to 0.72 from 0.8 so as to preserve the balance between the phase arms.

[0076] To give an example, the power in the first phase arm 304a is dropped from 41 .12 MW to 29.37 MW, so the converter cell 306 loadings in the second phase arm 304b and 304c are reduced by 28.6% to achieve balanced phase powers. Even though the three phase arms 304a, 304b, 304c have samevoltage, the modulation index of cells 306 in the first phase arm 304a are different from the second and third phase arms 304b, 304c due to the difference in number of converter cells 306.

[0077] Another possible approach to equalize the phase power may be to take away an equal number of DC units 308 from the healthy phases (i.e. , the second and third phase arms 304b, 304c) by setting the modulation index of respective converter cells 306 to zero. In doing so, the three phase arms 304a, 304b, 304c can operate with an equal number of converter cells 306. However, such approach may be most appropriately implemented when the power difference between the phase arms 304a, 304b, 304c is an integer multiple of the power absorbed / supplied by a DC unit 308, which may be the case when converter cells 306 and DC units 308 are bypassed / lost from a phase arm 304a, as shown in figure 3B.

[0078] For instance, in a case where converter cells 306 have associated DC units 308 of different characteristics such as dissimilar aging conditions, it may not always be possible to remove a specific number of converter cells 306 from other phase arms 304b, 304c to equalize the phase power. Also, if the DC units 308 are electrolyzers, shutting them down to equalize phase power may not be a beneficial choice, as restarting the electrolyzer back when needed can cause system delays due to its Balance-Of-Plant. Thus, it is preferred that the phase power equalization method 400 reduces an equal amount of power from all converter cells 306 in a phase arm 304a, 304b, 304c.

[0079] During equalization of phase arm active powers, as described by the method 400 above, or during any other stage of operation of an MMC, if the modulation index of a cell 306 exceeds 1 (m > 1), one or more further methods, which may be referred to as ‘modulation index stabilization’ may be enacted. The second and third example implementations of the presently disclosed control method can be considered as different types of modulation index stabilizers.

[0080] Modulation index stabilization may advantageously avoid overmodulation of cells and limit the modulation index of converter cells to 1 during dissimilarities in their loadings. Since all converter cells have samephase arm current, the input AC voltage (RMS) for the nthconverter cell in the xthphase arm of the mthMMC unit in a plant can be expressed as

[0081] where s™xis apparent power of nthconverter cell in an xthphase arm, s™ is total apparent power of xthphase arm, and v™ is xthphase arm voltage in RMS value. Thus, the modulation index of converter cell can be obtained as

[0082] where vdcis the DC bus voltage of an AC / DC FB cell in a converter cell such as that as shown in figures 1 or 3A-B. From the above equations, it is evident that phase arm voltage v™ and converter cell’s apparent power s™xaffects its modulation index.

[0083] Figures 5A, 5B, and 5C schematically show a progression of modulation index stabilization in a phase arm 504a, according to a second example implementation of the presently disclosed method. Figure 6 illustrates a method 600 being undertaken by a converter-level controller in response to the failure of three DC units 508 shown in figure 5A. Figure 5B shows an initial state after the fault, and figure 5C shows a resultant state after the state shown in figure 5B, as a result of the execution of the method 600.

[0084] In this technique, the apparent power of each converter cell 506 is controlled by setting the total apparent power for the phase arm 504a constant. In doing so, the loading factor of individual converter cells 506 can be adjusted such that their modulation index m is brought within the desirable limit. The effect of the loading factor on the modulation index is demonstrated for three different cases.

[0085] Figure 5A shows normal operating condition, where seven converter cells 506 in phase arm 504a operate at an equal loading with a modulation index of 0.8.

[0086] Figure 5B demonstrates a scenario where DC units 508 of three converter cells 508 are unavailable due to fault (shown as ‘FAIL’ in figure 5A).Under such a scenario, these three converter cells 506b do not impart active power but only provide reactive power. As a result, the loading factor of the four healthy converter cells 506a increases and the modulation index for the converter cells 506b with faulty DC units 508 decreases.

[0087] Due to the increase in loading factor, the modulation index for four healthy converter cells 506a exceeds 1 , as shown in figure 5B, increasing to 1.12. Figure 5C demonstrates the result of the method 600 where the apparent power of a converter cells 506 in the phase arm 504a is adjusted by varying their respective reactive power contribution.

[0088] The three converter cells 506b with faulty DC units 508 share the total reactive power for the phase arm 504a while remaining four healthy converter cells 504a impart only active power for their DC units 508. In doing so, the loading factor of the four healthy converter cells 506a decreases, which subsequently helps in reducing and limiting their modulation index to 1.0, as shown in figure 5C, without affecting the DC unit’s 508 operation or grid code requirements.

[0089] Put another way, the method 600 comprises, at step 610, determining a dissimilarity in DC loading, resulting from the fault in the DC units 508 shown in figure 5A.

[0090] The method 600 further comprises, at step 620, determining a first phase arm power to be delivered by the first phase arm 504a, involving the setting of the total apparent power to be delivered by the first phase arm 504a as a constant.

[0091] Step 630 then comprises determining that one or more cells 506a in the first phase arm 504a are overmodulated. Hence, determining a cell power to be delivered each cell 506 in the first phase arm 504a may involve unevenly distributing the first phase arm power among the cells 506 in the first phase arm 504a, at least in terms of their relative contributions of active and reactive power.

[0092] Step 640 then comprises reducing a cell reactive power for the one or more overmodulated cells 506a to thereby reduce the apparent power of the one or more overmodulated cells 506a, thereby controlling the cells 506 in the first phase arm 504a to deliver their respective determined cell power.

[0093] As discussed above, in step 640, reducing the cell reactive power for the one or more overmodulated cells 506a comprises determining a total first phase arm reactive power, and redistributing the cell reactive power from the one or more overmodulated cells 506a to other cells 506b in the first phase arm 504a (i.e. , those having failed DC units) to thereby maintain the first phase arm reactive power.

[0094] In an example, the MMC may further comprise converter cells without DC units in the first phase arm 504a, such that the reactive power contribution from these cells can be freely modified from an absorption or supply of reactive power, thereby achieving similar results as those described above.

[0095] Figures 7A and 7B schematically show a third example implementation of the presently disclosed control scheme. Figure 8 illustrates a method 800 being undertaken by a converter-level controller in response to the failure shown in figure 7A, to thereby transition the MMC to the state shown in figure 7B.

[0096] The method 800 shown in figures 7A and 7B is another type of modulation stabilization, which may be combined with the previously described method 600. However, according to this method 800, the phase arm voltage of the MMC is reduced by absorbing reactive power, which can then be compensated for by other MMCs in the plant. In doing so, the modulation index of converter cells 706 can be limited without affecting the DC unit’s 708 operation and grid code requirements.

[0097] The effectiveness of algorithm in limiting the mnxto 1 during the failure of three converter cells is demonstrated in figures 7A and 7B. Figure 7A shows normal operating condition, where all converter cells 706 of the three phase arms 704a, 704b, 704c of the MMC operate at an equal loading with a modulation index of 0.8, thereby achieving UPF at PCC. A failure of three converter cells 706 and their DC units 708 is indicated by the dotted box ‘FAIL’.

[0098] Figure 7B demonstrates a scenario where the three converter cells 706 of the first phase arm 704a are completely bypassed due to the fault. To limit the modulation index of remaining converter cells 706 in the first phase arm704a to 1 , the control method 800 reduces the phase arm voltage of the MMC.

[0099] Put another way, the method 800 comprises, at step 810, determining a dissimilarity in DC loading, resulting from the fault in the cells 706 shown in figure 7A.

[0100] The method 800 further comprises, at step 820, determining a first phase arm power to be delivered by the first phase arm 704a, which may result in a determination that one or more cells will be overmodulated.

[0101] Therefore, determining 820 the first phase arm power may comprise operating the MMC in an inductive mode to thereby reduce the first phase arm reactive power.

[0102] At step 830, determining a cell power to be delivered each cell 706 in the first phase arm 704a may involve distributing the revised first phase arm power among the cells 706 in the first phase arm 704a.

[0103] Step 840 then comprises effectively reducing a cell reactive power for the one or more overmodulated cells 706a to thereby reduce the apparent power of the one or more overmodulated cells 706a. In this way, it can be seen that the remaining cells in the first phase arm are limited to 1 .0 in figure 7B.

[0104] The cells 706 in the second and third phase arms 704b, 704c also have reduced modulation indexes - reduced to 0.57 - as a result of the balancing of phase powers as discussed in connection figures 3A-B and 4.

[0105] To give an example of the above, the MMC may be operated in an inducive mode, in a manner understood by those skilled in the art, by absorbing 10.32 MVAR from the PCC. As a result of the MMC operating in the inductive mode, the power factor at PCC drops below 1 .

[0106] A plant-level controller may identify this drop and triggers other MMCs in the plant to correct the power factor by providing the desired amount of reactive power to the PCC. As a result, these other MMCs in the plant may be operated in a capacitive mode by supplying, to follow the above example, 3 x 3.44 = 10.32 MVAR to the PCC. Since each other MMC supplies a fraction of the total reactive power to PCC for compensation, the increase in the modulation index of converter cells in these other MMCs is minimal.

[0107] The reactive power required by the MMC shown in figures 7A-B for reducing its phase arm voltage can also be provided by other available compensation devices in the plant such as shunt capacitors and / or STATCOM etc. Such an approach introduces the possibility to reduce the reactive power contribution from other MMCs in the plant, thereby increasing the modulation index margin of its converter cells. Therefore, the technique of reducing phase arm voltage effectively limits the converter cells modulation index to 1 without affecting the DC unit’s 708 operation or grid code requirements.

[0108] Figure 9 shows, in flow-chart form, an example control scheme for a plant such as the plant 100 shown in figure 1 , containing a plurality of MMC units 102-1 , 102-2, ... , 102-N.

[0109] It can be seen in figure 9 that, at a plant level, a plant-level controller sets the plant’s apparent phase power based on grid and / or DC unit requirements, and distributes the preset plant’s apparent phase power among units. As a default, this may be an even distribution.

[0110] For each of M MMC units, a converter-level controller receives the requested apparent power from the plant-level controller. If no dissimilarities are detected, a preset distribution (e.g., an even distribution) may be used to distribute the MMC’s apparent phase power among converter cells.

[0111] The converter-level controller may then communicate with cell-level controllers to compute a modulation index based on the requested apparent power and control, via a modulator and based on the computed modulation index, the converter cells to output switching signals, thereby delivering the requested apparent power.

[0112] In a case where a dissimilarity is detected, a ‘phase power equalizer’ is implemented, which may substantially correspond to the method 400 illustrated in connection with figures 3A-B and figure 4, where a minimum real / active phase power is identified and set as the active power for each of the three phases of the MMC.

[0113] If no overmodulation is detected, the method then involves the updating of the MMC unit’s apparent phase power, based on the equalization, which is reported back to the plant-level controller via a flag F1.

[0114] However, if an overmodulation is detected (mnx> 1), then modulation index stabilization is employed. ‘Algorithm T shown in figure 9 may substantially correspond to the method 600 discussed in connection with figures 5A-C and 6, and ‘Algorithm 2’ may substantially correspond to the method 800 discussed in connection with figures 7A-C and 8. As shown in figure 9 by the reversible half-arrows, these two algorithms may be implemented simultaneously. The extent to which each algorithm is employed, as discussed above, may depend on the amount of reactive power required for each phase.

[0115] After modulation index stabilization has been applied, the method may continue again to update the MMC unit’s apparent phase power, and subsequently onto the distribution of the apparent phase power among cells and the control of the cells to deliver said power.

[0116] As explained herein, such a control scheme, based on an individualized control of cells according to a distribution of phase power (active or reactive, evenly or unevenly), enables substantial tolerance of DC loading dissimilarity without requiring the disconnection of an MMC from the grid. Hence, the overall availability of a plant may be greatly enhanced. In a context where the plant is a hydrolyser plant, it will be appreciated that the hydrogen production from such a plant can be greatly enhanced.

[0117] While the present disclosure is susceptible to various modifications and alternative forms, specific embodiments are shown and described above by way of example in relation to the drawings, with a view to clearly explaining the various advantageous aspects of the present disclosure. It should be understood, however, that the detailed description herein and the drawings attached hereto are not intended to limit the disclosure to the particular form disclosed. Rather, the intention is to cover all modifications, equivalents, and alternatives falling within the scope of the following claims.

Claims

C L A I M S1 . A method for controlling a grid-connected modular multilevel converter, MMC, wherein the MMC comprises a plurality of phase arms, and each phase arm comprises a plurality of series-connected converter cells each having respective DC units, the method comprising: determining a dissimilarity in loading between DC units in a first phase arm; in response to determining the dissimilarity, determining a first phase arm power to be delivered by the first phase arm; determining a cell power to be delivered each cell in the first phase arm by distributing the first phase arm power among the cells in the first phase arm; and controlling the cells in the first phase arm to deliver their respective determined cell power.

2. The method according to claim 1 , wherein controlling a cell to deliver a determined cell power comprises adjusting the modulation index of the cell.

3. The method according to claim 1 or claim 2, wherein the cells in the first phase arm are controlled based on information associated with one or more DC units in the first phase arm.

4. The method according to any preceding claim, wherein determining a dissimilarity in loading comprises determining a failure of one or more DC units through DC voltage and / or current measurements.

5. The method according to any preceding claim, wherein determining the first phase arm power to be delivered by the first phase arm comprises determining a maximum deliverable active power for the first phase arm and setting a first phase arm active power to said maximum deliverable active power.

6. The method according to claim 5, further comprising determining a phase arm power for each of the other phase arms by equating their respective phase arm active powers to the first phase arm active power.

7. The method according to any preceding claim, further comprising determining that one or more cells in the first phase arm are overmodulated, and reducing a cell active power or a cell reactive power for the one or more overmodulated cells to thereby reduce the apparent power of the one or more overmodulated cells.

8. The method according to claim 7, wherein reducing a cell reactive power for the one or more overmodulated cells comprises determining a first phase arm reactive power, and redistributing cell reactive power from the one or more overmodulated cells to other cells in the first phase arm to thereby maintain the first phase arm reactive power.

9. The method according to claim 7 or claim 8, wherein reducing a cell reactive power for the one or more overmodulated cells comprises operating the MMC in an inductive mode to thereby reduce the first phase arm reactive power.

10. The method according to any of claims 7 to 9, wherein determining that one or more cells are overmodulated comprises determining that one or more cells have a modulation index exceeding one, and wherein the cell reactive power for the overmodulated cells is reduced to thereby return the modulation index to one or less.

11. An MMC converter, comprising: a plurality of phase arms, each phase arm comprising a plurality of series-connected converter cells, each converter cell having an associated DC unit;a converter-level controller configured to control the power output of each of the plurality of phase arms; and a plurality of cell-level controllers communicatively coupled to the converter-level controller and configured to control the power output of respective cells, wherein the converter-level controller and the plurality of cell-level controllers are configured to perform the method according to any preceding claim.

12. The converter according to claim 11 , wherein the converter-level controller is configured to monitor, via the cell-level controllers, the DC units in each phase arm for dissimilarities in loading.

13. The converter according to claim 11 or claim 12, wherein the DC units comprise one or more of a capacitor, a battery, a fuel cell, and an electrolyzer, preferably an electrolyzer.

14. The converter according to any of claims 11 to 13, wherein at least one DC unit is of a different type to at least one other DC unit.

15. The converter according to claim 14, wherein at least one DC unit is a DC source, and at least one other DC unit is a DC load.

16. The converter according to any of claims 11 to 15, further comprising one or more additional converter cells without associated DC units, or with bidirectional DC units.

17. A system comprising a plurality of MMC converters according to any of claims 11 to 16, and a system-level controller configured to control a power output of each MMC converter.

Images