WO2021228958A1

WO2021228958A1 - Method, system and device for power balancing in a power grid

Info

Publication number: WO2021228958A1
Application number: PCT/EP2021/062654
Authority: WO
Inventors: Edouard PERROY
Original assignee: Energy Pool Developpement
Priority date: 2020-05-13
Filing date: 2021-05-12
Publication date: 2021-11-18

Abstract

The present disclosure relates to a method of balancing power supply and demand in a power supply network comprising: modifying, by a first device an activation level of a first discrete load from a first level to a second level based on a first signal indicating an amount of imbalance in the power supply network and on a set of activation thresholds; generating, a first estimation of a change in a power level consumed by the first discrete load resulting from the change to the activation level; calculating a residue power level to be consumed by one or more continuous loads based on the first estimation and on a signal indicating the amount of imbalance in the power supply network; and generating a control signal to a first continuous load of the one or more continuous loads based on the calculated residue power level.

Description

DESCRIPTION

Method , system and device for power balancing in a power grid

The present patent application claims priority from the French patent application filed on 13 May 2020 and assigned application no. FR2004712, and from the Japanese patent application filed on 13 May 2020 and assigned application no. JP2020-084703, the contents of these applications being hereby incorporated by reference.

Technical field

[0001] The present disclosure relates generally to the field of electronical power management devices and systems, and in particular to a method, system and device for a demand side balancing of power supplied over a power grid.

Background art

[0002] A challenge that is recurrently faced by electrical power supply grids is to balance, at all times, the supply and demand in order to avoid overloading the transmission grid, which could lead to blackouts.

[0003] The system frequency is the foremost indicator of an instantaneous power imbalance on the grid. Indeed, an increase in consumption causes an increased power demand on synchronous production machines and thus causes a slowing of their speed of rotation. Inversely, a production surplus, and thus a frequency rise, will result from a reduced instantaneous power need.

[0004] In Europe, it is the role of the Transmission System Operator (TSO) to ensure the stability of the grid, by organizing and supervising actions and mechanisms related to frequency control. Other countries have entities that perform a similar role. In relatively large transmission grids, there are generally three layers of frequency control used to cope with frequency deviations: primary, secondary and tertiary. The Primary Frequency Control (PFC), also known as the Frequency Containment Reserve (FCR), is predominantly provided by the droop capability of large generators. However, the advent of Demand Response (DR) in the development of today's smart grids has made the demand side a new actor in the stability of an electrical power grid. Indeed, industrial sites are often willing to accept a certain flexibility in their power consumption in exchange for financial incentives such as a reduced energy cost.

[0005] There is, however, a technical difficulty in providing useful demand side power balancing functions, particularly in the case of relatively small industrial sites consuming less than 100 MW. There is thus a need in the art for a method, system and device allowing an industrial site to contribute usefully in a power balancing mechanism.

Summary of Invention

[0006] It is an aim of embodiments of the present disclosure to at least partially address one or more needs in the art.

[0007] According to one aspect, there is provided a method of balancing power supply and demand in a power supply network comprising: selecting, by a first device at a first client site, a first set of activation thresholds from among a plurality of sets of activation thresholds; modifying, by the first device, an activation level of a first discrete load from a first level to a second level based on a first signal indicating an amount of imbalance in the power supply network and based on the first set of activation thresholds; generating, by the first device, a first estimation of a change in a power level consumed by the first discrete load resulting from the change to the activation level; calculating a residue power level to be consumed by one or more continuous loads based on the first estimation and on the first signal or a second signal indicating the amount of imbalance in the power supply network; and generating a control signal to a first continuous load of the one or more continuous loads based on the calculated residue power level.

[0008] According to one embodiment, the first estimation indicates a regulation fault in the first discrete load, the method further comprising selecting, based on the first estimation, a second set of activation thresholds from among the plurality of sets of activation thresholds.

[0009] According to one embodiment, the first signal is a frequency signal indicating a system frequency of a supply voltage on the power supply network.

[0010] According to one embodiment, the first signal is a power modification command signal indicating a requested change in power consumption.

[0011] According to one embodiment, a deactivation of each activation level is applied with hysteresis, each activation threshold corresponding to a latch threshold at which the corresponding activation level is triggered, and being associated with a further release threshold at which the corresponding activation level is no longer applied.

[0012] According to one embodiment, the plurality of sets of activation thresholds are stored in a memory of the first device .

[0013] According to one embodiment, the first discrete load and the first continuous load are located at the first client site, and the calculation of the residue power level, and the generation of the control signal, are performed by the first device at the first client site.

[0014] According to one embodiment, the first discrete load is located at the first client site, and the first continuous load is located at a second client site, and the generation of the control signal is performed by a second device at the second client site.

[0015] According to one embodiment, calculating a residue power level to be consumed by one or more continuous loads is at least partially performed by a central power management system.

[0016] According to a further aspect, there is provided a device for balancing power supply and demand in a power supply network, the device comprising: a memory storing a plurality of sets of activation thresholds; and one or more circuits, and/or one or more processors under control of instructions stored in an instruction memory, configured to: select a first set of activation thresholds from among the plurality of sets of activation thresholds; modify an activation level of a first discrete load from a first level to a second level based on a first signal indicating an amount of imbalance in the power supply network and based on the first set of activation thresholds; generate a first estimation of a change in a power level consumed by at least the first discrete load resulting from the change to the activation level; and transmit the first estimation to a regulation system of a first continuous load.

[0017] According to one embodiment, the one or more circuits, and/or the one or more processors, are further configured to select, based on the first estimation, a second set of activation thresholds from among the plurality of sets of activation thresholds, wherein the first estimation indicates a regulation fault in the first discrete load.

[0018] According to one embodiment, the first signal is a frequency signal indicating a system frequency of a supply voltage on the power supply network. [0019] According to one embodiment, the first signal is a power modification command signal indicating a requested change in power consumption.

[0020] According to one embodiment, a deactivation of each activation level is applied with hysteresis, each activation threshold corresponding to a latch threshold at which the corresponding activation level is triggered, and being associated with a further release threshold at which the corresponding activation level is no longer applied.

[0021] According to one embodiment, the first discrete load and the first continuous load are located at a first client site, and the one or more circuits, and/or the one or more processors, are further configured to implement a regulation loop for controlling the first continuous load, by: calculating a residue power level to be consumed by at least one continuous load based on the first estimation and on the first signal; and generating a control signal to the first continuous load of the at least one continuous load based on the calculated residue power level.

[0022] According to a further aspect, there is provided a system for balancing power supply and demand in a power supply network, the system comprising: the above device, and the first discrete load, located at a first client site; and a second device, and the first continuous load, located at a second client site, the second device implementing a regulation loop for controlling the first continuous load, the second device being configured to: calculate a residue power level to be consumed by one or more continuous loads based on the first signal, or on a further signal indicating the amount of imbalance in the power supply network, and based on the first estimation of a change in a power level consumed by at least the first discrete load; and generate a control signal to the first continuous load of the one or more continuous loads based on the calculated residue power level.

[0023] According to one embodiment, the residue power level is further generated based on a feedback signal, the system further comprising a central power management system configured to generate the feedback signal.

[0024] According to a further aspect, there is provided A method of balancing power supply and demand in a power supply network comprising: modifying, by a first device at a first client site, an activation level of a first discrete load from a first level to a second level based on a first signal indicating an amount of imbalance in the power supply network and based on a first set of activation thresholds; generating, by the first device, a first estimation of a change in a power level consumed by the first discrete load resulting from the change to the activation level; calculating a residue power level to be consumed by one or more continuous loads based on the first estimation and on the first signal or a second signal indicating the amount of imbalance in the power supply network; and generating a control signal to a first continuous load of the one or more continuous loads based on the calculated residue power level.

[0025] According to one embodiment, the method further comprises, prior to modifying the activation level of the first discrete load, selecting, based on at least an availability of at least one further discrete load, the first set of activation thresholds from among a plurality of sets of activation thresholds.

[0026] According to a further aspect, there is provided a device for balancing power supply and demand in a power supply network, the device comprising one or more circuits, and/or one or more processors under control of instructions stored in an instruction memory, configured to: modify an activation level of a first discrete load from a first level to a second level based on a first signal indicating an amount of imbalance in the power supply network and based on a first set of activation thresholds; generate a first estimation of a change in a power level consumed by at least the first discrete load resulting from the change to the activation level; and transmit the first estimation to a regulation system of a first continuous load.

[0027] According to one embodiment, the device further comprises a memory storing a plurality of sets of activation thresholds, wherein the one or more circuits, and/or the one or more processors, are further configured to select, prior to modifying the activation level of the first discrete load, and based on at least an availability of at least one further discrete load, the first set of activation thresholds from among the plurality of sets of activation thresholds.

Brief description of drawings

[0028] The foregoing features and advantages, as well as others, will be described in detail in the following description of specific embodiments given by way of illustration and not limitation with reference to the accompanying drawings, in which:

[0029] Figure 1 is a block diagram illustrating a power supply system according to an example embodiment of the present disclosure;

[0030] Figure 2 is a block diagram illustrating a power supply system with sites having electrolysis equipment and furnaces according to an example embodiment of the present disclosure;

[0031] Figure 3 schematically illustrates an electrolysis system according to an example embodiment of the present disclosure; [0032] Figure 4 schematically illustrates a regulation system for continuous and discrete loads according to an example embodiment of the present disclosure;

[0033] Figure 5 is a graph representing an activation level of a discrete load as a function of frequency;

[0034] Figure 6 is a graph representing frequency variations over time, and corresponding system responses;

[0035] Figure 7 is a graph representing an example of frequency thresholds and response profiles according to an example embodiment of the present disclosure;

[0036] Figure 8 is a graph representing a simulated number of movements of each level of a discrete load as a function of the activation frequency and as a function of the persistence;

[0037] Figure 9 is a block diagram illustrating a computing device according to an example embodiment of the present disclosure;

[0038] Figure 10 schematically illustrates a regulation system for continuous and discrete loads according to a further example embodiment of the present disclosure; and

[0039] Figure 11 schematically illustrates a regulation device for continuous and discrete loads according to yet a further example embodiment of the present disclosure.

Description of embodiments

[0040] Like features have been designated by like references in the various figures. In particular, the structural and/or functional features that are common among the various embodiments may have the same references and may dispose identical structural, dimensional and material properties.

[0041] Unless indicated otherwise, when reference is made to two elements connected together, this signifies a direct connection without any intermediate elements other than conductors, and when reference is made to two elements coupled together, this signifies that these two elements can be connected or they can be coupled via one or more other elements .

[0042] In the following disclosure, unless indicated otherwise, when reference is made to absolute positional qualifiers, such as the terms "front", "back", "top", "bottom", "left", "right", etc., or to relative positional qualifiers, such as the terms "above", "below", "higher", "lower", etc., or to qualifiers of orientation, such as "horizontal", "vertical", etc., reference is made to the orientation shown in the figures.

[0043] Unless specified otherwise, the expressions "around", "approximately", "substantially" and "in the order of" signify within 10 %, and preferably within 5 %.

[0044] Figure 1 is a block diagram illustrating a power supply system 100 according to an example embodiment of the present disclosure.

[0045] In the example of Figure 1, the power supply system 100 comprises three client sites 102, 103 and 104, a central power management system 106, a markets server (MARKETS) 107 and a power grid operator server (POWER GRID OPERATOR) 108.

[0046] A client site in the power supply system 100 corresponds to a site that comprises one or more electrical loads. For example, one or more power supply contracts are in place involving the client sites 102, 103, 104 and operators of the power grid, setting a commercial relationship between the entities. While not illustrated in Figure 1, a client site may additionally comprise electricity storage and/or electricity generators. While three client sites 102, 103,

104 are illustrated in the embodiment of Figure 1, in alternative embodiments there may be any number of client sites managed by the central power management system 106.

[0047] Each of the client sites 102, 103, 104 comprises one or more continuous loads (C LOADS) 110, and/or one or more discrete loads (D LOADS) 112. In the example of Figure 1, the client site 102 comprises one or more continuous loads 110, the client site 103 comprises one or more discrete loads 112, and the client site 104 comprises both continuous loads 110, and discrete loads 112.

[0048] The term "discrete load" is used herein to designate a load that is either a bi-state load, in other words a load having just two operating states corresponding to active and inactive, or a load that is capable of operating in any of a plurality nd of states having different levels of power consumption from each other. For example, nd is an integer equal to between 2 and 50.

[0049] Examples of discrete loads include industrial ovens or furnaces or heating systems, or more generally thermostatic loads, ventilators with discrete power settings, etc. In some embodiments, the discrete loads comprise On-Load Tap Changers (OLTC). As known by those skilled in the art, an OLTC is a high voltage transformer that provides discrete levels of output voltage, and thus output power, by allowing a turn ratio of the transformer to be dynamically adjusted during operation. The one or more discrete loads 112 for example have a combined power consumption in the range 100 kW to 400 MW or more.

[0050] The term "continuous load" is used herein to designate a load that is controlled by an analog control signal generated by an analog regulation loop. A continuous load could alternatively be controlled by a digital control signal and/or using a digital regulation loop if the granularity is such that this control can be considered equivalent to that of an analog control. This is for example the case when the digital control signal, or digital regulation loop, is based on at least 12 bits, and preferably at least 16 bits.

[0051] Examples of continuous loads include electrolysis systems, variable speed motors, such as ventilators with variable power settings, etc. The one or more continuous loads 110 for example have a combined power consumption in the range 100 kW to 400 MW or more.

[0052] Each client site 102, 103, 104 also for example comprises an onsite monitoring and control interface (DR BOX) 114, which will be referred to hereafter as a demand response (DR) box, or simply DR box. The DR box 114 is for example a programmable logic controller (PLC), and is also for example configured to implement energy management, similar to the role of an energy management system (EMS).

[0053] The DR box 114 of the client site 102 for example comprises a regulator for continuous loads (C LOADS REG) 115, which for example provides control signals to the one or more continuous loads 110 via a continuous loads PLC 113, which for example locally manages operation of the one or more continuous loads 110. The DR box 114 of the client site 103 for example comprises a regulator for discrete loads (D LOADS REG) 116, which for example provides control signals to the one or more discrete loads 112 via a discrete loads PLC 117, which for example locally manages operation of the one or more discrete loads 112. The DR box 114 of the client site 104 for example comprises a regulator for both continuous and discrete loads (C+D LOADS REG) 120, which for example provides control signals to the one or more continuous loads 110 via a continuous loads PLC 113, and control signals to the one or more discrete loads 112 via a discrete loads PLC 117.

[0054] The DR boxes 114 are for example capable of communicating with the central power management system 106 via the internet, and with equipment at the client site via one or more field buses. For example, while not illustrated in Figure 1, the connection between the DR box 114 and the internet is via either a land-line switched communications network, such as via an ADSL (asymmetric digital subscriber line) modem, and/or via a wireless connection, for example comprising a cellular communications network.

[0055] While in the example of Figure 1 a single DR box 114 is provided for each site 102, 103, 104, in alternative embodiments some sites may comprise more than one DR box 114.

[0056] The central management system 106 for example comprises a control and data acquisition system 128, which is for example a supervisory control and data acquisition system (SCADA), responsible for receiving data from each of the DR boxes 114, and for providing control signals to the client sites 102, 103, 104 via the DR boxes 114. In particular, the SCADA 128 is for example responsible for transmitting control signals to the DR boxes of the client sites 102, 103, 104.

Furthermore, the SCADA 128 is for example responsible for acquiring and storing data measurements from the sites 102, 103, 104.

[0057] The central management system 106 also for example comprises a distributed energy resources management system (DERMS) 130, which is for example a computer platform configured to organize resource operations in relation with the loads of the various client sites of the system 100. The DERMS 130 also for example provides an interface with the markets server 107, and with the power grid operator server 108.

[0058] For example, the markets server 107 provides information on electricity prices for current and/or future periods, and also information on activations requested by the system operator 108. [0059] The power grid operator server 108 for example corresponds to a computer platform of an operator of the power grid supplying electricity to the client sites. In Europe, the power grid operator corresponds for example to the TSO, and/or to the distributed system operator (DSO). For example, the power grid operator server 108 provides activation orders to the DERMS 130, and the DERMS 130 provides control data, such as monitoring data and/or load statuses, to the power grid operator server 108.

[0060] According to embodiments described herein, one or more of the DR boxes 114 of the client sites, and optionally in combination with the SCADA 128, are used to control discrete and continuous loads in order to provide primary reserve (Rl) and/or secondary reserve (R2) functionality, as will now be described in more detail. The primary reserve (Rl) is also known as the Frequency Containment Reserve (FCR), and the secondary reserve (R2) is also known as the automated Frequency Restoration Reserve (aFRR). In the following, these will be referred to as the Rl and R2 pools respectively.

[0061] Figure 2 is a block diagram illustrating a power supply system 200 similar to the system 100 of Figure 1, and like features are labelled with like reference numerals, and will not be described again in detail. In the example of Figure 2, the system 200 comprises the client sites 102 and 103, and a further client site 204, which is similar to the client site 103. The continuous loads 110 of the client site 102 correspond to two electrolysis systems 220 (ELECTRO_l) and 222 (ELECTRO_2), the discrete loads 112 of the client site 103 correspond to two furnaces 224 (FURNACE_1) and 226 (FURNACE_2), which are for example electric arc furnaces, and similarly, the discrete loads 112 of the client site 204 correspond to two furnaces 228 (FURNACE_3), and 230 (FURNACE_4), which are also for example electric arc furnaces. [0062] The electrolysis systems 220, 222 each for example comprise a high voltage rectifier 206 for generating a DC current for driving cells of the electrolysis system.

[0063] Each of the electric arc furnaces for example comprises an On-Load Tap Changer 208 having a plurality of discrete output voltage levels that can be activated, generally in order, so as to vary the total power consumed by each furnace.

[0064] Each client site 102, 103, 204, for example comprises a power monitoring device (TRANSDUCER) 210 for monitoring an amount of imbalance in the power supply network 212 that supplies each of the client sites 102, 103, 204. In one example, the amount of imbalance is detected by detecting a frequency deviation D/ of the voltage on the power supply network with respect to a nominal level, which is for example 50°Hz, and the power monitoring device 210 is a transducer used for detecting this frequency deviation. However, in other embodiments, a signal indicating the amount of imbalance could be a different type of signal, obtained in a different manner. For example, particularly in the case of an R2 pool, the signal could be a power modification command provided by a power grid operator, such as the TSO, indicating an amount to be applied of an available power variability. For example, the value could be provided on a scale from -1 to +1, where -1 corresponds to downward power generation or an increase in power consumption by the maximum available change, and +1 corresponds to upward power generation or a reduction in power consumption by the maximum available change.

[0065] Whatever the type of signal, the signal is for example provided at a sampling rate of at least one sample per second, and in some embodiments at between one and ten samples per second, for example at five samples per second, corresponding to a sample every 200 ms. [0066] In the example of Figure 2, the communications link between the SCADA 128 and each of the DR boxes 114 is via a secure network (SECURE NETWORK) 214.

[0067] A pool of the three client sites 102, 103, 204 as represented in Figure 2 provides one example of a combination of discrete and continuous loads that may be regulated together in order to provide R1 and/or R2 functionalities. For example, each of the electrolysis systems 220, 222 has a power consumption of at least 1 MW, and for example of up to 20 MW or more, of which at least 5% can be varied in a continuous manner for R1 and/or R2 pooling purposes. Each of the furnaces FURNACE_1 to FURNACE_4 for example has a power consumption of at least 1 MW, and for example of up to 15 MW or more, and in some cases of up to 30 MW or more, of which at least 5% can be varied in discrete steps for R1 and/or R2 pooling purposes.

[0068] Figure 3 schematically illustrates the electrolysis system 220 of Figure 2 in more detail according to an example embodiment. The electrolysis system 222 is for example implemented by a similar circuit.

[0069] Industrial electrolysis systems are well known to those skilled in the art. The process of electrolysis is based on the management of a DC current Iline flowing through a line of cells Cl to Cn. The DC current is generated by the high voltage rectifier 206, which for example produces a current of at least 1 kA, and for example in the range 25 to 75 kA or more. The chemical decomposition of the electrolyte is driven by the intensity set-point, with the condition that, for each cell, the cell voltage Ucell is higher than a decomposition set point. Thus, by varying the consumption of the electrolysis system by up to between 5% to 15%, chemical decomposition can be maintained with a low to full industrial production . [0070] Figure 4 schematically illustrates a regulation system 400 for continuous and discrete loads according to an example embodiment of the present disclosure. For example, Figure 4 illustrates a regulation system 400 for regulating the continuous and discrete loads 220 to 230 of Figure 2.

[0071] A block 402 of the system 400 for example corresponds to devices and loads present at the client site 102, a block 403 of the system 400 for example corresponds to devices and loads present at the client site 103, and a block 404 of the system 400 for example corresponds to devices and loads present at the client site 204. The other elements of Figure 4 are for example implemented by the SCADA 128, although they could also be implemented elsewhere, for example at the client site 102.

[0072] The block 403 is for example implemented by the DR box 114 at the site 103, except for the loads 224 and 226 and corresponding PLCs 117. The block 403 for example comprises a distributer 406, which receives a measured frequency deviation D/, which is for example generated locally by the transducer 210 of the site 103 of Figure 2. For example, the frequency deviation D/ is calculated as D/ = f — f0, where / is a measured system frequency on the power network at the client site 103, and /0 is the targeted frequency, equal for example to 50 Hz, or to another level depending on the region.

[0073] The distributer 406 for example supplies the frequency deviation D/ to comparators implemented by hysteresis latches 408, 410, each associated with one of the discrete loads. The distributor 406 for example selectively supplies the frequency deviation D/ to those latches 408, 410 associated with active loads, and in this case any loads that are not to be used for R1 or R2 regulation, due to local or centralized activation rules, are not controlled. The distributer 406 and the hysteresis latches 408, 410 are for example implemented in software executed by the DR box 114, although hardware implementations would also be possible.

[0074] In the case of a discrete load having more than two states, there are for example a plurality of hysteresis latches 408, and a plurality of hysteresis latches 410. For example, the DR box 114 is configured to operate in the same way as a Schmitt trigger, in which the latch and release thresholds are programmable.

[0075] The hysteresis latches 408 apply a plurality of thresholds to the frequency deviation in order to determine the discrete level at which the corresponding load is to be activated. The set of thresholds of the latches 408 will be called [Fb}l. In some embodiments, this set of thresholds is defined by a set of pairs of values {flatch (i),p (i)}, where i is from 1 to nd, and nd is equal to the number of discrete levels of the discrete load, as already indicated. The value flatch(i) corresponds for example to the frequency at which the latch output goes high, and p(i) is a persistence value, indicating how far below the frequency flatch(i) the deviation D/ must fall before the latch output falls low, as will now be explained with reference to Figure 5.

[0076] Figure 5 is a graph representing an activation level (LEVEL) of a discrete load as a function of frequency f. The example of Figure 5 is based on an example of a load having four discrete levels 1, 2, 3 and 4, and a set of corresponding pairs of values {Fb}. A hysteresis latch associated with the first level of the discrete load i=l has frequency flatch equal to fll. This latch is released at a release frequency frelease equal to frl, and the persistence value pi is equal to the frequency difference between fll and frl. The other levels i=2,3,4 are defined in a similar manner.

[0077] In the example of Figure 5, each of the activation levels 1 to 4 is associated with a different persistence value However, in some embodiments, a same persistence value is for example used for a plurality of activation levels, and thus the latch and release frequencies of each of these activation levels can for example be defined by only its latch frequency, and the corresponding common persistence value. This leads to a relatively low memory storage requirement for storing the set of thresholds of each latch, and also simplifies the task of defining the threshold levels of the latches. Of course, in the case of a bi-stable load, only a single pair of values is for example defined.

[0078] With reference again to Figure 4, the hysteresis latches 410 also for example apply a plurality of thresholds to the frequency deviation in order to determine the discrete level at which the corresponding load is to be activated. The set of thresholds of the latches 410 will be called {Fb}2.

[0079] The outputs of the latches 408 are for example coupled to the load 224. For example, the DR box 114 generates a control signal for controlling the load 224, via the PLC 117, based on the comparison performed by the hysteresis latches. The load 224 for example has a transfer function HDBl(s), which determines the actual power consumed by the load at the selected level. Indeed, this actual power may vary, based on many factors, typically from -65% to +20% of the nominal power level associated with the selected level of the load. One or more signals are for example monitored at the load 224 and/or by the PLC 117, and provided to a real power extraction module 412, which is for example implemented in software by the DR box 114. For example, the DR box 114 is configured to determine an estimation of a change in the consumed power of the load 224 as a result of a change in the activation level of the load 224. For example, this estimation is based on readings captured during a sliding window of maximum values and gradient values of monitored current and/or voltages, as known by those skilled in the art. Low pass filtering is for example applied in order to remove noise related to process disturbances. The result is a signal SPD1 corresponding to an estimation of the actual change in the power consumed by the load 224 as a result of a change in the activation level.

[0080] Similarly, the outputs of the latches 410 are for example coupled to the load 226, which has a transfer function HDB2(s), which determines the actual power consumed by the load at the selected level. The actual power level consumed by the load 226 is for example estimated by a module 414, similar to the module 412, which generates a signal SPD 2 corresponding to an estimation of the actual change in the power consumed by the load 226 as a result of a change in the activation level of the load 226.

[0081] The power estimations SPD1 and SPD 2 from the modules 412, 414 are for example summed, and then transmitted by the DR box 114 of the site 103 to the continuous load regulation loop. As illustrated in Figure 4, in some embodiments, the sum of the power estimations SPD1 and SPD 2 is sent to a block 420, which is for example implemented by software executed by the SCADA 128 of Figure 1.

[0082] The block 404 is for example implemented by the DR box 114 of the site 204, except for the loads 228 and 230, and corresponding PLCs 117. The block 404 is for example implemented in a similar manner to the block 403, but with a distributer 406' that receives the frequency deviation D/' generated locally, for example by the transducer 210 at the site 204 of Figure 2, latches 410', 414' associated with the loads 228, 230, these loads having respective transfer functions HDC3(s) and HDC4(s), and modules 412' and 414', which generate signals SPD3 and SPD4 corresponding to estimations of the actual change in power consumed by the loads 228 and 230 respectively as a result of changes to the activation levels of the loads 228 and 230. These signals 5PD3 and 5PD4 are for example summed by the block 404, and provided to the continuous load regulation loop, for example via the SCADA 128.

[0083] The block 420 is for example configured to sum the power estimations generated by the blocks 403, 404 to generate an estimation 5PD of the total change in the power consumed by the discrete loads 224, 226, 228 and 230. The block 402 is for example implemented by the DR box 114 at the client site 102, except for the loads 220, 222 and corresponding PLCs 113. The block 402 for example comprises a gain module (Kreg) 422, which for example applies a gain to a measured frequency deviation D/". This frequency deviation D/" is for example generated locally by the transducer 210 of the site 102 of Figure 2, in a similar manner to the frequency deviation D/.

[0084] The gain Kreg for example corresponds to a targeted control gain in MW/Hz, and the block 422 provides a target power difference DR for the group of sites 102, 103 and 204. From this target DR, an estimation 5PDc of the actual power difference generated by the discrete loads of the sites 103, 204 is subtracted, in order to generate a target residual power difference PC to be obtained from the continuous loads In some embodiments, the estimation 5PDc corresponds to the estimation 5PD . In other embodiments, the estimation 5PDc is a corrected estimation generated by modifying the estimation 5PD, as will be described in more detail below.

[0085] This target value PC is for example provided to a distributer 424 of the block 402, which for example generates setpoints to the electrolysis systems 220, 222, via the corresponding PLCs 113, based on a desired sharing of the total power variation between the systems 220, 222. [0086] In some embodiments, the power consumption of the continuous loads 220, 222 are not only controlled based on the estimated change in power 5PD of the discrete loads, but also based on a difference between a global detected power consumption of the client sites 102, 103, 204 with respect to the reality, as will now be described in more detail. This for example permits compensation to be made for any other changes in consumption at the sites 102, 103, 204 not directly related to the loads 220 to 230, so as to ensure a desired overall response of the sites 102, 103 and 204.

[0087] The loads 220, 222 have transfer functions HCLl(s) and HCL2(s), which determine the actual power consumed by these loads under the given setpoint. The actual power consumed by each of the loads 220, 222 is for example measured at the loads, and estimations 5PC1 , 5PC2 are provided by the respective loads 220, 222 indicating an estimation of the actual change in the power consumed by the loads 220, 222 as a result of the change in the setpoint of these loads.

[0088] The estimations 5PC1, 5PC2 are for example summed to provide an overall estimation 5PC for the site 102, and provided to the block 420 of the SCADA 128. In the block 420, the estimation 5PC is for example added to the estimation 5PD, to provide a global estimation dR of the power change from the continuous and discrete loads of the three sites 102, 103 and 204.

[0089] The estimation dR is for example provided to a performance control module (PERF CTRL) 426, which for example also receives an estimation SDR of an overall power difference of the sites 102, 103, 204, detected for example by the power grid operator server 108 of Figure 1. This estimation SDR is for example provided based on meter values recorded at the client sites 102, 103, 204. The module 426 also for example receives a frequency deviation D/, which may correspond to the frequency deviation measured at one or more of the sites 102, 103, 204, or to an average of the measured frequency deviations at some or all of these sites. Based on this frequency deviation D/, the module 426 for example adjusts a gain value K0 applied by a gain module 428 to the frequency deviation D/, such that this gain results in a matching between the power estimations SDR and dR. The block 420, also for example comprises a gain module (Kreg) 430, which for example applies, to the frequency deviation D/, the same gain as the module 422 of the block 420, in order to generate the target power difference DR. The output of the gain module 428 is for example subtracted from the output of the gain module 430 in order to generate an error signal. This error signal is for example corrected by a transfer function (Hcorr) 432, which for example smooths the control and adjusts a time constant of the correction. The output of the function 432 is for example subtracted from the estimation 5PD in order to generate the corrected estimation 5PDc provided to the block 402.

[0090] Figure 6 is a graph representing frequency variations (FREQ [Hz]) of the system frequency f, shown by a curve 602, over time (TIME [s]), and the corresponding system responses according to an example embodiment. The example of Figure 6 is based on one or more discrete loads providing discrete levels associated with latch and release frequencies as shown in Figure 7 described below. The example of Figures 6 and 7 could for example be obtained using the system 400 of Figure 4.

[0091] A hysteresis latch for example has a latch frequency flatch of around 50.02 Hz, i.e. a deviation of +20 mHz, such that when the frequency exceeds this level, the activation level of a discrete load is adjusted, as represented by a shaded band 604. However, it can be seen that, as a result of the hysteresis, this activation level is not removed until the frequency falls to a level frelease that is significantly below 50.02 Hz, and for example to a level just below 50°Hz in the example of Figure 6. A further hysteresis latch for example has a latch frequency flatch of around 50.04 Hz, i.e. a deviation of +40 mHz, such that when the frequency exceeds this level, the activation level of a discrete load is adjusted again, as represented by a further shaded band 606.

[0092] Similarly, a hysteresis latch for example has a latch frequency flatch of around 49.98 Hz, i.e. a deviation of -20 mHz, such that when the frequency drops below this level, the activation level of a discrete load is adjusted, as represented by a shaded band 608. However, it can be seen that, as a result of the hysteresis, this activation level is not removed until the frequency rises again to a level frelease significantly above 49.98 Hz, and for example to a level just above 50°Hz in the example of Figure 6. A further hysteresis latch for example has a latch frequency flatch of around 49.96 Hz, i.e. a deviation of -40 mHz, such that when the frequency falls below this level, the activation level of a discrete load is adjusted again, as represented by a further shaded band 610.

[0093] Of course, the example based on frequency steps of 20 mHz is merely one example, the frequency steps for example being chosen to be between 10 mHz and 50 mHz, depending for example on the desired dynamic response, on the desired number of movements, and on the number of levels of the discrete loads .

[0094] Figure 7 is a graph showing the R1 response profile (Rl%) as a function of frequency (FREQ [Hz]), and represents an example of frequency thresholds and response profiles according to an example embodiment of the present disclosure. A line 702 illustrates a desired response, wherein a power reduction of -100% of the R1 reserve is applied when the frequency is at 49.8 Hz, a power increase of +100% of the R1 reserve is applied when the frequency is at 50.2 Hz, and the percentage of the R1 reserve varies in a linear manner as a function of frequency between these points.

[0095] A solid-line staircase 704 in Figure 7 represents an example of latch frequency thresholds flatch at which the hysteresis latches, such as the latches 408 of Figure 4, react in order to counteract an increased deviation from 50 Hz, while a dashed-line staircase 706 represents an example of the release frequency thresholds frealease at which the hysteresis latches are released to relax the load control as the frequency returns towards the level of 50 Hz.

[0096] A curve 708 in Figure 7 represents an example of the response of the continuous load, which for example increases or decreases linearly as the frequency deviation from 50 Hz increases. Furthermore, upon each change of the discrete load, the power consumption of the continuous load for example jumps by an equal and opposite amount.

[0097] Figure 7 also shows an example, for an ith hysteresis latch, of the frequencies flatch (i) and frelease(i), the interval between these frequencies corresponding to the persistence p(i) of the latch.

[0098] The greater the persistence p(i), the fewer the number of changes in the level of the discrete loads as a function of frequency variations. On the one hand, this is beneficial for many types of discrete loads. For example, in the case of electric arc furnaces, tap changers are used to vary the power consumption between the discrete levels, but moving parts of the tap changers are sensitive to wear, and involve heavy maintenance when a certain number of movements have been applied. Therefore, it may be desired to limit the number movements over given period, a typical example being to limit the number of movements to 25000 transitions per tap changer per year. On the other hand, the greater the persistence, the slower the reaction, and the less linear the response of the reserve regulation.

[0099] Figure 8 is a graph representing, for a period of one year, a simulated number of movements of each level of a discrete load as a function of the frequency flatch used to activate the level, and as a function of the persistence. In particular, curves 801 represent a persistence of 50 mHz, curves 802 represent a persistence of 45 mHz, curves 803 represent a persistence of 40 mHz, curves 804 represent a persistence of 35 mHz, curves 805 represent a persistence of 30 mHz, curves 806 represent a persistence of 25 mHz, curves 807 represent a persistence of 20 mHz, curves 808 represent a persistence of 15 mHz, curves 809 represent a persistence of 10 mHz, curves 810 represent a persistence of 5 mHz, and curves 811 represent a persistence of 0 mHz. A trace 812 represents a cumulative sum of the discrete load movements, corresponding to a latch or a release, and a line 814 indicates a maximum number Cm of movements, which is for example set at 100k movements for the four discrete loads 224, 226, 228, 230, i.e. 25k movements per load.

[0100] It can be seen from Figure 8 that it is possible to choose a persistence that is relatively low, such as a value of 20 mHz, while still remaining below the maximum number of 100k movements.

[0101] An algorithm for determining the persistence based on a given maximum number of movements is described in the publication entitled "Provision of frequency containment reserve through large industrial end-users pooling", E. Perroy et al., IEEE Transactions on Smart Grid (Volume: 11, Issue: 1, Jan. 2020). [0102] Figure 9 is a block diagram illustrating the DR box (DR BOX) 114 of the client sites 102, 103 and 204 of Figure

2 according to an example embodiment of the present disclosure As mentioned above, the DR box for example implements at least some of the functions of the blocks 403, 404 and 402 of Figure 4.

[0103] The DR box 114 for example comprises a processing device (P) 902 comprising one or more processors, such as one or more microprocessors or microcontrollers. The DR box 114 further comprises a memory (MEMORY) 904 and input/output interfaces (I/O INTERFACES) 906 and/or routers/modems (MODEMS) 907 linked to the processing device 902 via a bus 908. For example, the memory 904 is a non-volatile memory such as a FLASH memory, and stores firmware that is executed by the processing device 902 in order to implement the functions of the DR box 114. The memory 904 of the DR box 114 may further comprise volatile memory, such as a RAM (random access memory), for example a DRAM (dynamic RAM) or SRAM (static random access memory).

[0104] The SCADA 128 of Figure 2, which for example implements off-site functions of the system 400 of Figure 4, is for example implemented by similar hardware to the example of Figure 9, and may be implemented by one or more virtualized machines .

[0105] Figure 10 schematically illustrates a regulation system 1000 for continuous and discrete loads according to a further example embodiment of the present disclosure. The embodiment 1000 of Figure 10 is similar to the regulation system 400 of Figure 4, and like features have been labelled with like reference numerals, and will not be described again in detail.

[0106] However, in the example of Figure 10, each of the blocks 403, 404 for example comprises a respective memory 1002, 1004 coupled to the hysteresis latches 408, 410 and

408', 410'. For example, the memory 1002 of the block 403 stores a plurality of different versions of the parameter sets {Fb}l and {Fb}2 of the hysteresis latches 408 and 410 respectively, and similarly, the memory 1004 of the block 404 stores a plurality of different versions of the parameter sets {Fb}3 and {Fb}4 of the hysteresis latches 408' and 410' respectively. Furthermore, the block 420 for example comprises a module (LOAD VERIF/UPDATE) 1006 for verifying and updating the latch and release frequencies of the latches.

[0107] The module 1006 for example receives a signal STATE OF LOADS indicating the state of one or more of the discrete loads of the sites 103, 204, and for example indicating when one or more loads is offline for maintenance, or otherwise unavailable for use in frequency regulation. Additionally or alternatively, the module 1006 for example receives the power estimations 5PD1, 5PD2, 5PD3 and 5PD4 from the blocks 403 and 404, and is configured to detect, based on these power levels, when the behavior of any of the loads has deviated by more than a certain amount from expected levels. If so, one or more of the parameter sets {Fb}l, {Fb}2, {Fb}3 and {Fb}4 is for example updated as a consequence, for example by transmitting a signal from the module 1006 to the hysteresis latches 408, 410, 408', 410' in order to modify the version of the parameter set, stored in the memory 1002, 1004, to be applied by the corresponding latches. For example, if it is detected, based on the power estimation 5PD1, that the load 224 is not functioning as expected or is otherwise unavailable, new versions of the parameter sets {Fb}l, {Fb}2, {Fb}3 and

{Fb}4 are for example activated in the memories 1002, 1004, wherein these version exclude the load 224, and perform the discrete load regulation based only on the loads 226, 228, 230. An advantage of this is that it allows the frequency regulation to adapt to changing load conditions.

[0108] It would also be possible that, if it is found that the power reduction resulting from a given change of level of one of the loads is far lower than expected, the latch frequency flatch(i) associated with this change of level could also be dynamically adjusted as a consequence.

[0109] The solutions described above in relation with Figures 2 to 10 involve a pooling of discrete and continuous loads at a plurality of client sites in order to provide a global solution. However, for sites having both discrete and continuous loads, such as the site 104 of Figure 1, it is possible to provide a single device that regulates, on-site, the set-points to be applied to the discrete and continuous loads in order to perform power balancing. An embodiment of such a device will now be described with reference to Figure 11.

[0110] Figure 11 schematically illustrates a regulation device 1100 for continuous and discrete loads according to yet a further example embodiment of the present disclosure. In some embodiments, the device 1100 is implemented as an embedded solution in the DR box 114 of the site 104 of Figure 1. For example, the device 1100 locally implements the functions of the blocks 402, 403 and 404 of Figure 4, and in some cases, some of the functions of the block 420.

[0111] The device 1100 for example comprises a load controller (LOAD CONTROLLER) 1102 for supplying setpoints (OUTPUTS) to discrete and continuous loads. For example, discrete loads are controlled by a stage target signal stg_tgt generated by a discrete control signal generator (DISCRETE CTRL GEN) 1104, which for example implements hysteresis latches similar to the latches 408, 410, 408', 410' of Figure 4. Continuous loads are for example controlled by a continuous delta setpoint signal ASp indicating a change to be applied with respect to a current setpoint SpO. Alternatively, the continuous loads are controlled by a direct process setpoint Sp. These setpoints are for example generated by a regulator (REGULATOR) 1106, which for example receives the current setpoint SpO as an input (INPUT) from the load controller 1102 in the case that the continuous delta setpoint signal ASp is to be generated.

[0112] The discrete control signal generator 1104 for example receives the frequency deviation D/, which is for example generated by a frequency signal processing element (FREQ SIGNAL PROCESSING) 1108 based on a system frequency f measured by a sensor (FREQ. + POWER SENSOR) 1110, and received as an input (INPUT). The generator 1104 also for example receives, on a line 1112 from a parameter update module 1114, a matrix {Fb}D indicating the parameter sets to be applied by the hysteresis latches of the generator 1104 for controlling the discrete loads, for example defined by flatch frequencies, and persistence values. The generator 1104 for example comprises a memory for storing the parameter sets of the hysteresis latches to be applied to the detected frequency deviation Af.

[0113] The regulator 1106 for example receives a signal AP indicating the power change to be applied, which is for example generated based on one or more power measurements P from the sensors 1110, which indicate for example the real power consumed by the discrete loads. In some embodiments, the signal AP is generated by a corrector module (CORRECTOR) 1114, a continuous setpoint generator (CONTINUOUS SETPOINT GEN) 1116, and a performance control (PERF CTRL) module 1120, which for example perform functions similar to the modules 422, 426, 428, 430 and 432 of Figure 4. For example, corrector module 1114 receives a signal APcons. from the generator 1116, and the gain value KO from the performance control module 1120. The generator 1116 for example receives the frequency deviation D/ from the element 1108, and, on a line 1118 from the parameter update module 1114, a vector VbC indicating for example a static gain K of the power consumption of the one or more continuous loads as a function of the frequency, and, in some embodiments, ranges of activations of the continuous loads. For example, in some embodiments, the static gain K is a piecewise continuous function, where the level of gain is different between different frequency ranges. The corrector module 1114 and the performance control module 1120 both for example receive a correction value Pcorr provided by inputs (INPUTS) from a SCADA interface (SCADA INTERFACE) 1122, which for example communicates with the SCADA 128 of Figure 1. For example, the correction value Pcorr indicates an estimate of the actual difference in power associated with the client site 104, detected for example by the power grid operator 108 of Figure 1. Thus, this estimation is similar to the offset estimation output of block 432 based on SDR of Figure 4, but adapted for the client site 104.

[0114] A power measurement P from the sensor 1110, and an indication of a freed power level stg_ P of an activated stage, are for example provided as outputs (OUTPUTS) to the SCADA. For example, freed power level stg_ P is similar to the values 5PD1 to 5PD4 of Figure 4, and can be used if there are one or more other off-site continuous loads to be controlled in association with the discrete loads of the site 104, and/or if there is no continuous load at the site 104. For example, the SCADA 128 is configured, in such a case, to sum the value stg_ P with similar values from other sites, and to control the one or more other continuous loads via the input Pcorr of a corresponding device 1100 at another site. [0115] The performance control module 1120 for example receives the power and frequency measurements P and f from the sensor 1110, and the vector {VZ)]C on the line 1118. This vector is for example used by the module 1120 to determine a local baseline for performance control, using for example a linear regression model.

[0116] An advantage of embodiments described herein is that power balancing can be implemented in an efficient manner based on a combination of discrete and continuous loads at one or more client sites.

[0117] Furthermore, by performing, at each client site, a local real-time measurement of the frequency deviation, the real-time regulation autonomy of each site can be ensured.

[0118] Various embodiments and variants have been described. Those skilled in the art will understand that certain features of these embodiments can be combined and other variants will readily occur to those skilled in the art. For example, it will be apparent to those skilled in the art that the systems and devices described herein could be applied for any number of discrete and continuous loads present at one or more sites. It will also be apparent to those skilled in the art that whereas embodiments have been described in which a local measurement of a frequency deviation is used as a basis for detecting imbalance in a power supply network, in other embodiments, for example in the case of R2 pooling, another type of signal could be used as the bases for detecting the imbalance .

[0119] Furthermore, it will be apparent to those skilled in the art that the centralized load verification module 1006 could also be used in conjunction with the regulation device 1100 of Figure 11. [0120] Finally, the practical implementation of the embodiments and variants described herein is within the capabilities of those skilled in the art based on the functional description provided hereinabove.

Claims

1.A method of balancing power supply and demand in a power supply network (212) comprising:

- selecting, by a first device (114) at a first client site

(103, 204, 104), a first set of activation thresholds

({Fb}l) from among a plurality of sets of activation thresholds;

- modifying, by the first device (114), an activation level of a first discrete load (224) from a first level to a second level based on a first signal (D/) indicating an amount of imbalance in the power supply network and based on the first set of activation thresholds {{Fb}1);

- generating, by the first device (114, 1100), a first estimation {5PD1) of a change in a power level consumed by the first discrete load resulting from the change to the activation level;

- calculating a residue power level (APC) to be consumed by one or more continuous loads based on the first estimation {5PD1) and on the first signal (D/) or a second signal (D/") indicating the amount of imbalance in the power supply network; and

- generating a control signal to a first continuous load (220) of the one or more continuous loads based on the calculated residue power level.

2. The method of claim 1, wherein the first estimation indicates a regulation fault in the first discrete load (224), the method further comprising selecting, based on the first estimation {5PD1), a second set of activation thresholds from among the plurality of sets of activation thresholds.

3. The method of claim 1 or 2, wherein the first signal (D/) is a frequency signal indicating a system frequency of a supply voltage on the power supply network (212).

4. The method of claim 1 or 2, wherein the first signal is a power modification command signal indicating a requested change in power consumption.

5. The method of any of claims 1 to 4, wherein a deactivation of each activation level is applied with hysteresis, each activation threshold corresponding to a latch threshold (flatch) at which the corresponding activation level is triggered, and being associated with a further release threshold (frelease) at which the corresponding activation level is no longer applied.

6. The method of any of claims 1 to 5, wherein the plurality of sets of activation thresholds are stored in a memory (1002) of the first device (114).

7. The method of any of claims 1 to 6, wherein the first discrete load (224) and the first continuous load (220) are located at the first client site (104), and wherein the calculation of the residue power level, and the generation of the control signal, are performed by the first device (1100) at the first client site.

8. The method of any of claims 1 to 6, wherein the first discrete load (224) is located at the first client site

(103), and the first continuous load (220) is located at a second client site (102), and the generation of the control signal is performed by a second device (114) at the second client site.

9. The method of claim 8, wherein calculating a residue power level (APC) to be consumed by one or more continuous loads is at least partially performed by a central power management system (128).

10. A device (114) for balancing power supply and demand in a power supply network (212), the device comprising: a memory (1002) storing a plurality of sets of activation thresholds; and one or more circuits, and/or one or more processors under control of instructions stored in an instruction memory, configured to:

- select a first set of activation thresholds {{Fb}1) from among the plurality of sets of activation thresholds;

- modify an activation level of a first discrete load (224) from a first level to a second level based on a first signal (D/) indicating an amount of imbalance in the power supply network and based on the first set of activation thresholds ({Fb}1);

- generate a first estimation (5PD ) of a change in a power level consumed by at least the first discrete load resulting from the change to the activation level; and

- transmit the first estimation to a regulation system (420,

402) of a first continuous load (220).

11. The device of claim 10, wherein the one or more circuits, and/or the one or more processors, are further configured to select, based on the first estimation (5PD1), a second set of activation thresholds from among the plurality of sets of activation thresholds, wherein the first estimation indicates a regulation fault in the first discrete load (224).

12. The device of claim 10 or 11, wherein the first signal

(D/) is a frequency signal indicating a system frequency of a supply voltage on the power supply network.

13. The device of claim 10 or 11, wherein the first signal is a power modification command signal indicating a requested change in power consumption.

14. The device of any of claims 10 to 13, wherein a deactivation of each activation level is applied with hysteresis, each activation threshold corresponding to a latch threshold (flatch) at which the corresponding activation level is triggered, and being associated with a further release threshold (frelease) at which the corresponding activation level is no longer applied.

15. The device of any of claims 10 to 14, wherein the first discrete load and the first continuous load are located at a first client site (104), and wherein the one or more circuits, and/or the one or more processors, are further configured to implement a regulation loop (1116, 1114, 1106) for controlling the first continuous load, by:

- calculating a residue power level (APC) to be consumed by at least one continuous load based on the first estimation {SPD1) and on the first signal (D/); and

- generating a control signal to the first continuous load (220) of the at least one continuous load based on the calculated residue power level.

16. A system for balancing power supply and demand in a power supply network, the system comprising:

- the device of any of claims 10 to 14, and the first discrete load (224), located at a first client site (103); and

- a second device (114), and the first continuous load (220), located at a second client site (102), the second device implementing a regulation loop for controlling the first continuous load, the second device being configured to :

- calculate a residue power level (APC) to be consumed by one or more continuous loads based on the first signal (D/), or on a further signal (D/") indicating the amount of imbalance in the power supply network, and based on the first estimation (5PD ) of a change in a power level consumed by at least the first discrete load; and

- generate a control signal to the first continuous load of the one or more continuous loads based on the calculated residue power level.

17. The system of claim 16, wherein the residue power level (APC) is further generated based on a feedback signal (5PDc), the system further comprising a central power management system (128) configured to generate the feedback signal.