Imbalance. 07 January Settlement Plain English Version (DRAFT)

Transcription

1 Imbalance 07 January 2016 Settlement Plain English Version (DRAFT)

2 Release Date Version No. Summary of Changes 27/11/ First published version for Market Rules Working Group. Second published version for Market Rules Working Group. - Updated and added sections previously TBC on topics of: o Imbalance Component; o Prices for Accepted Orders; o Trade in the Opposite Direction to the TSO; o Start and No-Load costs; o Treatment of Losses; o Units with the Ability to Offer Negative Output Range; o Special Rules for Hydro and Pumped Storage; o Supplier Units. - Removed section on Treatment of Interconnectors. - Clarity in equations for calculating QBOA or derivative quantities 07/01/ around differences between qdq and qd profiles: where qdq is an adjusted profile using other profiles (PN, qd etc.), while qd is a base profile either coming from the instruction profiling tool or from some consideration of PN for the 0 th instruction. - Corrected examples on TOTSO and substitutive PN functionality to more accurately reflect calculations, also removed TOTSO functionality from non-firm example as it would only be using the substitutive PN functionality. - Change to approach on refined approach volume calculations, instead of using opposite calculation assumptions, include an extra provision in the original calculation assumptions which caps/floors volume calculation at the FPN. - Clarification in Net Settlement and Aggregation section, removing references to NGU and clarifying that PNs for the TSSU would be optional. CONTENTS 1 Purpose Core Cash Flow Equation Consultation Paper Development of Granularity and References Final Core Cash Flow Equation with All Settlement Functionality Imbalance Component and Imbalance Settlement Period vs Ex-Ante Trading Period Premium Component and Discount Component Form of Bids and Offers Differences Within Settlement Period, Bands and Multiple Acceptances

3 4.3 Open Instructions and Closed Settlement Substitutive PNs and Trade in the Opposite Direction to a TSO Action Quantities for Pricing and Settlement Prices for Accepted Orders Refined Approach Limiting Premium / Discount Functionality Trade in the Opposite Direction to the TSO Non-Firm Decremental Undelivered Quantities Biasing PN Versus Ex-Ante Ensuring Against Double Counting Curtailment Uninstructed Imbalances Information Imbalance Charge Start and No Load Costs Mechanism Contiguous Operating Period Treatment of Losses Treatment of Unit Types Aggregators and Net Settlement of Trading Sites Units with the Ability to Offer Negative Output Range Orders Special Rules for Hydro and Pumped Storage Uninstructed Imbalances Treatment of Priority Dispatch Supplier Units

4 1 PURPOSE Imbalance settlement is concerned with calculating the volumes and cash flows in the balancing market and imbalances. Specifically it looks at two aspects: - Calculating volumes and resulting cash flows for actions taken by the TSO in the balancing market; and - Calculating volumes and resulting cash flows for imbalances by participants (delivered or not against positions procured in markets). In both of these aspects complex algebra is required to enact functionality, handle particular cases, and provide the market signals, defined in the energy market design (high level and detailed level). The purpose of this document is to outline the rules for imbalance settlement for the I-SEM in the style of a Plain-English guideline to feed into a number of processes: - The market rules development, so that the high level concepts, and detailed implementation, of each function required can be easily understood and assessed for approval or improvement; - The market rules legal drafting, so that the approach of implementing the functionality is developed to the level of detail required to be an adequate start point for the legal drafting of the final market rules; - The design of the market systems, so that the functionality is developed to the level of detail required to be an adequate start point for the drafting of the requirements and design for the systems, and the plain-english explanation of the intended purposes and outcomes of these functions ensure that any additional detail required to be defined in this process is developed in keeping with the market design. The finalised version of this document is intended to represent the baselined approach to the implementation of the imbalance settlement aspects of the design. 3

5 2 CORE CASH FLOW EQUATION 2.1 CONSULTATION PAPER From the consultation paper, the final point of the high level settlement equations for the cashflow included the general form with balancing market volumes at an integrated settlement period basis (9.2). There were also more detailed versions where functionality to limit the premium or discount depending for certain ineligible balancing market volumes. The following equation represents the high level cashflow equation 9.2 (with the changes in notation required): ( ) ( ( )) ( ( ) ( )) ( ( ) ( )) The more detailed version, focussing in on the balancing market components, is as follows: For generation unit incs: - Imbalance component: - Premium/Discount component: For generation unit decs: ( ) ( ) ( ( ) ( ) ) - Imbalance component - Premium/Discount component: ( ) ( ) ( ( ) ( ) ) The premium/discount component can more generally be represented by: For generation unit incs: For generation unit decs: ( ) ( ) Separating into different equations (for inc and dec orders, for imbalance and premium/discount components) builds in the function of ensuring the participant gets best price if inc d or dec d. 4

6 The QBOA calculation from the consultation paper algebra builds in the following functions: - Premium/Discount should only be on delivered volumes - Premium/Discount limited for non-firm constrained volumes - Premium/Discount limited where participant biases PN versus ex-ante position. This approach already requires it to be known if the unit is being dec d or inc d before it decides which equation to use. It is not a simple case of just incorporating both P/D components as current into a single equation. For example, regardless of whether a unit is inc d or dec d, if an action is taken then its QBOA can be non-zero for both the inc premium/discount and dec premium/discount component. It therefore needs a means of multiplying the desired component by zero, and the pricing parts of these calculations cannot offer this. It is not possible to ensure that Max(PBO PIMB, 0) consistently gives zero for decs while Min(PBO PIMB, 0) consistently gives zero for incs. Either of the following are required: - To allow for a single equation, a new calculation which guarantees a zero multiplier on the undesired component needs to be added. - To keep the current different equations, some means of deciding which equation to use and not to use in that instance will be needed. The second of these allows for tracking of when both incs and decs are taken on a unit in the same period. Each instance of a change in the information in those equations requiring it will trigger a calculation, the result of the calculation in that instance will be stored. There are then two choices as to what happens in instances where the design requires a change to this original value: - 1. Follow-the-BOA approach, where it is the same BOA as originally stored whose values are being changed - 2. Instance approach, where each instance calculates a new BOA, and the original BOAs are not overwritten. Anywhere the design discusses an overwriting, e.g. where PN changes reduce the volume of a previous QBOA, this is then actually done by the second instance calculating a new QBOA. The sum of all of these instances at the end should then enact the functions desired. In some instances the calculation may need to know if it is the first action in that settlement period; for example in cases where IDM trades are in the direction opposite to a BM trade. In that example, no calculation would be needed where a PN change is made while no QBOA was accepted, but a calculation would be needed where PN changes are made following a QBOA acceptance. The next step for building the core equation is to ensure the correct calculation of QBOA is included, incorporating the additional level of functionality required. This needs to be cognisant of the level of granularity available and required of different variables. This granularity has spot point in time, price / quantity band, and submission number elements. Then the additional functionality required of the settlement needs to be layers onto the core equation, or as separate equations where required. 2.2 DEVELOPMENT OF GRANULARITY AND REFERENCES 5

7 The following work attempts to build into the balancing market settlement equations the level of granularity available in terms of having different accepted orders, data submissions and bands. This is in order to identify gaps, e.g. where a level of granularity is available for one parameter and not another, but the interaction of these parameters requires something to make them comparable: For generation unit incs: - Imbalance component: - Premium/Discount component: ( ) ( ) ( ( ) ( ) ) For generation unit decs: - Imbalance component - Premium/Discount component: ( ) ( ) ( ( ) ( ) ) Here the premium/discount component does not take into account that the quantities of accepted orders will be different for different price bands. Each QBOA may need to take into account the actual PN at the time of its acceptance. QEX represents the sum of all ex-ante trades from the different markets. βo in this case is meant to indicate the relevant submission present in the system at the time that order o was accepted (e.g. there could be numerous updates to PN over time, but for calculating a particular QBOA it is the PN at the time the instruction which incurs QBOA, or submissions thereafter, which matter). A point not considered in this representation is the fact that QBOA needs to be calculated within an hour of the relevant trading period for use in imbalance pricing. Information such as the metered output and ex-ante quantities would not be known at this stage (it would be possible to know a unit s firm access quantity as this is a constant value established ex-ante). Therefore, the core equation to calculate QBOA cannot include the functionality of limiting the premium/discount on non-delivered volumes, non-firm volumes and volumes arising from PNs biased versus ex-ante volumes. Therefore, the functionality of the core BOA and the additional P/D limiting functionality will be split out. Ignoring the additional premium/discount functionality for the moment, the high level cash flow equation for the market becomes the following: 6

8 ( ) ( ( +) ( ( ) ( )) ( ( ) ( )) 2.3 FINAL CORE CASH FLOW EQUATION WITH ALL SETTLEMENT FUNCTIONALITY The following equation is a snapshot of all the settlement functionality developed in imbalance settlement, and will be used as a reference throughout the document where the detail behind the individual components will be laid out and discussed: EAQ PN DQ MQ FAQ ( ) ( ) ( ( ) ( ( ))) ( ( ) ( ( ))) (( ) ) (( ) ( )) (( ) ( )) 7

9 3 IMBALANCE COMPONENT AND IMBALANCE SETTLEMENT PERIOD VS EX-ANTE TRADING PERIOD Relevant aspects of core cash flow equation: - Ex-Ante Component: Not in scope of imbalance settlement, included as a reference to the cash flows from the ex-ante markets against which affect the impact of imbalance cash flows. - Imbalance Component: Applies the imbalance price to any difference between the ex-ante volumes procured and the metered volumes delivered. Includes instructed imbalances from dispatch instructions (implements part of decision to remunerate these at the better of the imbalance price or the bid/offer price, rest of implementation through premium/discount component), and uninstructed imbalances from PN submission, underdelivery or overdelivery. EAQ PN DQ MQ FAQ ( ) ( ) ( ( ) ( ( ))) ( ( ) ( ( ))) (( ) ) (( ) ( )) (( ) ( )) The detailed design decision stated that the functionality to implement Option (iii), which is an average price (hourly or half hourly, weighted or unweighted) of the imbalance settlement periods over the trading period duration of the ex-ante trade, should be implemented, and the decision to use this functionality would be utilised should there be no means of participants managing their exposure through sufficient granularity of ex-ante products. It then states that this average price may be best implemented through the appropriate allocation of the ex-ante volumes into the 8

10 settlement periods, i.e. the split of the trades and application of the individual prices to the imbalances gives effect to the average price across the whole imbalance for the trading period duration. This means that a relationship between the amount of imbalance in each settlement period to which each price would be applied, and the weighting factors determining the average price, must be established. The approach to the imbalance component must be able to take into account the total ex-ante position from trades in all markets over the period in question in order (for example an hour for the TPD of a DAM product) to calculate the total imbalance over the period which is to be profiled into different ISPs to enact the average price being applied to the total volume over the period. How exante products which are particular to a subset of the ISPs within the larger period over which the average is being considered are treated needs to be determined. The average price as it would be calculated over the trading period duration would look as follows: Where: ( ) - WFPIMB γ is the weighting factor, the value for which will be set during the parameter setting exercise. Examples of potential values include: o 1 for unweighted average; o QNIV γ for total system imbalance volume weighted average; o QDEMAND γ for total system demand weighted average. - The generalised period of h is specified here as α for period over which the average price is intended to be calculated (i.e. an hour or half hour), and γ for the imbalance settlement period. In order to calculate the split of the ex-ante quantities which give effect to this average imbalance price, and equivalent price (which instead is weighted by the level of imbalance on the unit rather than the weighting factor applied) can be derived and compared with the original as follows: ( ) ( ) Therefore it must follow that for each imbalance settlement period γ: Also, QIMB can be defined as follows: ( ) 9

11 Where: - QEX uα is the sum of the position from all ex-ante market orders o within a trading period duration α. Based on this, the value of the individually split out ex-ante quantities can be calculated as follows: (( ), This split of the ex-ante quantities then results in imbalance quantities in each settlement period which, when applied against the respective imbalance price, results in an average imbalance price over the trading period duration equal to PIMB α. This average price approach should be used in the case where there are no ex-ante products available at the granularity of the imbalance settlement period to directly hedge against imbalances in those periods. It means that if the net imbalance over the period α considered is zero, then the imbalance in the individual settlement periods γ is zero. The following graph shows an example of this where the imbalance settlement period is 30 minutes, but no products of that granularity are available, and hourly average pricing is used: Day-ahead 1hr (profile) T1 Profile 1hr into 2 half hr Intraday hr (profile) T2 Profile 1hr into 2 half hr Ex-Ante (post profile) α1 (T1 + T2) α2 (T1 + T2) Metered Quantity (no profile) 30 min 30 min If products of ISP granularity become available, this average pricing regime may not be needed. In this case the larger granularity ex-ante products can be split evenly into the individual ISPs. This can be thought of as taking the ISP level position of a unit due to their ex-ante trade, in the same way as taking the hourly position of a unit could be splitting a multi-hour day-ahead market product into the positions in each hour period resulting from the product. The following figure shows an example of how this could be done, where the imbalance settlement period is 30 minutes, but products of that granularity are available, and no average pricing is used: 10

12 Day-ahead 1hr (even split) T1 Split 1hr evenly into 2 half hr Intraday hr (even split) T2 Split 1hr evenly into 2 half hr Intraday half hr (no split) T3 Ex-Ante (post split) (T1 + T2) / 2 + T3 (T1 + T2) / 2 Metered Quantity (no split) 30 min 30 min Whether or not the average should be over an hour or half hour (i.e. whether α should be considered 60 minutes of 30 minutes) could be on the following basis: - If ISP is 30 minutes and the finest granularity of ex-ante product is one hour then α should be 60 minutes; - If ISP is 30 minutes and the finest granularity of ex-ante product is 30 minutes then α should be 30 minutes; - If ISP is 15 minutes and the finest granularity of ex-ante product is 30 minutes then α should be 30 minutes; - If ISP is 15 minutes and the finest granularity of ex-ante product is 15 minutes then α should be 15 minutes. The decision determining if products of sufficient granularity are available and therefore stopping the use of the average pricing approach could be taken by the SEM Committee. Depending on how and when this decision is taken or implemented, there are a number of potential cases which could arise: - Where although there are technically products of ISP granularity available to directly assist in hedging the imbalance in a period, it may be decided that it is not sufficient to change from average pricing; - If it is decided that the average should be over an hour when the imbalance settlement period is 15 minutes and products of that granularity are available in the ex-ante markets; The following graph shows an example of the complexity which would have to be explored in these additional potential cases: 11

13 Day-ahead 1hr (profile) T1 Profile 1hr into 4 quarter hr Intraday half hr (profile) T2 Profile half hr into 2 quarter hr Intraday quarter hr (no profile) T3 T4 T5 Ex-Ante (post profile) α1t1, β2t2 & T3 α2t1 & β2t2 α3t1 & T4 α4t1 & T5 Metered Quantity (no profile) 15 min 15 min 15 min 15 min If these cases arise, the approach would include the products of finer granularity (30 minutes, 15 minutes) into the total ex-ante quantity over the period considered for the average, resulting in the volume for those products being profiled into adjacent settlement periods. For example, if a 30 minute product was bought for the first half hour in an hour period, but hourly averaging of pricing was retained, then this product s quantity would be profiled into the first and second half hour of that hour period in order to enact the correct average price for imbalances over the hour. In this case by retaining average pricing, the products available could be thought of as not being sufficient to hedge imbalances in individual periods, and therefore should instead be thought of as contributing to hedging the imbalance over the averaging period considered. Keeping the volume of an ISP granular product within the settlement period to which it is relevant would not enact the correct average price over the hour. The core settlement equation then looks like this (the period in the equation is changed from the specific ISP γ to the general h period, but this is not an operational step, it is just meant to highlight that the calculation can be carried out over any general "period", as is included throughout the core equation, rather than having the calculation timeframe hardcoded as the imbalance settlement period or some other time period): ( ) ( ( )) ( ( ) ( )) ( ( ) ( )) 12

14 4 PREMIUM COMPONENT AND DISCOUNT COMPONENT Relevant aspects of the core cash flow equation: - Functionality for calculating volumes of energy procured in the balancing market for incs and decs for any desired period including allowing changes to PNs to substitute between balancing market volumes and intraday market volumes - Calculating a Premium Component for Incs if the offer price for a balancing market inc BOA is greater than the imbalance price - Calculating a Discount Component for Decs if the bid price for a balancing market dec BOA is less than the imbalance price - Functionality for determining the relevant price for each BOA. EAQ PN DQ MQ FAQ ( ) ( ) ( ( ) ( ( ))) ( ( ) ( ( ))) (( ) ) (( ) ( )) (( ) ( )) 4.1 FORM OF BIDS AND OFFERS The SEMC decision on the form of offers and bids includes the provision for a two cost curve solution based on the understanding that decremental costs may not be the equal and opposite of incremental costs in the same MW range. This means a participant can submit a set of commercial offer data to the TSO in the form of two separate PQ curves, one containing a range a break points and inc prices, and the other containing a range of break points (not necessarily the same) and dec prices. 13

15 From a settlement point of view a method consistent with the approach required for calculating qboa requires a common set of breakpoints in either direction. The approach developed integrates the quantities from the up and down curves into a single curve. This means that when an inc crosses a dec breakpoint, the algebra will calculate two distinct BOAs but based on the direction of the order each will have the same price. Equally, any dec in this manner will apply the price based on the dec breakpoints but have a common price when it create two BOAs by crossing an inc. The following tables illustrate this approach: i Price Up Quantity up Price Down Quantity Down i Quantity Price Up Price Down While this approach appears to work for settlement, it would be a separate discussion how the two curve approach may be implemented in the scheduling tools. The approach outlined above only considers those prices and quantities submitted by the participant. While it may be included in the rules and business procedures that this must be done up until the max capacity or availability of a unit, the general case where this is not included should be considered. This could follow the functionality of the SEM, where the prices of the last PQ submissions would continue on for the quantities between the quantity of the last PQ submissions and the participant s max availability. 4.2 DIFFERENCES WITHIN SETTLEMENT PERIOD, BANDS AND MULTIPLE ACCEPTANCES To get an understanding of the calculation of QBOA, which should be calculated per band and as a function of time (e.g. per minute-by-minute differences), this section outlines the level of granularity required in changes to the premium/discount component. For this we need to be able to calculate how to split items into the parts of relevant bands. If a quantity, for example minute-by-minute PN, covers multiple bands in the settlement period under consideration, the banded version of that quantity must: 14

16 - Be zero for minutes where the quantity is within bands below; - Be equal to some representation of the value of the quantity for minutes where it is within the relevant band (this could be the actual quantity, or the quantity relative to the bottom of the band); and - Be equal to some representation of the maximum of the band for minutes where the quantity is within bands above. The algebra for this approach is as follows: Where: ( ) { * ( ) ( )+ ( ) ( )} { { ( ) ( ) ( )} ( ) ( )} ( ) ( ) ( ) ( ) ( ) qpn uβoh (t) is the time stamped PN profile last submitted by the participant at the time the instruction was issued (and hence the order was accepted). This equation must be developed to calculate the QBOA while taking into account the functionality of limiting the premium / discount for non-delivered quantities, non-firm capacities and biasing PNs with respect to ex-ante positions. This has to deal with how these limits operate on a function-oftime basis or an integrated settlement period basis. This will also take account of the requirement that the price / quantity bands are relative to 0 for the I-SEM. On some occasions, qd will be = qpn if that is the initial instruction issued to the unit, however it may not be the first instruction may be to follow a different profile than the PN and we will want to calculate a qboa(t) for that. In this case qd u(o-1)h would be equal to zero. Instead of building qpn explicitly into the equation, a means of implementing the required functionality is to assume that the 0 th instruction of a unit in a trading period is its PN, rather than 0, at the time of first incurring a BOA in that trading period. This would be a settlement construct only: the TSO would not physically issue the instruction, but it would be implied at the time an instruction is issued this means that the qd u(o 1)h (t) where o 1 = 0 would be guaranteed to be qpn uoβh (t). This would mean that for each period h, the accepted order counter o resets to 1 rather than being a continuous counter for all accepted orders. Where it is desired that the top line of the equation should cancel with the bottom line of the equation (i.e. where volumes for BOA are no longer incurred), the lines follow BOUR to cancel. This single equation results in positive values for inc actions and negative values for dec actions. It therefore also requires additional equations to isolate the inc profiles and dec profiles, which would be as follows: ( ) ( ( ) ) ( ) ( ( ) ) 15

17 Where qao uoih (t) is the profile of accepted orders (inc s) and qab uoih (t) is the profile of accepted bids (dec s). This approach allows for multiple acceptances of bids or offers in the balancing market for the same settlement periods, and same spot times within those settlement periods. Each acceptance results in a new BOA (i.e. a different subscript number, where an initial acceptance is o = n and the subsequent acceptance is o = n+1), and therefore a new qboa uoih (t) is calculated from the new dispatch profile inferred by the acceptance versus the previous dispatch profile. The calculation of qboa can be triggered after each settlement period boundary, calculating all volumes within that settlement period for all dispatch profiles resulting in that period. How instructions and closed acceptances interact is further discussed in Section 0. From these, the quantities required for settlement can be determined by integrating the profiles over the settlement period, or by breaking the functions of time into minute-by-minute quantities, with the results for each of the minutes within the settlement period adding up to the total result for the settlement period. Adding this, and the required additional granularity, to the general equation leads to the following: ( ) ( ( )) ( ( ) ) ( ( ) ) The volumes considered above are the Actual Volume of orders for imbalance pricing, rather than any Adjusted Volume used in the high level algebra for imbalance settlement from the consultation paper, to which the premium or discount component of settlement is attributed. The following figures illustrate the volumes being calculated under this approach. The following is the legend and colour scheme for the lines used: Max{Min*qD uo (t) BOUR ui (t)+ BOUR u(i ) (t)} Order Line Max{Min{qD u(o ) (t) BOUR ui (t)} BOUR u(i ) (t)} Physical Notification Line Dispatch Quantity (e.g. Instruction Profile) Line BOA Volume Band Upper / Lower Range Line Reference Line Physical Notification Line 16

18 Order 1 Band 3 DQ Order 2 MW DQ Order 1 i = 5 i = 4 PN i = 3 i = 2 i = 1 h h + 1 Order 1 Band 4 DQ Order 2 MW DQ Order 1 i = 5 i = 4 PN i = 3 i = 2 i = 1 h h + 1 Order 1 Band 5 DQ Order 2 MW DQ Order 1 i = 5 i = 4 PN i = 3 i = 2 i = 1 h h

19 Order 2 Band 5 DQ Order 2 MW DQ Order 1 i = 5 i = 4 PN i = 3 i = 2 i = 1 h h OPEN INSTRUCTIONS AND CLOSED SETTLEMENT The detailed design decision on Open vs Closed instructions stated that open instructions will be maintained in terms of communications between participants and TSOs, but the orders they incur will be closed in settlement according to the minimum quantity implied by the unit s technical offer data. The interaction between instructions and order acceptances needs to be explored for a number of cases: - Receipt of an instruction opening and closing an order (closed acceptance with minimum quantity implied, e.g. by technical offer data or if instruction includes stay at level X for at least time T ). This kind of order can be opened or closed in real-time or through earlier instructions; - Maintaining a previous instruction opening an order (i.e. a lack of an instruction causing a continuous open acceptance, closing according to a set rule rather than minimum quantity implied e.g.: closing acceptance minute-by-minute as explained in SEMC decision paper; closing acceptance on receipt of new instruction; closing acceptance at imbalance settlement period boundary). This kind of order can only be opened in real-time, but it can be closed either in real-time or through earlier instructions; - Receipt of a new instruction closing the previous order, opening and closing a new order, i.e. multiple acceptances (closed acceptance of entire previous order, closed acceptance of new order with minimum quantity implied). This kind of order can be opened or closed in realtime or through earlier instructions; - Receipt of a new instruction closing the previous order with no new order opened, i.e. understanding that the TSO wants to end all incurring of BOAs, with the participant to follow their PN. This kind of order can be closed in real-time or through earlier instructions. All of this needs to be considered of in the context of the decision on the ability for participants to update PNs and commercial offer data. Participants will be able to update their commercial offer data for all non-accepted volumes in the balancing market, and update their PNs, up until intraday gate closure, i.e. one hour before the time in question. 18

20 What also becomes clear from the above is that there is not a direct correlation between receipt of a dispatch instruction and instigation of an BOA. BOAs can be opened through the lack of issuing an instruction, in the case of continuous open acceptance, and the means of implementing the closed acceptance of these (e.g. minute-by-minute or at settlement period boundaries) will influence what drives the definition of a BOA also. The above can be generalised by stating that there are two different types of order in the balancing market which need to be treated differently: - Type 1 Implied Closed Acceptance: These are BOAs instigated by a dispatch instruction. They imply a number of points based on the technical offer data of a unit, in following a ramp from their previous position to their new position for opening the instruction, staying at that position for a certain length of time if there is one stated in the instruction, and following the ramp from their new position to the position which closes the instruction. This kind of order can be opened and closed at any time while the intraday market is still open, or in the balancing timeframe; and - Type 2 Continuous Open Acceptance: These are BOAs instigated by the lack of a dispatch instruction, i.e. where a previous instruction to go to a certain output level has not been closed by a subsequent instruction at the point in time required by the minimum implied acceptance of the first order a subsequent order is opened. This kind of BOA can only be opened in real-time, where it is not possible for a participant to update their PN or COD there may not be a need to have a continuous minute-by-minute closure and acceptance of one and opening of another. The prices applied to this order cannot change until the settlement period boundary, therefore within the settlement period in question the order should be closed either when a subsequent instruction is made, or at the settlement period boundary. The following figures illustrate some (but not all) of the volumes being calculated from such acceptances: Order 1 Band 3 MW DQ Order 1 DQ Order 2 i = 5 i = 4 PN i = 3 i = 2 i = 1 h h

21 Order 2 Band 3 MW DQ Order 1 DQ Order 2 i = 5 i = 4 PN i = 3 i = 2 i = 1 h h + 1 These figures show how the volume can be calculated for an individual settlement period, as the prices do not change within that period from the time the BOA is opened after gate closure. However, it does not show how the acceptance of an order from open instructions can freeze the prices in subsequent trading periods for which the possibility of updating prices and PNs are still open. The following section intends to explain this. For open instructions, the accepted volume is the minimum volume implied by the instruction; that is, by comparing the PN profile with a dispatch profile which represents the following: - Ramp from previous level to instructed level; - If min-on or min-off conditions apply then keep unit at that level for that time; and - Ramp from instructed level back to PN. Prices and PNs at the time of the instruction are then relevant to this accepted volume (i.e. if the instruction is issued before gate closure, it would be for this volume that the prices are frozen and time-stamped PNs apply for Trade in the Opposite Direction to the TSO). If this instruction is not closed in real time, (i.e. the point in time has been reached where in settlement the dispatch profile is ramping back to PN, but no instruction to actually do so in operation has been issued), then the minimum volume implied by the instruction can be thought of as keeping the unit at that output level for an additional minute before ramping to PN. This is repeated until such a time as a new instruction is issued. As this new volume is being accepted in real-time, the prices and PNs which apply in that settlement period are the last ones submitted before gate closure, not the ones frozen or applicable at the time of the initial instruction. If the ramping profile of the minute-by-minute minimum acceptance traverses into another settlement period which is still open to having prices updated, the accepted volume under the curve in that later settlement period similarly has its price frozen and has an applicable PN as determined through timestamps. An example of this is as follows. The following sequence of events over time occurs: 20

22 - At time t1 in real-time, BOA 1 is opened through a dispatch instruction to go to a certain output level; - At time t2 in real-time, BOA 2 opened as instruction is not closed; - At time t3 in real-time, participant resubmits prices, which apply from Hr3 onwards where gate is not yet closed; - At time t4 in real-time, TSO closes instruction by instructing participant to return to PN. The following figures illustrate this: Gate Closure for Hr2 Gate Closure for Hr3 MW DQ t2 t3 t4 Continuous open acceptance i = 5 i = 4 t1 Initial Acceptance Closed i = 3 PN i = 2 i = 1 HR1ISP1 HR1ISP2 HR2ISP1 HR2ISP2 Hr1 Hr2 HR3ISP1 Hr3 Gate Closure for Hr2 Gate Closure for Hr3 t2 t3 t4 i = 5 MW DQ i = 4 t1 i = 3 PN i = 2 i = 1 HR1ISP1 HR1ISP2 HR2ISP1 HR2ISP2 Hr1 Hr2 HR3ISP1 Hr3 With Price Set 1 before t3 With Price Set 2 after t3 The dashed lines ramping down from DQ show the implicit minute-by-minute closed acceptance of volumes as each minute passes in real time for the open acceptance. The solid lines ramping down from DQ show the instances where there is a need to calculate a volume, and freezing prices for accepted volumes at the time of their acceptance, while ensuring open instructions don t restrict the 21

23 ability for participants to resubmit prices and PNs and have them taken into account for relevant subsequently accepted volumes. The price within technically feasible envelope is fixed at the price which is present when it was accepted. For open instructions this means that as each minute passes in real time where the instruction is not closed, the envelope is keeping the unit on for the next minute and ramping the unit down to PN. Any changes in price within Hr1 do not affect the prices of orders in Hr1 or Hr2, because the gate closure for resubmitting prices for Hr2 is the start of Hr1, and an hour earlier for resubmitting prices for Hr1. On the basis of the timestamps of when a price resubmission has been received, and when an order is accepted (in the case of open instructions, every minute in real-time the volume for the technically feasible envelope from keeping the unit on for the next minute), it can be determined which prices are applicable to which quantities, and calculate the volumes within the bands for each order and relevant price submission accordingly. In the following example, a dispatch instruction is issued before the last hour. Note that this example does not indicate a likely operating regime; it is only used to highlight which prices are relevant for which volumes, taking into account the open instruction. The following timelines show to series of events, the results of which are reflected in the following figures: - t1 before Hr1: Instruction issued by TSO to participant to go to a certain output level; - t2 before Hr1 but after t1: participant updates their price information, applicable from Hr2 onwards where gate is not yet closed; - t3: the time for the instruction to begin according to instruction profiling (e.g. notice times etc.), order 1 is opened; - t4: order 2 opened in real time as instruction is not closed; - t5: TSO closes instruction in real time by instructing participant to return to PN. Gate Closure for Hr2 Gate Closure for Hr3 MW DQ t4 t5 Continuous open acceptance i = 5 i = 4 t3 Initial Acceptance Closed i = 3 PN i = 2 i = 1 HR1ISP1 HR1ISP2 HR2ISP1 HR2ISP2 Hr1 Hr2 HR3ISP1 Hr3 22

24 Gate Closure for Hr2 MW t3 DQ t4 Gate Closure for Hr3 t5 i = 5 i = 4 i = 3 PN i = 2 i = 1 HR1ISP1 HR1ISP2 HR2ISP1 HR2ISP2 Hr1 Hr2 HR3ISP1 Hr3 With Price Set 1 before t2 With Price Set 2 after t2 In Hr1, the prices for the open instruction volumes are not the latest set of prices resubmitted, because the time the prices were resubmitted was after the gate closure for Hr1, but before the gate closure for Hr SUBSTITUTIVE PNS AND TRADE IN THE OPPOSITE DIRECTION TO A TSO ACTION The settlement algebra must consider functionality to allow for a change in PN (due to an intraday market trade for example) after an offer or bid has been accepted. How this has handled needs to consider the following scenarios: - Allowing the volume of a bid or offer to increase or decrease depending on the change in PN; - Completely substituting out the balancing market price for the intraday price for substituted volumes (i.e. when the volume of the balancing market accepted order is reduced); - Allowing a change in PN to change the type of order accepted (for example if an inc offer was accepted, following which the PN profile increased to above the dispatch profile without a change in dispatch profile, this is now a dec bid accepted); - Accounting for changes in PN which increase the volume of an accepted bid or offer to enact functionality for trades in the opposite direction to the TSO to limit the premium/discount component for these volumes The change in the PN only needs to be taken into account in the premium/discount component, as the result of any trade in the ex-ante markets are accounted for in the imbalance component through the ex-ante quantities directly. Therefore the change in PN over time needs to be accounted for only in the calculation of volumes for bid offer acceptances, QBOA. The approach would also need to ignore any PN submissions prior to that which was taken in the first instruction incurring a BOA. One added advantage of having the PN appear in the formulation of qd u(o-1)h (t) is that functionality which handles the change in PN following a TSO action can be put in place with this reference, i.e. 23

25 reducing or increasing volumes with substitutive PNs for changes in the same direction as the BOA, and calculating volumes arising from PN changes which represent trades in the opposite direction to the TSO. The following figure outlines which issues need to be considered together in this functionality: MW DQ Order PN1 3 DQ Order PN2 4 PN2 1 2 PN1 h h + 1 The issues highlighted as requiring consideration are: 1. Trade in the opposite direction to the TSO: Changes in PN submissions after an order has been accepted which increases the volume of the order accepted; 2. Substitutive PNs: Changes in PN submissions after an order has been accepted which decreases the volume of the order accepted; 3. Closed acceptances and open instructions: There are two order types one initiated by a dispatch instruction which results in an implied closed acceptance, and another initiated by the lack of a dispatch instruction which results in a continuous open acceptance. How each are treated for trade opposite the TSO, and for substitutive PNs, are different due to differences in how and when they are accepted; 4. Calculation of qd uoih (t) and qd u(o-1)ih (t) for different order types: As discussed in point 3, there are two types of orders. As each type are accepted at different times and therefore reference different PNs, the typical treatment for multiple acceptances does not apply. In order to calculate the correct volume of an order, two things are essential: - The dispatch profile of that order (qd uoih (t)) must not be equal to the profile to be taken away from the dispatch profile (qd u(o-1)h (t)) when an order is being incurred; - The dispatch profile must equal the profile to be taken away from the dispatch profile when no order is being incurred their values must net to zero in those moments. The following diagrams are intended to highlight the desired profiles for a number of cases, considering the above issues, with the intention of highlighting the general rules which should apply to all cases. For the moment the function of dividing the BOAs into bid and offer bands is not considered. The volumes of the orders are shaded in purple, and the profiles which are used to 24

26 define them in the qboa calculation are highlighted in red (for qd uoh (t), the dispatch profile of that order) and green (for qd u(o-1)h (t), the profile taken away from the dispatch profile for that order). The following examples apply the substitutive PN functionality, but not the trade in the opposite direction to the TSO functionality: Sub PN Inc Example 1: - Order 1 implied closed acceptance from DI1 issued before IDM gate time of PN1; - Order 2 continuous open acceptance from not issuing DI, started in real time with FPN = PN2. Also valid for an implied closed acceptance from issuing DI2 in real time, instruction of the style go to X for at least time Y, where Y is beyond the period h. Order 1 MW DQ PN2 PN1 h h + 1 Order 2 MW PN2 PN1 h h Description of Order 1 Profiles: o qd uoh (t) follows maximum of FPN and profile defining the closed accepted order; o qd u(o-1)h (t) follows the previous profile, which for order 1 is the FPN; o Applies Substitutive PN functionality. 25

27 - Description of Order 2 Profiles: o qd uoh (t) follows maximum of profile defining the previous closed accepted order, and profile defining the continuous open acceptance order (which in this example ran up to the settlement period boundary); o qd u(o-1)h (t) follows the previous profile defining the previous closed accepted order; o Applies Substitutive PN functionality. Sub PN Inc Example 2: - Order 1 implied closed acceptance from DI1 issued before IDM gate time of PN1; - Order 2 continuous open acceptance from not issuing DI, started in real time with FPN = PN2. Also valid for an implied closed acceptance from issuing DI2 in real time, instruction of the style go to X for at least time Y, where Y is beyond the period h. MW Order 1 +ve DQ -ve (when using dec assumptions, for reference) PN1 PN2 h h + 1 Order 2 MW DQ PN1 PN2 h h Description of Order 1 Profiles: o qd uoh (t) follows maximum of FPN and profile defining the closed accepted order (for inc assumptions, volume which would have arisen if the profiles were using dec 26

28 assumptions included for reference). When closing the profile according to the minimum technically feasible volume, the profile defining the closed accepted order reverts back to the FPN; o qd u(o-1)h (t) follows the previous profile, which for order 1 is the FPN; o Applies Substitutive PN functionality. - Description of Order 2 Profiles: o qd uoh (t) follows maximum of profile defining the previous closed accepted order, and profile defining the continuous open acceptance order (which in this example ran up to the settlement period boundary); o qd u(o-1)h (t) follows the maximum of the PNs and profile defining the previous closed accepted order until that previous order is closed, when it then follows the FPN. o Applies Substitutive PN functionality. The following examples apply both the substitutive PN functionality and the trade in the opposite direction to the TSO functionality: Sub PN and TOTSO Inc Example 1: - Order 1 implied closed acceptance from DI1 issued before IDM gate time of PN1; - Order 2 continuous open acceptance from not issuing DI, started in real time with FPN = PN2. Also valid for an implied closed acceptance from issuing DI2 in real time, instruction of the style go to X for at least time Y, where Y is beyond the period h. Order 1 MW DQ PN2 PN1 h h

29 Order 2 MW PN2 PN1 h h Description of Order 1 Profiles: o qd uoh (t) follows maximum of PNs and profile defining the closed accepted order; o qd u(o-1)h (t) follows the previous profile, which for order 1 is the max of the PN at the time of acceptance and the FPN; o Applies both Trade in the opposite direction to the TSO and Substitutive PN functionality. - Description of Order 2 Profiles: o qd uoh (t) follows maximum of profile defining the previous closed accepted order, and profile defining the continuous open acceptance order (which in this example ran up to the settlement period boundary); o qd u(o-1)h (t) follows the previous profile defining the previous closed accepted order; o Applies Substitutive PN functionality (no TOTSO). Sub PN and TOTSO Inc Example 2: - Order 1 implied closed acceptance from DI1 issued before IDM gate time of PN1; - Order 2 continuous open acceptance from not issuing DI, started in real time with FPN = PN2. Also valid for an implied closed acceptance from issuing DI2 in real time, instruction of the style go to X for at least time Y, where Y is beyond the period h. 28

30 Order 1 MW DQ -ve (when using dec assumptions, for reference) PN1 PN2 h h + 1 Order 2 MW DQ PN1 PN2 h h Description of Order 1 Profiles: o qd uoh (t) follows maximum of the PNs and profile defining the closed accepted order; o qd u(o-1)h (t) follows the previous profile, which for order 1 is the max of the PN at the time of acceptance, and the FPN; o Applies both Trade in the opposite direction to the TSO and Substitutive PN functionality. - Description of Order 2 Profiles: o qd uoh (t) follows maximum of the PNs, profile defining the previous closed accepted order, and profile defining the continuous open acceptance order (which in this example runs up to the settlement period boundary); o qd u(o-1)h (t) follows the maximum of the PNs and profile defining the previous closed accepted order until that previous order is closed, when it then follows the FPN. o Applies Substitutive PN functionality but not Trade in the opposite direction to the TSO functionality. Because Order 2 was accepted in real time where the relevant PN was PN2, the change from PN1 to PN2 does not count as a trade in the opposite direction to the TSO. 29

31 Sub PN and TOTSO Inc Example 3: - Order 1 implied closed acceptance from DI1 issued before IDM gate time of PN1; - Order 2 implied closed acceptance from DI2 issued before IDM gate time of PN1 (multiple acceptance), instruction of the style go to X for at least time Y, where Y is beyond period h. Order 1 MW PN1 PN2 h h + 1 Order 2 MW PN1 PN2 h h Description of Order 1 Profiles: Same as Example 2; - Description of Order 2 Profiles: o o qd uoh (t) follows maximum of the PNs, profile defining the previous closed accepted order, and profile defining the continuous open acceptance order (which in this example runs up to the settlement period boundary); qd u(o-1)h (t) follows the maximum of the PNs and profile defining the previous closed accepted order until that previous order is closed, when it then follows the FPN. 30

32 o Applies both Substitutive PN and Trade in the opposite direction to the TSO functionality. Because Order 2 was incurred through a DI before intraday gate closure at the time where PN1 was the relevant PN, the change from PN1 to PN2 counts as a trade in the opposite direction to the TSO. At this point a number of observations can be made, which need to feed through to the rules and algebra to implement this functionality. One observation in particular is that only the substitutive PN functionality should be required of volumes to be used in pricing, and as the core volume which represents the BOA in settlement. Additional functionality should then be considered outside of this core to ensure that the volumes resulting from trade opposite the TSO are captured in settlement. A scenario which also must be considered is where the change in a PN results in the order being completely substituted. The base scenario to consider for this would be where the participant carries out this entirely substituting change in PN by submitting qpn which is higher than qd. If no further instruction is issued, and therefore the understanding in operation is that the unit must follow qd despite their revised qpn, this would mean that instead of the TSO accepting an inc order, they would now be accepting a dec order. There appears to be one equation for calculating qboa(t), but two sets of assumptions for the inputs to that equation depending on whether the volume for an inc or a dec is being calculated. This is complicated by the fact that an inc and a dec may need to be calculated for the same period. Therefore the calculation needs to be carried out twice, once with each set of assumptions. The correct volumes from this set of calculations are found by taking the positive results from the calculation with inc assumptions, and the negative results from the calculation with dec assumptions. This is because the inc assumptions cannot capture the functionality for substitutive PNs for decs, and vice versa for dec assumptions. 4.5 QUANTITIES FOR PRICING AND SETTLEMENT Considering the previous discussions, calculating the volume of energy procured in the balancing market needs to only take the functionality required for substitutive PNs into account. This means that after a BOA has been incurred in the balancing market, a change in the PN either results in an increase or a decrease to the volume of that BOA. The other functionality required can be captured separately to the main volume calculation in different functionality required to limit the premium / discount component. With this in mind, generalising the functionality from the previous sections into the core equations for balancing market volumes gives the following: Where for incs: ( ) { * ( ) ( )+ ( ) ( )} { { ( ) ( ) ( )} ( ) ( )} ( ) ( ( ) ( ) ( )) ( ) ( ) ( ) ( ) 31