Evolution of ancillary services needs to balance the Belgian control area towards 2018 May 2013 This study analyses the needs for ancillary services to balance the Belgian control area in 2018 and the capacity of the existing resources in the Belgian system to meet these needs. It is performed upon request by the CREG in its decision (B)120621-CDC-1162 on the approval of the applicable reserve volumes for 2013, issued on the 21th of June 2012. This study does not address the long-term issues of security of supply and generation adequacy; instead it focuses on the reserves required for the TSO to perform real-time balancing. Ancillary services study - horizon 2018 Contents 1 Glossary ......................................................................................5 2 Executive summary .....................................................................7 2.1. 2.2. Evolution of system reserve needs...................................................... 7 2.1.1. Frequency containment reserves (FCR) .................................... 7 2.1.2. Frequency restoration reserves (FRRa and FRRm) ..................... 7 2.1.3. Replacement reserves (RR)....................................................11 Available reserve resources in 2018 ...................................................12 2.2.1. FCR and FRRa resources........................................................12 2.2.2. Upward manual FRR .............................................................13 2.2.3. Downward manual FRR .........................................................13 3 Introduction and objectives ......................................................15 3.1. 3.2. 3.3. European network codes ..................................................................15 Needs for different types of reserves..................................................16 Available reserve resources ..............................................................16 4 Context......................................................................................19 4.1. 4.2. 4.3. Responsibilities of TSO and BRP ........................................................19 Increasing need for system flexibility .................................................20 Game changers for dimensioning of reserves ......................................21 4.3.1. Increase in installed capacity of VRE .......................................21 4.3.2. HVDC interconnector.............................................................22 5 Estimated FCR needs in 2018 ....................................................23 6 Estimated FRRa and FRRm needs in 2018 .................................25 6.1. 6.2. 6.3. 6.4. 6.5. Methodology ...................................................................................25 6.1.1. Methodology ........................................................................25 6.1.2. Desaturation of FRRa by FRRm...............................................26 Inputs used for FRR dimensioning......................................................27 Scenarios .......................................................................................28 Resulting FRRa and FRRm needs in 2018 ............................................31 FRR dimensioning conclusions ...........................................................33 7 Replacement Reserves ..............................................................35 8 Reserve resources available in the system ................................37 8.1. 8.2. Flexibility........................................................................................37 Available reserve resources in 2018 ...................................................38 8.2.1. FCR ....................................................................................38 8.2.2. FRRa...................................................................................39 8.2.3. Upward FRRm ......................................................................41 8.2.4. Downward FRRm ..................................................................42 9 Conclusions ...............................................................................45 References ....................................................................................46 Annex 1: Extract from CREG decision ............................................47 Annex 2: FRR dimensioning methodology .....................................48 Annex 3: Imbalance drivers ..........................................................50 Historical imbalances ................................................................................50 HVDC interconnector ................................................................................51 Outages of units and HVDC interconnector ..................................................52 Unit outages ...................................................................................53 HVDC interconnector........................................................................53 2018 estimated residual imbalances due to FOs ..................................53 Page 3 of 56 Ancillary services study - horizon 2018 PV and wind residual forecast errors ...........................................................54 PV forecast errors ............................................................................54 Wind forecast errors ........................................................................54 PV and wind ramping imbalances.......................................................55 Page 4 of 56 Ancillary services study - horizon 2018 1 Glossary ACE Area Control Error ACER Agency for the Cooperation of Energy Regulators BRP Balancing Responsible Party CCGT Combined Cycle Gas Turbine CHP Combined Heat and Power Generation CIPU Contract for the Injection of Power Units CREG Commission for the Regulation of Electricity and Gas DA Day-ahead DSM Demand Side Management DSO Distribution System Operator ENTSOe European Network of Transmission System Operators for electricity FCR Frequency Containment Reserves, currently called primary reserves (R1) FO Forced Outage FRR Frequency Restoration Reserves FRR- Downward Frequency Restoration Reserves FRR+ Upward Frequency Restoration Reserves FRRa Automatic Frequency Restoration Reserves, currently called secondary reserves (R2) FRRm Manual Frequency Restoration Reserves, currently called tertiary reserves (R3) GCT Gate Closure Time GT Gas Turbine GWh GigaWatthour HUB Platform for the exchange of electrical energy within the Belgium control area. HVDC High Voltage Direct Current ID Intraday MW MegaWatt N-1 Largest single instantaneous incident in a control block resulting in a system imbalance considered in the dimensioning of the FRR. NC LFC&R Network Code Load Frequency Control and Reserves NRV Net Regulation Volume NWE North Western-Europe OCGT Open Cycle Gas Turbine Pdef Deficit probability PV PhotoVoltaic production unit Qh Quarter hour(ly) RES Renewable Energy Sources Page 5 of 56 Ancillary services study - horizon 2018 RG CE Regional Group Continental Europe RR Replacement Reserves RT Real-time SI System imbalance TJ Turbojet TSO Transmission System Operator UK United Kingdom VRE Variable Renewable Energy resources Page 6 of 56 Ancillary services study - horizon 2018 2 Executive summary This study was requested by CREG in its decision on the applicable reserves volumes for the year 2013 [1]. The objectives of the study are to: estimate the system reserves needs in 2017/2018; verify whether sufficient reserve resources are expected to be available in the power system to cover the reserve needs in 2017/2018. For the avoidance of doubt, this study is not addressing the issue of security of supply and generation adequacy. It is strictly looking at the specific reserves to be secured by Elia in the context of its responsibility to ensure the availability of appropriate ancillary services, as defined by the Electricity Law, (art. 8 §1) and the Federal Grid Code (art. 231 and 232). 2.1. Evolution of system reserve needs 2.1.1. Frequency containment reserves (FCR) The dimensioning of FCR (currently called primary reserves) is performed at ENTSOe level. The contribution of the Belgian control area to these reserves for the Continental Europe electricity system was historically within a range of about 90 to 106 MW. This need is expected to remain more or less stable until 2018, although a small increase can be expected since: the European TSOs identified that, due to a deterioration of system frequency quality, the risk of having insufficient FCR available in the system to cover incidents increased during last years [2, 6]; in the future the FCR share for Belgium is likely to be calculated on the basis of the sum of net generation and consumption whereas before this was calculated on the net generation only [2, 6]. This evolution might result in a small increase in FCR capacity as Belgium is typically a net importer of electricity. As a result an FCR range of 95 to 110 MW is projected for 2018. 2.1.2. Frequency restoration reserves (FRRa and FRRm) Balancing responsible parties (BRPs) are responsible for balancing their perimeter on a 15minute time interval in the Belgian system [4]. BRPs have to nominate a balanced perimeter in day-ahead and have to perform intraday adjustments according to more accurate intraday forecasts and actual measurements of production and off-take, as the uncertainty on the final balancing position of the perimeter decreases towards real-time. Any residual system imbalance is in last instance resolved by Elia [3, 4] by deploying a combination of automatic1 and manual2 frequency restoration reserves (resp. FRRa and FRRm). As a result, any assessment of the required reserve volumes for FRRa and FRRm crucially depends on assumptions regarding the behaviour of BRPs. This study assumes that major additional -compared to the actual situationefforts are performed by BRPs and other market parties in order to minimize the residual imbalances in the Belgian control area. The resulting reserve needs set forth in this study are therefore only valid under this strong assumption. A very high increase in reserve needs and according costs for society is expected in case of insufficient efforts and investments or in case of status quo. The massive penetration of variable renewable energy sources (VRE) such as wind and PV, for which the output is defined by weather conditions and not by system off-take, increases the need for system flexibility to enable BRPs to balance their perimeter. VRE can however also offer part of the required flexibility to the grid, subject to availability. This study assumes that BRPs have access to –and make use of- sufficient flexibility to balance the expected position of their perimeter on a 15-minute time interval. As a result no structural residual imbalances due to a lack of flexibility to cover the predicted output of VRE is taken into account in the reserves dimensioning. 1 2 Currently called secondary reserves Currently called direct activated tertiary reserves Page 7 of 56 Ancillary services study - horizon 2018 It is thus assumed that BRPs balance this variable output within their perimeter by themselves. Furthermore all BRPs are expected to invest in (and improve) highly accurate intraday forecasts of VRE production and consumption within their perimeter. Under this assumption Elia only has to resolve the residual imbalances due to unpredictable or partially predictable events such as near to real-time forecast errors of load and production (VRE, conventional,…), outages of load and generation units and/or HVDC interconnectors,... The main drivers behind the evolutions in system reserve needs towards 20183 are expected to be: the massive increase of VRE capacity in the system, resulting in increased forecast errors and possible increased volatility of the residual system imbalance; the integration of the 1000 MW HVDC interconnector between UK and Belgium in the system, called the NEMO project, creating both very large positive and negative imbalances in case of an outage in respectively export or import mode. Schedule changes might also result in an increase in volatility of the residual system imbalance. Different scenarios for the evolution of the system reserve needs towards 2018 were simulated. The table below gives an overview of the assumptions taken for each of the scenarios. As already explained above all of the simulated scenarios assume major additional efforts and investments by market parties in forecasting, flexibility and market participation, compared to the actual situation. All simulated scenarios therefore have identical assumptions for: Developments in forecasting, metering, profiling Balancing in day-ahead timeframe It has to be emphasized that these efforts and investments in forecasting, flexibility and market participation, will only take place in case efficient and strong incentives are given to BRPs and market parties in order to minimize residual system imbalances by investing in –and exploiting all- system flexibility and by accessing electricity markets in the dayahead and intraday timeframe to balance their position. These incentives are given by the imbalance tariffs set by Elia. It is currently observed however that during some periods the current incentives tend to be inadequate, which has to be resolved in the near future in order to avoid a significant and very costly- increase in system reserve needs. Incentives given to market parties have to be adequate in order to ensure the sustainable integration of VRE in the system without requiring a continuous structural increase in system reserve needs due to a lack of system flexibility. In contrast to the above common assumptions, the simulated scenarios differ in: Balancing in intraday timeframe: the amount of available flexibility in a very short intraday notice enabling BRPs to adjust the position of their perimeter according to improved ID forecasts and actual metering; Intra-hourly balancing: the amount of available flexibility on a 15-minute timescale enabling BRPs to balance the ramping of VRE and HVDC interconnector within the hourly timeframe. The low reserve needs scenario assumes that a high share of intraday and 15-minute flexibility is available, whereas less of such flexibility is assumed to be available in the high reserve needs scenario. More detailed information on the different scenarios can be found in paragraph 6.3 and in Annex 3. 3 This study takes the assumption that NEMO will be commissioned in 2018. Page 8 of 56 Ancillary services study - horizon 2018 Assumptions for simulation Low reserve needs scenario High reserve needs scenario Developments in forecasting, metering, profiling High High High High High Low High Low Investments of BRPs in accurate intra-day forecasts of VRE production and off-take. Investments in smart metering and load profiling to have a clear real-time view on the actual off-take, injection and balancing position of the system and BRP perimeter. Balancing on day-ahead timeframe Ability of BRPs to balance the day-ahead expected position of their perimeter, including the variable output of VRE, ramping of off-take,… This requires sufficient investments in system flexibility to incorporate the high shares of future VRE capacity (in combination with the standard daily ramping of off-take). Balancing on intraday timeframe Ability of BRPs to adjust the position of their perimeter in ID according to more accurate ID forecasts of VRE production and off-take (smart metering,…). This depends on the amount of ID flexibility within the perimeter of the BRP (load and generation) and on the liquidity of ID markets. Intra-hourly balancing Ability of BRPs to balance the ramping of VRE (wind, PV,…) and HVDC interconnectors4 within the hour. This depends on: the amount of 15-minute flexibility within the BRP perimeter (load and generation) the presence of a liquid 15-minutes ID market. The figures and table below show the 2013 - 2018 interpolated system needs for FRRa and FRRm, required to resolve residual imbalances for the different scenarios under the above assumptions. The grey dotted lines show the very high increase in reserve needs in case the above assumed significant efforts and investments, common for all scenarios, do not (or only partially) take place, which would result in structural residual imbalances caused by a lack of system flexibility or insufficient forecasting quality of VRE and system off-take. 4 Depending on the final market design on the balancing responsibility for the planned BE – UK HVDC interconnector (NEMO). Page 9 of 56 Ancillary services study - horizon 2018 Scenario FRRa [MW] FRRm downward [MW] FRRm upward [MW] 2013 reference 140 695 1120 2018: low reserve needs scenario 152 1138 1078 2018: high reserve needs scenario 192 1331 1321 Insufficient efforts & investments Up to >300 MW Up to >1750 MW Up to >1700 MW These reserves are not necessarily pre-contracted by the TSO; for instance they can be offered as non-pre-contracted reserves in the balancing market; also, the reserve needs represent volumes with an availability of 100%, but equivalent volumes for the case of an availability less than 100%, taking portfolio effects into account, can be calculated. It can be concluded that the reserve needs of the system heavily depend on the BRP behaviour. In order to avoid a very steep increase in future reserve needs (and according costs for society) as indicated by the grey dotted lines, it is of crucial importance that: BRPs invest in best practice ID forecasting of VRE production and off-take; BRPs make active use of markets on all timescales to balance their perimeter; BRPs pro-actively foster the development of flexibility in their portfolio (load & generation) and bring this flexibility to the markets (day-ahead, intraday and balancing market); TSO, DSOs, BRPs and other market parties perform additional efforts to achieve accurate metering and load profiling (DSO responsibility). Adequate incentives by the imbalance mechanism are crucial to achieve this: all imbalance volumes must be exposed to imbalance prices; real-time market reaction on imbalance prices to support the balance of the BRP perimeter and the Belgian system must be incentivized; imbalance prices should be sufficiently high to incentivize investments in –and the actual use of- system flexibility by all market parties. Additional incentives might be required to achieve this as current incentives tend to be inadequate during some time periods. In case insufficient system flexibility is made available to balance the output of VRE, or in case of insufficient investments are made in accurate (intraday) forecasts, the reserve needs of the system will increase significantly and are expected to be much higher than in the simulated low and high reserve needs scenarios. In addition the simulated scenarios for the estimation of the 2018 reserve needs showed that: The need for FRRa will increase towards 2018 due to an expected increase in volatility of residual imbalances caused by ramping of VRE within the hour, the ramping of the future HVDC interconnector between Belgium and UK, forecast errors,... The extent of increase of FRRa will depend on the amount of quarter hourly flexibility available in the system. This stresses the importance of o sufficient investments in quarter hourly flexibility in the BRP perimeter (load and generation); o the development of a liquid coupled ID market with a time resolution of 15 minutes and short gate closure times (GCT) for the cost-effective integration of high shares of VRE in the system. Page 10 of 56 Ancillary services study - horizon 2018 All performed simulations assumed a very efficient desaturation of FRRa by FRRm. This requires very flexible FRRm, being activated very often and almost on a continuous basis, and a very liquid FRRm balancing market. In addition there is the need for FRRm to cover rare but large incidents, which allows also less flexible FRRm (in terms of number of activations,…). The need for upward FRRm shows an increase in the high reserve needs scenario and even a very slight decrease in the low reserve needs scenario. This can be explained as follows: o The 2012 baseline, used to simulate the 2018 needs, showed a significant decrease in negative residual system imbalances compared to 2011 (which was used as baseline to calculate the 2013 reserve needs); o There is a collateralization of the increased needs for upward FRRm due to additional VRE capacity with the upward FRRm required to cover the large N-1; o High investments in system flexibility, high increase in forecasting quality and ID flexibility and major additional efforts to minimize residual imbalances in the low reserve needs scenario compared to the current 2013 situation. The need for downward FRRm increases significantly due to: o the integration of additional VRE (positive and negative forecast errors); o the introduction of a large positive N-1 incident as result of an outage of the 1000 MW HVDC interconnector between Belgium and UK in full export mode; o the 2012 baseline, used to simulate the 2018 needs, showed a significant increase in positive residual system imbalances compared to 2011 (used as baseline for the 2013 needs). Residual positive system imbalances caused by a lack of (activated) downward flexibility are already observed today in the Belgian electricity system. This shows again the importance of adequate and strong market incentives to develop such downward flexibility in the next years. The very high increase in downward reserve needs indicate the need to develop a significant amount of downward flexibility in the system. This is elaborated further in Chapter 8. Elia will perform efforts in order to avoid a systematic long position of the residual system imbalances as observed in 2012. This would result in a small reduction of the increase of the downward FRRm needs and a small increase of the upward FRRm needs, although the general conclusions of this study remain valid 5. 2.1.3. Replacement reserves (RR) Replacement reserves (RR) are the slower manually activated reserves meant to (partially) replace the FRR after 15 minutes. In the Belgian system, market parties are authorised and incentivised to restore the balance of their perimeter even in real-time. Therefore market parties fulfil the replacement role in the Belgian system. As a result Elia does not contract RR, nor foresees to contract RR in the future. The future NC LFC&R [2] confirms that RR are optional and that the market design determines whether or not RR are required. Provided that this fundamental feature of the Belgian balancing market design is preserved, contracting of RR is not considered in the Belgian market design. 5 The average value of the 2012 residual system imbalance was 64 MW (long). The impact of assuming an average balanced residual system imbalance on the above calculated reserve volumes is assumed to be less than 50 MW due to probabilistic combinatorial effects in the dimensioning process. Page 11 of 56 Ancillary services study - horizon 2018 2.2. Available reserve resources in 2018 The second question addressed in this study is whether the reserve resources that will be most likely available in 2018 are sufficient to cover the 2018 simulated system reserve needs. The study makes a snapshot of the expected 2018 situation and does not account for the evolution of the reserve resources between 2013 and 2018. This study is not addressing the issue of security of supply and generation adequacy. It is strictly looking at the specific reserves to be secured by Elia in the context of its responsibility to ensure the availability of appropriate ancillary services, as defined by the Electricity Law, (art. 8 §1) and the Federal Grid Code (art. 231 and 232). Only reserve resources within the Belgian system are considered, due to the inherent uncertainty linked with the cross-border procurement of reserves. Elia will however continue to investigate and develop the cross-border procurement of reserves. In addition to check whether, from a system capability view, sufficient resources will be available to cover the system reserve needs, it is also of crucial importance to check whether the reserves resources will be sufficiently diversified to enable their economic efficient procurement. The following sections assume that sufficient additional investments and efforts are performed to minimize the residual system imbalances. Therefore the considered system reserve needs are those of the low and high reserve needs scenario. It is clear that, in case insufficient investments or efforts are performed, that the system reserve needs will be much higher than considered in this section, resulting in the need for significant additional investments in reserve resources. 2.2.1. FCR and FRRa resources From the point of view of system capability, under the assumptions that at least 3 of the existing CCGT units will still be available in the system in 2018 and that part of the FCR will still be delivered by demand, for FCR and FRRa it is expected that the 2018 reserve needs can be covered, for both the low and high reserve needs scenario. It has to be emphasized however that, in case of no investments in new FCR and FRRa capability, the margins for FCR and FRRa will reduce, leading to a non-sustainable situation, especially for the high reserve needs scenario. From an economic point of view, as already observed today, the actual FCR and (especially) FRRa resources do not allow an efficient procurement, as the majority of the capability is concentrated on CCGT units. Therefore procurement costs for spinning reserves are fully dependent of the CCGT market conditions, defined by the evolution of the clean spark spread. In an efficient reserve market the spinning reserves (FCR, FRRa) should be provided by running demand facilities or by power plants being selected in the merit order, such as wind parks, CCGT units, coal units, CHPs, biomass units,… depending on the market situation. Procurement of spinning reserves on power plants that aren’t selected in the merit order should be avoided as it leads to very high procurement costs due to the must run character. In order to further minimize the procurement costs of FCR and FRRa the most efficient power plants have to provide downward reserve capacity while the least efficient ones provide the upward capacity. It can be concluded that, in order to enable an economically efficient procurement of FCR and FRRa, the participation of wind, CHP units, biomass units, and where possible load, to such services must become a reality. Important pre-conditions for this to happen include: smart support schemes for RES and CHP units must set an attractive framework for the participation of these units in the ancillary service market, which is currently not always the case; new and refurbished power plants need to have FCR and FRRa capability in order to ensure the continuity of reserve capability of the Belgian production park and to ensure that the newest -and therefore often the most efficient- units are able to provide FCR and FRRa; Page 12 of 56 Ancillary services study - horizon 2018 new operation strategies for pump storage hydro units to optimize the economic efficiency of FCR and especially FRRa delivery have to be investigated further. 2.2.2. Upward manual FRR For upward FRRm, from the point of view of system capability, it is expected that the current reserve resources which are likely to be still available in 2018 will be insufficient to cover the reserve needs. This is because of the combination of on the one hand increasing needs for upward FRRm in the high reserve needs scenario and on the other hand the assumed decommissioning of the turbojets (210 MW upward FRRm) and some older OCGTs which currently provide upward FRRm. In the high reserve needs scenario about 360 MW of additional upward FRRm capability has to be developed. Moreover, from an economic point of view, a diversification of the upward FRRm resources is required in order to avoid excessive dependency on OCGT units and therefore on gas and electricity prices for the procurement of upward FRRm. It has to be noted that FRRm has two main functions: continuous de-saturation of the activated FRRa, requiring upward FRRm by: o flexible spinning resources (biomass, RES, CHPs, CCGTs/OCGTs, pump storage units,…) or very flexible demand facilities in terms of activation frequency; o very flexible non-spinning units -in terms of numbers activation frequencysuch as pump storage units,… covering rare but very large imbalances (typically outages of power units), allowing FRRm by less flexible units in terms of activation frequency such as the interruption of demand facilities,… The required new upward FRRm capability has to be created by: participation of new/refurbished OCGTs (converted CCGTs) in the upward FRRm market; participation of load in the upward FRRm market; cross-border procurement of FRRm neighbouring TSOs for FRRm capacity; further diversification (biomass, RES, CHPs,…) in the FRRm market. 2.2.3. and sharing agreements with Downward manual FRR The Belgian system already encounters difficulties due to a lack of (activated) downward system flexibility –being referred to as situations of incompressibility- during some periods in summer and even in winter. In addition the massive integration of VRE in the system and the HVDC interconnector between Belgium and UK are expected to significantly increase the need for downward flexibility by +/-700 MW in the high reserve needs scenario. Downward FRRm should in principle not be pre-contracted, as the need indicates an excess of generation in the system. A framework in which downward flexibility is offered by the available generation units or demand facilities (increase of consumption) at a price reflecting their marginal activation costs for downward regulation power is more sustainable on the long term and is required to set proper incentives for investments in downward flexibility. This is in line with the current framework for offering downward FRRm in accordance with the CIPU contract [8]. The imbalance tariffs then reflect these price signals (single marginal pricing), incentivizing BRPs and market players to invest in sufficient downward flexibility, to balance their portfolio on a 15-minute basis and to support the balance of the control area. In case of absence of sufficient downward flexibility offered by market players additional incentives in the imbalance tariffs might be needed as an intermediate measure to incentivize the development –and use of- downward flexibility. It can be concluded that a significant amount of downward manual FRRm flexibility has to be further developed towards 2018 by: Page 13 of 56 Ancillary services study - horizon 2018 maximizing the downward flexibility of existing plants (lowering minimum stable power output,…), for instance by converting CCGTs in OCGTs; requiring new and refurbished generating units to have low minimum stable power output, to be able to shut down and start quickly and to have high ramping capabilities; requiring VRE units to offer downward flexibility (required for the sustainable integration of a high shares of VRE in the system); Furthermore the same conclusions as for upward FRRm on its different functions, being desaturating the activated FRRa and covering rare but large incidents, hold for downward FRRm. Page 14 of 56 Ancillary services study - horizon 2018 3 Introduction and objectives In its decision (B)120621-CDC-1162 on ‘the request for approval of the dimensioning methodology for, and the determination of, primary, secondary and tertiary control energy for 2013’ [1], submitted by Elia, the CREG requested Elia to perform a study on both the system reserve needs and the reserve resources that will be most likely available in the Belgium system on a time horizon of 5 years (2017 - 2018)6. The objective of the study is to estimate the system reserve needs towards 2017 – 2018 and to evaluate whether the resources that will be most likely available in the Belgium system on the same time horizon will be sufficient to cover the needs. The request from CREG originates from the general concern that some market parties announced the decommissioning or mothballing of some flexible power plants, currently providing reserve capacity to the grid, in combination with an expected increase of reserve needs in the future. Furthermore the issue of system adequacy is currently being assessed by the government, which may ultimately lead to new investments and therefore new reserve resources. For the avoidance of doubt, this study is not addressing the issue of security of supply and generation adequacy. It is strictly looking at the specific reserves to be secured by Elia in the context of its responsibility to ensure the availability of appropriate ancillary services, as defined by the Electricity Law, (art. 8 §1) and the Federal Grid Code (art. 231 and 232). The next paragraphs of this chapter give a short introduction to the new reserve terminology used throughout this study (paragraph 3.1) and to the main questions being addressed in this study, being the estimation of the future reserve needs (paragraph 3.2) and the reserve resources likely to be available in the future (paragraph 3.3). Chapter 4 then elaborates on the Belgian balancing market framework, the increasing needs for system flexibility due to the integration of high shares of VRE and the game changers for the dimensioning of reserves in the future. Chapter 5 deals with the estimated future need for frequency containment reserves. The future needs for automatic and manual frequency restoration reserves are discussed in chapter 6, while chapter 7 briefly deals with replacement reserves. Chapter 8 elaborates on the reserve resources that will be most likely available in the system in the future. Finally chapter 9 concludes on the most important results of this study. 3.1. European network codes The draft of the NC LFC&R submitted for public consultation [2] strives for the panEuropean harmonization of reserves terminology, which has been taken into account in the elaboration of this paper. The table below gives a mapping of the current and future terms: Old term New term Purpose Primary reserves Frequency Containment Reserves (FCR) Contain the system frequency after the occurrence of an incident or imbalance within the Synchronous Area. Secondary reserves Automatic (FRRa) FRR Tertiary reserves Manual (FRRm) FRR Reserves with an activation time less than 15 minutes which are used to restore the ACE of the control block to zero in case of an imbalance in the block. Slow tertiary reserves Replacement Reserves (RR) 6 Frequency Containment is a joint action of all the TSOs of the Synchronous Area. The FRR consists of an automatic and a manual part. Optional reserves with an activation lead time exceeding 15 minutes that have to prepare the FRR for further imbalances. See Annex 1 for the detailed request. Page 15 of 56 Ancillary services study - horizon 2018 3.2. Needs for different types of reserves Chapters 5, 6 and 7 of this study determine the future reserve needs by projecting the currently known system evolutions to 2018. The study makes a snapshot of the 2018 situation and does not account for the evolution of reserve needs between 2013 and 2018. At this stage it is important to stress the high degree of uncertainty regarding the estimation of reserve needs on a 5 year time horizon because of: the impact of the decommissioning or mothballing of existing flexible power plants, the integration of new flexible power plants in the system and the expected shift towards demand side management on system operation and residual system imbalances faced by the TSO; the on-going work in the Network Codes, especially NC LFC&R and NC Balancing, for which the ultimate goal is a high degree of cooperation between Control Blocks in balancing and the creation of a single European balancing market; and the uncertainty on the impact of future changes in the system, such as the integration of an HVDC interconnector between Belgium and UK, and the massive increase of VRE in the system, on system operation and residual system imbalances. In order to address part of this uncertainty, Elia focused on the Belgian system only, without considering future cross-border collaborations that might impact the system reserve needs. Elia will however further investigate cross-border collaborations in the future, especially as interconnection capacity with neighbouring areas will increase. It is also important to note that this study assumes that towards 2018 there will be significant additional efforts and investments in system flexibility, market participation, production and load forecasting and accurate (smart) metering. The resulting system reserve needs set forth in this study are only valid under these assumptions. In case these assumptions are not fulfilled it is expected that the reserve needs will increase dramatically towards 2018. Furthermore Elia investigated several scenarios in order to reflect systems in which much or almost no flexibility is available within the intraday timeframe as explained further on in the document. Given the high degree of uncertainty, it is important to note that the results of this study only show general evolutions in the need for reserves, but that it is crucial for the TSO to continuously monitor the evolutions and to perform a detailed assessment of the reserve needs at least every year for the next year. As such the results of this study have only an indicative character. 3.3. Available reserve resources Chapter 8 of this study analyses whether the reserve resources that are likely to be still available in the system in 2018 are sufficient to cover the reserve needs. The study makes a snapshot of the expected 2018 situation and does not account for the evolution of the reserve resources between 2013 and 2018. It has to be noted also that Elia, as TSO, is not responsible for system adequacy, nor for new investments in flexible power plants in the Belgium system, which is the role of the government. In addition to the reserve resources capacity [MW] question from a system capability point of view, it is important to look also at the diversification of the reserve resources able to deliver the different types of reserves. This is also addressed in chapter 8. The creation of a liquid reserves market requires that different types of reserve resources are able to deliver the same type of reserves. A highly concentrated reserves market, in which only one specific resource is able to deliver a certain type of reserves, leads to the risk that procurement costs for reserves are fully coupled to market evolutions (e.g. clean spark spread in case of procurement of FCR/FRRa on CCGT units) and that insufficient reserve capacity will be available in case of reduced profitability and according reduced availability or decommissioning of the reserve resources. In an economic optimal situation either the available load (demand side management) or the power plants that are selected in the merit order deliver the spinning reserves (FCR and FRRa), in which case the power plants operating at lowest cost deliver the downward part of the reserves and the power plants operating at highest cost deliver the upward part Page 16 of 56 Ancillary services study - horizon 2018 of the reserves. Tertiary reserves are delivered by load, non-spinning power plants or by flexible spinning power plants. The results of this study have to be considered in a context with a high degree of uncertainty on: future market evolutions, such as fuel costs, electricity spot prices,… which will impact the available resources in the system; the amount of demand side management (DSM) available in 2018; evolution towards smart support schemes for RES and CHP units in order to facilitate their participation in the reserves market; technical ability of wind plants or other renewables to provide FCR/FRRa in 2018; … Due to the uncertainty related to the cross-border procurement of reserves, this study only considers reserve resources available within the Belgian system. Elia will however continue to investigate the option of cross-border exchange of reserves. Page 17 of 56 Ancillary services study - horizon 2018 Page 18 of 56 Ancillary services study - horizon 2018 4 Context 4.1. Responsibilities of TSO and BRP Article 157 of the Federal Grid Code [4] states that BRPs are responsible to perimeter on a 15-minute time interval. BRPs have to do so by accessing system flexibility in consecutive timeframes. As real-time approaches, BRPs insight on the final balancing position of their perimeter and have to adjust accordingly in the intraday timeframe. balance their the available get a clearer their position The TSO resolves the residual imbalances in the system due to unpredictable or limited predictable events. These are caused by forecast errors of load, VRE production,… still existing near to real-time and by outages of large loads, production units and HVDC interconnectors7 that, given their size, cannot be compensated immediately by the BRPs in question. BRPs are incentivised to balance their perimeter before and even in real-time via the real time price signals of the imbalance mechanism. This is illustrated in the figure below: The imbalance mechanism must give adequate incentives to BRPs and market parties in order to invest in accurate intraday forecasting of load and VRE production and to invest in flexibility in their perimeter (by exploiting all flexibility of existing load and generation assets and by investing in new flexibility resources if required). BRPs need this flexibility to balance the variable output of VRE, the ramping of system off-take, and forecast errors of load and production,… in their perimeter and to adjust the position of their perimeter in the intraday timeframe in accordance with more accurate forecasts. The imbalance mechanism must ensure at all times that BRPs also deploy the required flexibility to balance their perimeter in order to minimize residual imbalances in the system. The reserve needs of the system -and according costs for society- are determined by the amount of residual imbalances. The imbalance mechanism is therefore an important tool to ensure that sufficient investments in accurate forecasts and system flexibility are made by BRPs and other market parties. This ensures the sustainable integration of the planned VRE capacity in the system and minimizes costs for system reserves that are required to resolve residual imbalances. The Belgian market design allows -and incentivizes- the real-time (passive) reaction of market parties (with a physical position) on imbalance prices, thereby supporting the system balance and reducing residual system imbalances. This is of key importance to foster investments in system flexibility by market parties and to minimize residual imbalances. The dimensioning of reserves is based on residual imbalances and does not account for imbalances caused by a structural lack of system flexibility or lack of qualitative forecasts. The financial incentives given by the imbalance mechanism must be sufficiently strong to avoid such situations and to make sure that BRPs exploit all reasonable measures to balance their perimeter. The current incentives however tend to be inadequate during some time periods. 7 Market arrangements for the outage of the NEMO HVDC interconnector are not defined yet. Depending on the final framework either Elia or a BRP can be responsible for balancing the cable. Page 19 of 56 Ancillary services study - horizon 2018 4.2. Increasing need for system flexibility System flexibility is defined as the combined flexibility of generation to meet the instantaneous load and of load to meet the instantaneous amount of generation. System flexibility is also provided by the day ahead and intraday electricity markets by the import and/or export of electricity, whereas in addition these markets enable the exchange of system flexibility between market parties in the Belgian system (which is also enabled by the HUB). Historically the need for system flexibility was mainly defined by the daily peak and offpeak fluctuations of the system load. In such case the import and export of electricity, together with the centralized production units were operated in a way to match the load. The system consisted mainly of baseload units, running all day, and peak units running only during peak hours. This situation changed considerably due to the integration of VRE in the system (see chapter 6). The available production output of these resources is defined by weather conditions, regardless the instantaneous amount of load. As a result the need for system flexibility increased, as the rest of the system now needs to be flexible enough to cover the load minus the invariable RES production, for which the profile is more fluctuating and more challenging to cover than the historic peak – off-peak profile. Following cases give some examples for the increased need of system flexibility due to the integration of wind and solar: Suppose that during the upward ramping of the load in the morning (6 - 8 A.M.) the amount of wind production decreases significantly. This would mean that the import, together with the upward ramping of other power plants (or demand side reaction), now have to cover both the decrease of wind production and increase of load. This results in an increased need for system flexibility; In case the VRE production and baseload production exceeds demand, energy has to be exported or baseload (or VRE) production has to be modulated, which again indicates the increasing need for system flexibility. VRE however can –and must be incentivized to- also offer some flexibility to the grid (especially downward reserves), subject to availability. VRE has to provide some of the required system flexibility to enable its sustainable integration in the system. The table below gives an idea of the observed and/or forecasted ramping of onshore and offshore wind production and of photovoltaic production in function of the installed capacity. It can be concluded that especially offshore wind production has a high variability, with fluctuations of its output of more than 13% of its installed capacity on a 15-minute basis and more than 30% on a 60-minute basis. This represents a real challenge for the system when looking at the expected massive increase of VRE, and especially of offshore wind production in the next years. Maximum8 ramping 15 minutes 30 minutes 1 hour Estimated 2012 – 2018 additional installed capacity [GW]9 PV 4,6% 8,7% 16,9% 2 Onshore wind 4,9% 7,7% 12,2% 0,9 Offshore wind 13,5% 21,3% 31,8% 1,9 [% of installed capacity] The ramping of renewables in timescales longer than 1 hour is even much higher, requiring a very flexible system that is able to deal with this variability. Literature states that the amount of system flexibility available in the system in a timeframe of 6 to 36 hours [5] this is the capability of the system to alternate between situations with full PV and wind production to situations without PV and wind production - is the limiting factor for the integration potential of VRE in the electricity system. 8 Covering 99-percentile of the historical observed absolute values of the ramps. For PV only daytime was taken into account in order to give realistic values. 9 Based on current estimations. These values change during time and require close follow-up by Elia. Page 20 of 56 Ancillary services study - horizon 2018 The figure below gives a qualitative insight in the relationship between output of VRE, the need for system flexibility deployed by the BRP to cover this variability and the residual system imbalance which results from the part of the forecast error, still existing near realtime, that is not covered by the BRP. For illustrative purposes, the figure assumes a constant off-take of the grid and shows the required flexibility to cover e.g. the decreasing output of VRE output in order to keep the system balanced. Although constant off-take is assumed, it has to be noted that (part of) the required flexibility can also be provided by DSM. It can be concluded that the need for system flexibility for the BRP, to offset the predicted variable output of VRE, is many times bigger than the residual system imbalance to be covered by the TSO. The BRP (and other market parties supporting the balance of the system) are assumed to offset any forecast error in real-time. The dimensioning of reserves only relates to the residual system imbalances, while it is assumed that BRPs will deploy sufficient flexibility to offset the variability of the VRE output and perform best efforts to resolve forecast errors of VRE and load. This shows again the importance of adequate and strong price incentives given by the TSO to BRPs for resolving the residual system imbalances. These incentives ensure that BRPs will invest in –and actually deploy- the required flexibility, thereby leaving only small residual imbalances in the system to be covered by the reserves of the TSO. 4.3. Game changers for dimensioning of reserves This paragraph summarizes the most important system evolutions having an impact on the reserve needs towards 2018. More details on this topic can be found in Annex 3. 4.3.1. Increase in installed capacity of VRE It is estimated that the installed wind capacity will increase with +/-2,8 GW between 2012 and 2018 (0,9 GW onshore and 1,9 GW offshore). The PV capacity is expected to increase with an additional 2 GW between 2012 and 2018. This results in a total predicted amount of more than 8GW of VRE capacity installed in 2018. The significant increase in installed capacity of PV and wind production in 2018 will have an impact on the size and nature of the residual system imbalances, used as input for the dimensioning of the reserves, due to: Forecast errors of VRE output: The actual output of VRE is difficult to predict. The uncertainty on the forecasts (resulting in forecast errors) decreases towards real-time. The dimensioning of reserves accounts for residual system imbalances caused by forecast errors of VRE Page 21 of 56 Ancillary services study - horizon 2018 output still existing near to real-time for which the BRP has only limited possibility to offset them by its own. Ramping of VRE output: It is assumed that BRPs fully compensate the ramping (variability) of VRE on an hourly timeframe by themselves as sufficient hourly flexibility (1-hour energy blocks on the spot and intraday market,…) is expected to be available. Therefore ramping of VRE in an hourly timeframe is assumed not to introduce hourly residual system imbalances. Depending on the amount of 15-minute flexibility available in the system (e.g. by having a liquid 15-minute ID markets, flexible spinning units, demand flexibility,…), the ramping of a large amount of VRE inside an hour might introduce quarter hourly residual system imbalances (see Annex 3 for more details). In a system where quarter hourly flexibility is scarce, BRPs will only be able to offset these quarter-hourly imbalances to a limited extent. This increases both the size and volatility10 of the residual system imbalances. The observed residual imbalances for recent years showed that, although the size of the residual imbalances increased during last years, the volatility remained more or less constant. This is shown in respectively the left and right graph below. This indicates that the increase of volatility of the residual system imbalance due to this effect is –up to now– very limited and negligible. Further follow up of this effect by the TSO is required. 4.3.2. HVDC interconnector A second evolution in the system impacting the need for reserves is the introduction of a 1000 MW HVDC VSC interconnector between Belgium and UK (NEMO project)11. Following two effects impact the required amount of reserves: HVDC interconnector outages: The forced outage of an HVDC interconnector can occur in situations of full import or export. In case of full import the loss of the HVDC interconnector results in a deficit of 1000 MW, which is comparable to the loss of a nuclear unit in the current system. In case of full export, the loss of the cable introduces an excess of 1000 MW, which is completely new for the system. Ramping of HVDC cable: The ramping of the planned HVDC cable is expected to increase both the size and volatility of the residual system imbalances, as explained further in Annex 3. 10 Volatility is measured by the difference between consecutive quarter-hourly residual system imbalances. 11 This study takes the assumption that NEMO will be commissioned in 2018. Page 22 of 56 Ancillary services study - horizon 2018 5 Estimated FCR needs in 2018 The required amount of FCR is defined on the level of ENTSOe. Currently the total amount of FCR within Continental Europe is 3000 MW, covering the instantaneous outage of two of the largest units within Europe. The share for Elia for the Belgian control area is historically within a range of about 90 – 106 MW and is determined by the amount of the net generation of the Belgian area compared to the net generation of the entire Continental Europe system. This ratio was more or less constant with only very limited changes. In the report of the Ad-hoc Team Operational Reserves [6] the impact of deterministic frequency deviations on the dimensioning of the FCR is identified. Deterministic frequency deviations are instantaneous frequency deviations occurring during the hourly changes of the energy schedules. These frequency deviations occur due to the fact that the hourly energy blocks on the market do not allow load-following and due to insufficient alignment of simultaneous ramping of different units within the system. These frequency deviations are deterministic (always occur around the change of the hour) and, due to their specific nature, disappear automatically as time progresses. The figure below shows, on the top left, the distribution of the 2010 historical frequency deviations for RG CE. The red curve shows the frequency deviations at the change of the hour, the blue curve shows the one within the hour. The image beneath shows a normalized distribution, which shows that practically all the large frequency deviations in the Continental Europe system (RG CE) are located around the change of the hour. The solution for these imbalances lays rather in a re-design of the market rules and a better alignment of the ramping of units, addressing the main causes of these imbalances, than in the increase of operational reserves. These deterministic imbalances are quite big (several hundreds of MWs) and very fast (phenomena within a few minutes). Therefore they can only be resolved by the fast FCR. During such frequency deviations 50% up to 70% of the FCR is activated. This results in a situation in which insufficient FCR is available to cover the outage of the 2 largest units. The final report on operational reserves by ENTSOe [6] and the current draft of the NC LFC&R [2] require that the risk of having insufficient FCR is monitored for each Synchronous Area, in order to be able to increase the FCR in case an allowed risk threshold is exceeded. Simulations performed show that, in case the yearly number of minutes where the average frequency deviation exceeds more than 75 mHz increases from 2360 (2010 value) to 4000 minutes, the required FCR would also increase from 3000 MW to 3070 MW. In addition the current draft of the NC LFC&R [2] states that the FCR contribution of TSOs shall be calculated in the future on the basis of the sum of net generation and consumption of control blocks, which also leads to a small increase for Belgium as importer of electricity. Therefore a 5 MW increase in FCR is assumed to the historical range of ± 95–105 MW. Based on this information, the amount of FCR to be provided by Elia in 2018 is estimated to be within a 95 - 110 MW range. Page 23 of 56 Ancillary services study - horizon 2018 Page 24 of 56 Ancillary services study - horizon 2018 6 Estimated FRRa and FRRm needs in 2018 6.1. Methodology 6.1.1. Methodology In contrast to FCR, the required amount of FRR (FRRa and FRRm12) is determined on the level of the TSO. Historically TSOs used deterministic dimensioning criteria to determine the total amount of FRR. A TSO dimensioned its FRR in order to cover the biggest outage due to the loss of a single grid element. Such a deterministic method doesn’t take the increasing complexity of system operation due to the integration of variable and limited predictable VRE into account. Elia changed its reserves dimensioning methodology from deterministic to probabilistic in 2012. The reserve volumes for 2013, as calculated in 2012, were based on this probabilistic methodology. The determination and evaluation of reserves for 2013 on this basis was approved by the CREG [1]. In the probabilistic methodology, the total required FRR (FRRa and FRRm) is determined as the amount of FRR that covers the expected residual system imbalance in both directions during 99,9% of time13. This allows to take imbalances due to forecast errors of load and VRE into account. The methodology is explained in the figure below. The N-1 principle is however kept as a strict minimum for the amount of total FRR, as required by ENTSOe. Some general principles for the dimensioning of the reserves can be found in the ENTSOe Operation Handbook Policy 1 [7]. The TSO dimensions the FRRa in order to reach a satisfactory balancing quality for its Control Block. As FRRa is activated on an automated basis and is faster than FRRm, it is a very important reserve to ensure a satisfying balancing quality on a continuous basis. Finally the combination of FRRa and FRRm must allow the TSO to respect the ACE quality targets agreed upon between the TSOs of the Synchronous Area. The FRRa covers fast components of the residual system imbalance, which are modelled by the difference between the residual system imbalances of two consecutive quarter hours, being the volatility of the residual system imbalances. The actual amount of FRRa is determined as the required volume to cover Y% of the volatility of the residual system imbalances. This is shown in the figures below. A more detailed explanation of this methodology can be found in Annex 2. 12 FRRm corresponds to manual activated reserves that can be deployed within 15 minutes such as non-contracted manual bids according to CIPU, R3 production, interruptible industrial clients,… 13 As Elia only procures upward FRRm, the yearly dimensioning only considers the upward FRRm dimensioning. Page 25 of 56 Ancillary services study - horizon 2018 6.1.2. Desaturation of FRRa by FRRm The dimensioning model for the FRRa is based on a theoretical model in which an efficient desaturation of FRRa by FRRm is assumed. During recent years Elia performed increased efforts in order to improve the desaturation of FRRa by FRRm, with satisfying results. An efficient desaturation of FRRa by FRRm allows to achieve a satisfying balancing quality with a relatively limited amount of FRRa, which is scarce in Belgium and has a very high reservation cost (see later). This is illustrated in the figure below, which shows that, although the amount of system imbalances nearly doubled last years, Elia managed to keep the ACE quality under control by performing more activations of FRRm, while the amount of FRRa remained more or less constant. The graphs below shows the absolute energy contents [GWh] of the system imbalance, of the activated FRRa and FRRm volumes and of the ACE for the last 4 years. These graphs show the importance of FRRm to de-saturate and support the FRRa. Preliminary results of the Elia participation to iGCC (only 2 months in 2012), also indicate an enhancement of the ACE quality. Module 1 of iGCC avoids the activation of FRR control power whenever there is an opposite demand for FRR control power in the different participating control areas. Page 26 of 56 Ancillary services study - horizon 2018 The above figures show that, despite an increase of the system imbalance of 515 GWh in 2012 compared to 2009 (+54%!), the increase in ACE was kept limited to 100 GWh due to the increased amount of manual activations. It is expected that the desaturation of FRRa by FRRm will become more challenging in the future, given the steep increase in the required amount of manual activations. Elia therefore continuously investigates efficient ways to de-saturate the FRRa, to increase the liquidity on the FRRm products being able to de-saturate the FRRa on a nearly continuous basis and to increase the liquidity of the FRRa market itself to enhance the balancing quality. 6.2. Inputs used for FRR dimensioning The figure below gives an overview of the considered imbalance drivers used as input to simulate the expected residual system imbalances for the year 2018. It also shows the way these inputs are then combined to determine the expected residual system imbalances. The model used for each imbalance driver and the assumptions taken in the different simulated scenarios (see paragraph 6.3) are further explained in Annex 3. Finally the distribution of the total expected residual system imbalances is used as input in the probabilistic dimensioning model for FRR. The considered imbalance drivers are: Historical observed system imbalances for 2012 as baseline. As imbalances due to forced outages are simulated separately, the 2012 quarter hourly time-series is compensated for periods with observed outages in 2012. Imbalances caused by incremental installed capacity (compared to 2012) of wind and PV production units, being: o imbalances due to forecast errors remaining near real-time; o quarter hourly imbalances due to ramping of PV and wind within the hourly timeframe. Imbalances caused by interconnector (NEMO). forced outages of units and the planned HVDC Ramping imbalances caused by the introduction of an HVDC interconnector in the system. Wind turbines automatically shut down in case the wind speed exceeds their respective cut-off speeds during a certain period of time. This is especially a risk for offshore wind parks given the high wind speeds they face. As a result in case of a storm a high amount of offshore wind production might instantaneously shut down, creating large imbalances in Page 27 of 56 Ancillary services study - horizon 2018 the system. It has to be noted however that the instantaneous outage of all offshore wind parks due to the occurrence of a storm is not considered as an N-1 incident as: storms can be predicted and therefore preventive measures can be taken by BRPs and/or the TSO to reduce the risk. This will be further investigated by Elia during 2013. Elia identified different cut-off thresholds for the different offshore wind parks. Furthermore it is expected that not all wind mills of a wind park simultaneoulsy shut down. This is illustrated by the following figure, which shows the forecast and measurement of the Belgian offshore wind parks on 30/1/2013. new wind turbine technology exists that avoids the instantaneous shutting down of wind parks in case wind speed exceeds the cut-off speed. These wind turbines then gradually reduce their output instead of going down to zero immediately. The occurrence of a storm, and according forecast errors in case of the shutting down of wind mills, is however taken into account in the calculation of the expected residual system imbalance. In the different scenarios, significant improvements are expected for the offshore production forecasts, which in such case also apply for forecast errors due to the shutting down of offshore wind parks due to storms. Elia has only limited experience with the behaviour of offshore wind parks in case of storm and will further investigate this phenomenon in the future to improve the modelling of the storm risk in the dimensioning of the reserves and to take appropriate measures. 6.3. Scenarios The expected residual system imbalances and therefore the required volume of FRRa and FRRm depend on the assumptions taken for the future system evolutions. Therefore different scenarios were simulated, covering different possible evolutions. As already explained above all the simulated scenarios assume significant additional efforts and investments (compared to the current situation) by BRPs and other market parties in order to increase the system flexibility and day-ahead and intraday forecasting quality of load and VRE production. This is required to ensure the sustainable integration of high shares of VRE in the system and to minimize residual system imbalances. Therefore none of the scenarios assume residual imbalances as a result of a structural lack of system flexibility, which is a very strong and optimistic assumption. If residual imbalances are caused by a structural lack of system flexibility, system reserve needs will increase dramatically, which is not taken into account in any of the simulated scenarios. In all of the three simulated scenarios it is assumed that: BRPs invest in highly accurate intraday forecasts of VRE production and off-take to minimize residual system imbalances; BRPs pro-actively foster the development of flexibility in their portfolio (load & generation) and bring this flexibility to the markets (day-ahead, intraday and balancing market). This allows the BRPs to balance at least the expected position of their perimeter in day-ahead. Moreover it is assumed that BRPs actually deploy all available system flexibility in order to minimize residual system imbalances. Therefore no structural residual imbalances due to a lack of (day-ahead) flexibility is taken into account; Page 28 of 56 Ancillary services study - horizon 2018 TSO, DSOs, BRPs and other market parties perform additional efforts to achieve accurate metering and load profiling (DSO responsibility) resulting in a better realtime view on the actual off-take and injection within the system and the BRP perimeter. Adequate incentives by the imbalance mechanism are crucial to achieve this: all imbalance volumes must be exposed to imbalance prices; real-time market reaction on imbalance prices to support the balance of the BRP perimeter and the Belgian system must be incentivized; imbalance prices should be sufficiently high to incentivize investments in –and the actual use of- system flexibility by all market parties. Additional incentives might be required to achieve this as current incentives tend to be inadequate during some time periods. Within the above assumptions, three different scenarios were elaborated to estimate the need for FRRa and FRRm in 2018, reflecting systems in which different amounts of intraday flexibility and flexibility within the hourly timeframe are available to the BRPs. The ‘low reserve needs’ scenario represents a system in which ID flexibility and flexibility within the hourly timeframe is abundant, thereby fully enabling BRPs to adjust their perimeter in the ID time scales according to more accurate ID forecasts. The presence of a liquid 15-minute ID market with short gate closure times (GCT) minimizes the imbalances caused by the ramping of VRE within the hour and by the ramping of NEMO (see also Annex 3). In the ‘high reserve needs’ scenario the ID system flexibility and flexibility within the hour timeframe becomes scarce. This affects the ability of BRPs to balance their perimeter in the intraday timescale, which results in an increase of residual imbalances in the system. Furthermore this affects the ability of BRPs to offset the ramping of VRE within the hour and to offset imbalances due to the ramping of NEMO. The ‘medium reserve needs’ scenario has to be considered as an intermediate one. The assumptions for the different scenarios are shown in the table below: Assumption Development profiling of forecasting, metering Low reserve needs Medium reserve needs High reserve needs High High High High High High High Medium Low and Investments of BRPs in accurate intra-day forecasts of VRE production and off-take. Investments in smart metering and load profiling to have a clear real-time view on the actual off-take, injection and (balancing) position of the system and BRP portfolio. Balancing on day-ahead timeframe Ability of BRPs to balance the day-ahead expected position of their perimeter, including the variable output of VRE, ramping of off-take… This requires sufficient investments in system flexibility to incorporate the high shares of future VRE capacity (in combination with the standard daily ramping of off-take). Balancing on intraday timeframe Ability of BRPs to adjust the position of their perimeter in ID according to more accurate ID forecasts of VRE production and off-take (smart metering,…). This depends on the amount of ID flexibility within the perimeter of the BRP (load and generation) and on the liquidity of ID markets. Page 29 of 56 Ancillary services study - horizon 2018 Intra-hourly balancing. Ability of BRPs to balance the ramping of VRE (wind, PV,…) and HVDC interconnectors within the hour. This depends on: amount of 15-minute flexibility within the BRP perimeter (load and generation); presence of a liquid 15 minutes intra-day market. High Medium Low The following paragraphs give a more detailed explanation on the assumptions and the model parameters for the different assumptions (more details are given in Annex 3). Assumptions taken for the improvement in the (ID) forecasting quality of VRE: It is assumed that the incentives given by the imbalance mechanism are sufficiently strong so that all BRPs, in all scenarios, make use of very accurate ID forecasts. The reference forecast used to model the 2018 incremental wind forecast errors is a day-ahead forecast. Therefore a significant increase in forecasting quality (up to 40%), due to the use of very accurate intra-day forecasts in 2018 is taken into account. The reference forecast used to model PV forecast errors is based on an 8 A.M intraday forecast. A performance increase of 25%, as a result of accuracy improvements towards 2018, was taken into account. Ability of BRPs to balance their portfolio on the basis of the DA expected position of their perimeter: All three scenarios assume that BRPs have sufficient flexibility available (and actually make use of it) to balance (at least) the DA predicted position of their perimeter (ramping of load and VRE,…). This means that no structural (predictable) residual imbalances are introduced in the system due to a lack of system flexibility. This represents a significant change from the current situation in which (balancing) incentives given to the market tend to be insufficient during some periods, resulting in imbalances due to a lack of (activated) system flexibility by market parties. Ability of BRPs to adjust their perimeter according to more accurate ID forecasts: The ability of BRPs to adjust their perimeter in ID based on more accurate ID forecasts depends on the amount of ID flexibility available in the system. ID flexibility refers to flexibility: o within the BRP perimeter, such as flexible generation units, demand side management,…; and o provided by intraday markets and the HUB. The coupling of the NWE intraday market, planned before 2018, might lead to a significant increase in ID market liquidity. The three different scenarios represent systems having different amounts of ID flexibility available. It is assumed that the incentives given by the imbalance mechanism ensure that BRPs also fully use the available ID flexibility to minimize any residual system imbalance. This represents a significant change compared limited liquidity and use of ID markets by potential of more accurate ID forecasts is not in the question whether current price signals are adequate. to the current situation in which the BRPs allows to conclude that the fully exploited yet. This also results given by the imbalance mechanism Ability of BRPs to offset ramping of VRE within the hour and to offset ramping of NEMO: Page 30 of 56 Ancillary services study - horizon 2018 It is observed and expected that less flexible power plants (mainly gas plants in Belgium) are, and will be, available in the system during some time periods. The running hours of these units are decreasing as a result of the increase of VRE production. This results in a reduction of flexibility with a resolution smaller than 1 hour (current market resolution), which affects the ability of BRPs to offset the ramping of VRE within the hour. This might eventually lead to an increase in both the size and volatility of the quarter hourly residual imbalances (although this is currently not observed). The same holds true for ramping of HVDC interconnectors. The introduction of a liquid 15-minutes ID market, such as in Germany and/or investments in 15-minutes flexibility (DSM, VRE,…) are important to solve this issue. The three scenarios represent systems in which different amounts of flexibility within the hourly timeframe are available. In all scenarios it is assumed that the incentives given by the imbalance mechanism are sufficiently strong to incentivize BRPs to make use of all available 15-minute flexibility. 6.4. Resulting FRRa and FRRm needs in 2018 The simulated reserve needs for 2018 represent the reserve needs for the Belgian system and are independent of the volumes pre-contracted by Elia. The Framework Guidelines on Electricity Balancing [11], and the final report on Operational Reserves by ENTSOe [6] confirm that TSOs do not have to pre-contract all required reserves. This is for example the case for systems in which a liquid balancing market exist in which non pre-contracted flexibility is offered to the TSO. This is especially the case for downward FRRm, as in case of the activation of these reserves, there is already an excess of generation in the system. In such case the correct market mechanisms should be in place so that the TSO can require that the BRPs lower their production (or increase the load), at the actual downward regulation cost for the BRPs. This cost is then reflected in the imbalance tariffs to give adequate incentives. The system reserve needs stated below assume a 100%-availability. Equivalent reserve volumes for an availability less than 100%, taking portfolio effects into account, can be calculated. This is performed by taking the availability of the reserves into account in the calculation of the deficit probability of the reserves to cover the expected residual system imbalance (and its volatility). Scenario FRRa [MW] FRRm downward [MW] FRRm upward [MW] 140 695 1120 No ramping 152 1138 107814 With ramping 155 1135 1078 No ramping 159 1231 1131 With ramping 172 1238 1138 No ramping 166 1327 1317 With ramping 192 1331 1321 Up to >1750 MW Up to >1700 MW 2013 reference 2018: low reserve needs scenario 2018: medium reserve needs scenario 2018: high reserve needs scenario Insufficient efforts & investments 14 Up to >300 MW Volume almost equal to N-1 deterministic dimensioning criterion. Page 31 of 56 Ancillary services study - horizon 2018 The table above shows the amount of required FRRa and FRRm for the three different scenarios as well as the 2013 reference. Up to now Elia didn’t notice an impact of the increase in VRE capacity in the Belgium system on the volatility15 of the residual system imbalance. This might however be the case if a significant amount of additional VRE capacity is introduced in the system. Therefore two different cases are simulated for each scenario. In the first case the imbalances due to ramping of VRE and the HVDC interconnector within the hourly timescale are not taken into account (‘No ramping’), while in the second case (‘With ramping’) they are. The figures below show the interpolated ranges for the FRRa and the upward and downward FRRm system needs for the two extreme scenarios (low reserve needs scenario without ramping impact and high reserve needs scenario with ramping impact) between 2013 and 2018. 15 The volatility of the system imbalance is calculated as the delta between the average system imbalance of the actual and the next quarter hour. Page 32 of 56 Ancillary services study - horizon 2018 In case insufficient system flexibility will be available to balance the variable output of the planned additional VRE capacity, or in case of insufficient investments are made in accurate ID forecasts, the reserve needs of the system (and costs for society) will increase significantly and are expected to be much higher than in the above simulated scenarios. The reserve needs of the system heavily depend on the BRP behaviour. In order to avoid a very steep increase in future reserve needs (and costs for society), as indicated by the grey dotted lines in the figures, it is of crucial importance that: BRPs invest in best practice (DA and ID) forecasting of VRE production and off-take in their perimeter; BRPs make active use of markets on all timescales to balance their perimeter; BRPs pro-actively foster the development of flexibility in their portfolio (load & generation) and bring this flexibility to the markets (day-ahead, intraday and balancing market); TSO, DSOs, BRPs and other market parties perform additional efforts to achieve accurate metering and load profiling (DSO responsibility). Adequate incentives are crucial to achieve this: all imbalances should be exposed to imbalance prices; real-time market reaction on imbalance prices to support the balance of the BRP perimeter and the Belgian system must be accepted and incentivized; imbalance prices should be sufficiently high to incentivise investments in –and the actual use of- system flexibility by all market parties. Additional incentives might be required to achieve this; the price signals given by the imbalance mechanism must be sufficiently strong at any point in time to incentivize BRPs to make use of all flexibility to balance their perimeter. This represents a significant change from the current situation in which it is observed that the incentives by the imbalance mechanism are not always sufficiently strong to achieve this. 6.5. FRR dimensioning conclusions The sustainable integration of the planned capacity of VRE in combination, resulting in a limited increase in system reserves needs (and according costs for society), requires: o additional investments in more accurate forecasts of load and real-time measurements of load (smart metering) and VRE production, both in dayahead and intraday timescales; o sufficient investments in system flexibility by BRPs and market parties (maximizing flexibility of current generation and load, investments in new flexible power plants,…) to balance the VRE output. System reserve needs (and costs) will explode in case no accurate forecasts or insufficient system flexibility are available (or are not being used) by BRPs to balance the variable output of PV and wind production within their perimeter. Gradual strong price signals by the imbalance mechanism, representing the market cost for deployed balancing energy, in case of residual system imbalances must give proper incentives to all involved market parties to invest in accurate ID forecasts of load and VRE and to develop the required system flexibility. The ability of BRPs to adjust their perimeter according to more accurate ID forecasts depends strongly on the amount of ID flexibility within the system. In case of the low reserve needs scenario, ID flexibility is abundant resulting in a rather limited increase of reserve needs in the system. The high reserve needs scenario assumes scarcity of ID flexibility in the system, resulting in a significant increase of reserve needs. This shows again the importance of adequate incentives by the imbalance mechanism for market parties to develop and use ID system flexibility. Page 33 of 56 Ancillary services study - horizon 2018 The need for FRRa will increase towards 2018. This is mainly due to an expected increase in volatility of residual imbalances because of the ramping of VRE within the hour and the ramping of the planned HVDC interconnector between Belgium and UK. The increase of FRRa therefore depends on the amount of quarter hourly flexibility available in the system. This stresses the importance of o sufficient investments in quarter hourly flexibility in the BRP perimeter (load and generation); o the development of liquid ID markets with a time resolution of 15-minutes and GCT near real-time for the cost-effective integration of high shares of VRE and HVDC interconnectors in the system. This effect has to be closely followed by the TSO during next years. All performed simulations assume an efficient desaturation of FRRa by FRRm. This can only be achieved in case of the presence of a very flexible and liquid balancing market. This will become more challenging in the future, as Elia already experienced during recent years. The reason for this is the high number of FRRm activations required, as a result of the increase in amplitude and frequency of large residual system imbalances. As a result two different functions for FRRm are identified: o the first function of FRRm is to cover rare but very large imbalances due to instantaneous incidents or very large (but rare) forecast errors. o the second one being the de-saturation of the activated FRRa on a more or less continuous basis. Whereas historically the FRRm was mainly used to fulfil the first function, thereby acting as a sort of ‘contingency’ reserve, Elia now seeks additional volumes of the FRRm fulfilling the second function, by exploiting different possibilities. The FRRm reserve resources for the different functions may have different characteristics, e.g. in activation frequency and duration, and may therefore be provided by different types of resources. One of the ways explored by Elia to capture additional FRRm able to de-saturate the activated FRRa is the creation of a bid-ladder to capture more flexibility in the system, thereby enabling all resources, such as biomass, wind production, CHPs, demand,…to offer this kind of flexibility. The need for upward FRRm shows an increase in the high reserve needs scenario and even a very limited decrease in the low reserve needs scenario. This can be explained by the fact that: o the 2012 baseline, used to simulate the 2018 needs, showed a significant decrease in negative residual system imbalances compared to 2011, which was used as baseline for the calculation of the 2013 needs; o there is a collateralization of the increased needs for upward manual FRR due to additional VRE capacity and the positive FRR required to cover the N-1; o the major increase in forecasting quality and ID flexibility in the low reserve needs scenario compared to the 2013 situation. The need for downward manual FRR increases significantly due to: o the integration of additional VRE capacity; o the introduction of a large positive N-1 incident as result of an outage of the 1000 MW HVDC interconnector between Belgium and UK in full export mode; o the 2012 baseline, used to simulate the 2018 needs, showed a significant increase in positive residual system imbalances compared to 2011 (used as baseline for the 2013 needs). The development of sufficient downward flexibility in the system represents a major challenge for the next years. Page 34 of 56 Ancillary services study - horizon 2018 Elia already observes residual positive system imbalances caused by insufficient downward flexibility. This shows again the importance of adequate market incentives to develop additional downward flexibility. The increase in downward flexibility has to be achieved by reducing the minimum stable power of existing power plants (e.g. transformation of CCGT to OCGT), by requiring RES units to offer downward flexibility, by new investments in downward flexibility,... This is elaborated further in Chapter 8. Elia will perform efforts in order to avoid a systematic long position of the residual system imbalances as observed in 2012. This would result in a small reduction of the increase of the downward FRRm needs and a small increase of the upward FRRm needs, although the general conclusions of this study remain valid 5. 7 Replacement Reserves Replacement Reserves (RR) are the slower manually activated reserves meant to (partially) replace the activated FRR after 15 minutes. As authorised in the future NC LFC&R [2], Elia does not currently contract RR. Indeed, in the Belgian system, market parties are authorised and incentivised to restore the balance of their perimeter up to real-time. Therefore market parties fulfil the replacement role in the Belgian system. Provided this fundamental feature of the Belgian balancing market design is preserved, the contracting of RR is not considered in the Belgian market design. Page 35 of 56 Ancillary services study - horizon 2018 Page 36 of 56 Ancillary services study - horizon 2018 8 Reserve resources available in the system This chapter analyses whether the reserve resources that are reasonably expected to be available in the system in 2018 are sufficient to cover the 2018 reserve needs for the different simulated scenarios. In addition some qualitative insights are given in required evolutions to enable the economic efficient procurement of the required reserves in 2018. In this perspective it is investigated whether the reserves portfolio will be sufficiently diversified to avoid excessive exposure to specific market conditions for the procurement of reserves. This study only considers reserve resources available in the Belgian system. Cross-border reserve resources are not taken into account given the uncertainty of cross-border exchange of reserves and the relatively long time horizon until 2018. The development of cross-border exchange of reserves will however become very important in the future and will be considered by Elia. The planned increase in interconnection capacity might also increase the potential for cross-border exchange of reserves. As already mentioned in paragraph 3.3 there is a high degree of uncertainty regarding the current reserve resources that will still be available in 2018. This depends e.g. on the evolution of economic parameters such as fuel and electricity prices,… for power plants, on profitability of industrial demand facilities,… This is shown e.g. by the current declared intentions of decommissioning or mothballing of some gas fired plants in Belgium due to reduced profitability. Furthermore there is a link with possible new investments required to ensure system adequacy in the future. For the avoidance of doubt, this study is not addressing the issue of security of supply and generation adequacy. It is strictly looking at the specific reserves to be secured by Elia in the context of its responsibility to ensure the availability of appropriate ancillary services, as defined by the Electricity Law, (art. 8 §1) and the Federal Grid Code (art. 231 and 232). The results in this section are based on an extrapolation of the current trends and evolutions, but can change significantly in function of the evolution of economic parameters. 8.1. Flexibility High system flexibility is the key requisite for the integration of large shares of VRE in the system. The results of this study point out the high ramping capabilities of wind and PV, which have to be covered by flexible resources together with flexibility on the DA and ID electricity markets. A lack of (activated) system flexibility generally results in structural and very large residual imbalances. Although these imbalances can be predicted, they are not resolved by the BRPs. The reserve dimensioning however only accounts for residual imbalances due to unpredictable or partially predictable events. The creation of a highly flexible system requires following continued efforts: new units must have high ramping capabilities, very low minimum stable power and the ability to start and stop quickly and frequently. Furthermore new units with low marginal production costs must be able to provide spinning flexibility to the system; development of DSM system flexibility; the right incentives must be in place in order to require RES units themselves to provide system flexibility; set incentives for BRPs to offer (and actually use) flexibility by giving adequate price signals for imbalances; Maximize the flexibility of the existing centralized production assets and demand facilities; and Set the framework for the creation of a coupled, liquid ID market with GCT near real-time and a 15-minute resolution. The results of the study show that the reserve needs of the system are highly dependent of the system flexibility and on the incentives for BRPs to balance their portfolio on a 15Page 37 of 56 Ancillary services study - horizon 2018 minute level. The results of the study performed here are only valid for a system without structural flexibility issues. 8.2. Available reserve resources in 2018 The question whether sufficient reserves resources will be available in the system in 2018 goes further than having sufficient reserve capacity [in MW], from a system capability point of view, available on generating units and demand facilities within the grid. A diversified reserve portfolio for each reserve type is crucial for the creation of a liquid and sustainable reserves market and to ensure their economically efficient procurement. In an efficient reserve market the spinning reserves are provided either by demand facilities and/or by power plants that are selected in the merit order. These can be renewables, CCGT units, coal units, CHPs, biomass units,… depending on the actual market conditions. The most efficient power plants provide downward reserve capacity while the least efficient ones provide upward reserve capacity. Diversification of the reserves portfolio can be achieved by: Creating the attractive framework for RES to provide reserves (smart support schemes, reserve products,…); Activating demand side management to provide reserves; Attracting new players and new reserve resources in the systems; and Requiring new/refurbished power plants to have FCR, FRRa and FRRm capability. Elia currently observes a highly concentrated, illiquid market for both FCR and FRRa, as these reserves (especially FRRa) are almost only offered by CCGT units for which running hours are decreasing under current market conditions. The procurement of FCR and FRRa on CCGT units therefore introduces high must run costs in the system as costs fully depend on the evolution of the clean spark spread. Even though many changes were already introduced in the market to facilitate diversification of reserve resources and that the technical possibility for several resources to offer reserves already exists (no R&D required), a continued increased effort of all parties is required to diversify the reserve resources and to ensure a well-functioning, liquid reserves market. 8.2.1. FCR The estimated range for FCR in 2018 is [95 – 110 MW]. The following resources actually provide FCR and are expected to be still available in the system in 2018: FCR resource CCGT units Resources still available in 2018 The issue of adequacy –and whether additional support for flexible capacity is required- is currently being addressed by the Belgian Government. It is assumed that at least 3 existing CCGT units will be still available in the system in 2018, despite their current observed reduced profitability and transformation of some CCGT units towards OCGT peak plants. It has to be emphasized that the above assumption does not imply that 3 CCGT units are sufficient to ensure system adequacy. It is however considered that the probability of having at least 3 CCGT units in the system is very high. Nuclear The planned decommissioning of part of the nuclear capacity towards 2018 is considered to have a negligible impact on the FCR providing capability of the remaining nuclear production park. Load Successful pilot project of delivery of FCR by demand facilities was launched in 2013. Pump-storage hydro plants Still available in the system in 2018 and capable of delivering a significant amount of FCR, however only in turbine mode and subject to energy constraints. Page 38 of 56 Ancillary services study - horizon 2018 For 2013 Elia procures part of its FCR in France. The cross-border procurement of FCR is less complex than for other reserves given the full harmonization of the product on European level. As already stated before this study however does not consider crossborder resources given the related uncertainty. From the point of view of system capability for FCR it is expected that the current reserve resources, which are expected to be still available in 2018, will allow to cover the 2018 reserve needs. In the above table it is assumed that the current providers of R1 load –mainly big industries- will still be operational within the Belgian control area in 2018, which also depends from external factors (economy,…). From an economic point of view though, as already observed today, these reserve resources do not allow an efficient procurement of the reserves, as a large share of the capability is concentrated on CCGT units. Therefore procurement costs for FCR (spinning reserves) are highly dependent of the CCGT market conditions. These market conditions are mainly determined by the evolution of the clean spark spread. Spinning reserves such as FCR have to be provided by units that are selected in the merit order, or by demand facilities, to avoid high must run costs. These units can be wind parks, CCGT units, coal units, CHPs, biomass units… depending on the market situation. The most efficient power plants provide downward reserve capacity while the least efficient ones provide the upward capacity. This is the main reason why procurement costs for FCR significantly increased during recent years as market conditions for CCGT unit worsened. Therefore it is of key importance that some of the following FCR resources are developed in the future: FCR resource To be developed towards 2018 CHPs CHP units have a competitive advantage compared to gas units given the additional value for the underlying (heat) process. Therefore FCR provision by CHPs decreases must run costs compared to CCGT units. New production capacity & refurbished units New centralized production units, required to ensure system adequacy, must have FCR capability in order to further diversify the FCR market and to ensure continuity in the FCR capability of the Belgian production park. Refurbished power plants need to have FCR capability where technically possible. Load The participation of load in the FCR market has to be developed further as this might avoid high must run costs to be paid to conventional generation units to provide FCR. RES units During periods of high RES infeed, less (or none) conventional spinning units will be available in the system to provide FCR. FCR then has to be delivered by RES units (biomass, wind,…) or load in order to avoid high must run costs. Smart support schemes are required to foster participation of RES units (biomass, PV, wind units) in reserves markets, which is not always the case for the current support schemes. It is important that BRPs and market parties are incentivized to develop these new reserve resources. Smart support schemes for RES and CHP units must enable the efficient participation of RES in the ancillary services market. 8.2.2. FRRa The estimated range for FRRa in 2018 is [152 – 192 MW]. The following resources actually provide FRRa and are expected to be still available in the system in 2018: Page 39 of 56 Ancillary services study - horizon 2018 FRRa resource CCGT units Resources still available in 2018 The issue of adequacy –and whether additional support for flexible capacity is required- is currently being addressed by the Government. It is assumed that at least 3 existing CCGT units (~200 MW of FRRa capacity) will be still available in the system in 2018, despite their current observed reduced profitability and transformation of some CCGT units towards OCGT peak plants. It has to be emphasized that the above assumption does not imply that 3 CCGT units are sufficient to ensure system adequacy. It is however considered that the probability of having at least 3 CCGT units in the system is very high. Pump storage units The current pump-storage units are able to deliver all the required FRRa, however only in turbine mode and subject to energy constraints. CHP units Currently only accounts for a marginal participation in the FRRa market. The cross-border procurement of FRRa is very complex given the variety of FRRa products and market designs across Europe. Furthermore the cross-border procurement of FRRa requires cross-border transmission capacity to be continuously available. This study therefore doesn’t consider the cross-border procurement FRRa. From the point of view of system capability for FRRa it is expected that the current reserve resources, which are expected to be still available in 2018, will allow to cover the 2018 reserve needs for both the low and high reserve needs scenario (resp. 152 and 192 MW). It has to be emphasized however that, in case no further FRRa resources are developed towards 2018, the presence of 3 CCGT units for the delivery of FRRa in the high reserve needs scenario results in very limited margins. The unavailability of one or more CCGT units will then affect the ability of the TSO to procure the required FRRa volumes. Also from an economic point of view, as already observed today, these reserve resources do not allow an efficient procurement of the reserves, as a large share of the capability is concentrated on CCGT units. Therefore procurement costs for FRRa are fully dependent of the CCGT market conditions, related to the evolution of the clean spark spread. FRRa has to be provided by units that are selected in the merit order, or by demand facilities, to avoid high must run costs. These units can be wind parks, CCGT units, coal units, CHPs, biomass units, pump storage… depending on the market situation. The most efficient power plants provide downward reserve capacity while the least efficient ones provide the upward capacity. Procurement of spinning reserves on units that aren’t selected in the merit order leads to high must run costs and must be avoided. The procurement costs for FRRa therefore depend on the evolution of the clean spark spread. Therefore it is of key importance that some of the following FRRa resources are developed in the future: FRRa resource To be developed towards 2018 CHPs CHP units have a competitive advantage compared to gas units given their value for the underlying (heat) process. Therefore FRRa provision by CHPs results in less must run costs compared to CCGT units. New production capacity and refurbished units New investments required to ensure system adequacy must have FRRa capability in order to further diversify the FRRa market and to ensure continuity in the FRRa capability of the Belgian production park. Refurbished units need to have FRRa capability where technically possible. Further investigation of operating modes for pump storage units to enable an economic more efficient delivery of FRRa. Pump storage units Page 40 of 56 Ancillary services study - horizon 2018 Load The participation of demand side in the FRRa market should be investigated, knowing however that it is likely to be more difficult than for FRRm or FCR. RES units During periods of high RES infeed, no (or few) conventional spinning units will be available in the system to provide FRRa. FRRa then has to be delivered by RES units (biomass, wind,…) or load in order to avoid high must run costs. Smart support schemes are required to foster participation of RES units (biomass, PV, wind units) in reserves markets, which is not the case for the current support schemes. 8.2.3. Upward FRRm The estimated range for upward manual FRR in 2018 is [1078 – 1321 MW]. Currently following resources in the Belgian control area are providing the majority of the upward FRRm: Turbojets; Gas turbines; Interruptible industrial clients (single and aggregated units) connected to the TSO grid. At the end of 2012 about 850 MW of power plants (gas turbines and turbojets) was qualified to deliver FRRm (400 MW contracted), while about 350 MW of FRRm was offered by interruptible industrial clients (261 MW contracted for balancing purposes). The following resources actually provide upward FRRm and are expected to be still available in the system in 2018: Upward manual FRR resource R3 production units Interruptible industrial clients Resources still available in 2018 640 MW mainly on OCGT units under the assumptions that: o turbojets (210 MW) will be decommissioned; o loss of FRRm capability due to the decommissioning of some older GTs will be offset by the transformation of CCGTs to OCGTs. 350 MW manual upward FRRm capability is assumed to remain available in the system. Elia relies also on non-contracted incremental bids in the framework of the CIPU contract16 [8] and on mutual support contracts with neighbouring TSOs. In 2013 Elia launched a pilot project to gather experience in the participation of aggregated DSO connected load in the FRRm market. A bid ladder project for FRRm was started to attract more FRRm bids compared to the current CIPU framework. From the point of view of system capability, for upward FRRm it is expected that the current reserve resources which are expected to be still available in 2018, will be insufficient to cover the reserves in both the low and high reserve needs scenarios. This is because of the combination of increasing needs for upward FRRm, in the high reserve needs scenario, and the assumed decommissioning of the turbojets (210 MW upward FRRm). Depending on the scenario there is a need to develop up to 360 MW of additional upward FRRm capability. In the above table it is assumed that the interruptible industrial clients –mainly big industries- will still be operational within the Belgian control area in 2018, which also depends from external factors (economy,…). In addition, from an economic point of view, a diversification of the resources is required in order to avoid excessive dependency on OCGT units and therefore on gas and electricity prices for the procurement of upward FRRm. 16 All power plants exceeding 75 MW are obliged to offer the remaining flexibility margins to Elia. Page 41 of 56 Ancillary services study - horizon 2018 The table below gives an overview of the different options for the development of 360 MW of upward FRRm and for the further diversification of upward FRRm resources towards 2018: Upward manual FRR resources To be developed towards 2018 Gas turbines Conversion of CCGT units to OCGT units (in addition to the already assumed conversions to offset the loss of upward FRRm capability due to the decommissioning of some older OCGTs). Load Upward FRRm can be delivered relatively easy by demand facilities (reduction of offtake) Cross-border cooperation Sharing17 of up to neighbouring TSOs. New production capacity & refurbished units New centralized production units must have upward FRRm capability in order to: RES units 300 MW of upward FRRm with o diversify the FRRm market; o ensure continuity in the FRRm capability of the Belgian production park. Refurbished units need to have upward FRRm capability where technically possible. Not considered to offer upward FRRm flexibility due to environmental aspects and low marginal production costs (high de-rating costs). The upward FRRm has two functions: The continuous de-saturation of the activated FRRa, requiring upward FRRm by: o Flexible spinning resources (biomass, RES, CHPs, CCGTs/OCGTs, pump storage units,…) or very flexible demand in terms of activation frequency; o Very flexible non-spinning units, in terms of activation frequency, such as pump storage units,… Cover rare but very large imbalances (typically outages of power units), requiring FRRm by less flexible units in terms of activation frequency such as the interruption of demand facilities (in case of a limited number of activations per year),… The required new upward FRRm capability has to be created by: participation of load and new OCGTs (transformed CCGTs) in the upward FRRm market; cross-border procurement of FRRm and sharing agreements with neighbouring TSOs for FRRm capacity; participation of other units (biomass, RES, CHPs,…) in the FRRm market. 8.2.4. Downward FRRm The estimated range for downward FRRm in 2018 is [1138 - 1348 MW], representing a significant increase of +/- 700 MW compared to 2013. During recent years Elia noticed a significant increase in the need for downward reserves, driven by the increase of VRE in the system in combination with a lack of downward system flexibility and/or insufficient incentives for BRPs to access the available downward flexibility. Downward FRRm should in principle not be pre-contracted, as the need indicates an excess of generation in the system. A framework in which downward flexibility is offered by the available generation units at a price reflecting their marginal (activation) costs for downward regulation power, or by demand for increasing consumption, is more sustainable on the long term and is required to set proper incentives for investments in downward 17 Concept introduced in [2] and [6] allowing TSOs to share part of their FRRm. Page 42 of 56 Ancillary services study - horizon 2018 flexibility. This is in line with the actual framework for offering downward FRRm in accordance with the CIPU contract. The imbalance tariffs then reflect these price signals (single marginal pricing), incentivizing BRPs and market players to invest in sufficient downward flexibility, to balance their portfolio on a 15-minute basis and to support the balance of the control area. In case of absence of sufficient downward flexibility offered by market players, additional incentives in the imbalance tariffs might be needed as an intermediate measure to incentivize the development –and use of- downward flexibility. Downward manual FRRm flexibility has to be further developed towards 2018. The following table gives an overview of the different resources to be developed: Downward FRRm resources New production capacity and refurbished units To be developed towards 2018 New centralized production units must have very low stable minimum power, high ramping capabilities and have the capability to start and stop quickly and frequently in order to be able to offer sufficient downward flexibility. Refurbished units need to have downward FRRm capability where technically possible. Existing power plants Increased efforts to reduce the minimum stable power of the units are required. RES units Participation of RES in downward FRRm is required to incorporate the massive increase of RES in the system. Smart support schemes are required to foster the participation of RES units (biomass, PV, wind units) in downward FRRm. Cross-border cooperation Sharing agreements17 of downward FRRm in order to compensate the instantaneous loss of an HVDC interconnector in full export have to be envisaged (300 MW max). Load The potential for load is uncertain as this would require an increase in off-take, which is, at first sight, less straightforward than the reduction of off-take as to deliver upward FRRm. The sustainable integration of increasing shares of VRE in the system is only possible in case sufficient downward flexibility is available in the system. Currently Elia observes insufficient (activated) available downward flexibility in the system during some periods. Therefore it is very important to give adequate price signals to BRPs and market players reflecting the actual cost of – and need for – downward flexibility. Furthermore the same conclusions as for upward FRRm on its different functions, being desaturating the activated FRRa and covering large incidents, hold for downward FRRm. Page 43 of 56 Ancillary services study - horizon 2018 Page 44 of 56 Ancillary services study - horizon 2018 9 Conclusions This study is not addressing the issue of security of supply and generation adequacy. It is strictly looking at the specific reserves to be secured by Elia in the context of its responsibility to ensure the availability of appropriate ancillary services, as defined by the Electricity Law, (art. 8 §1) and the Federal Grid Code (art. 231 and 232). Elia resolves residual system imbalances in the system while BRPs are responsible for balancing their perimeter on a 15-minute interval up to real-time and are incentivised to do so. The Belgian market design allows for –and incentivizes- real-time reaction of market parties (with physical position) on the imbalance prices, thereby reducing residual system imbalances. This market design is of crucial importance to enable the sustainable integration of high shares of VRE in the system and to foster the required investments in system flexibility and accurate intraday forecasting of load (smart metering) and VRE production by all actors. The system reserve needs depend on the amount of residual system imbalances and are therefore highly dependent on the BRP behaviour to balance their perimeter on a 15minute basis. This study takes a strong assumption that BRPs and other market parties will perform significant additional efforts and investments to increase the system flexibility, thereby enabling them to balance the increasing shares of VRE in their perimeter, in addition to significant investments in highly accurate intraday forecasts of VRE production and system off-take. As such residual system imbalances are minimized and no residual system imbalances due to a structural lack of system flexibility are therefore considered in this study. Adequate and efficient incentives for BRPs to be balanced on a 15-minute basis are of crucial importance to achieve such a situation, which means that quarter-hourly imbalance prices must be carefully monitored and their calculation further tuned in the future. Furthermore it is important that all imbalances are exposed to the imbalance mechanism. In case such required efforts and investment by market parties in terms of forecasting, flexibility and market participation are not made (for instance due to lack of incentives), system reserve needs (and according costs) will rise dramatically as residual system imbalance will increase. This falls beyond the scope of this study and is therefore not considered. The study shows that a high amount of intraday flexibility and of system flexibility on a 15minutes timescale is very important to minimize the increase in system reserve needs. Whereas there is only a slight increase expected for FCR, depending on the scenario the needs for FRRa, upward and downward FRRm increase significantly towards 2018. Especially the increase in downward FRRm is very high, in all simulated scenarios, due to the introduction of a large positive dimensioning incident, being the outage of the planned 1000 MW HVDC interconnector between Belgium and UK (NEMO) in export mode, and due to increased forecast errors of VRE. The study assumes a very efficient de-saturation of FRRa by FRRm. This will require very flexible FRRm that can be activated on an almost continuous basis. In addition there is a need for FRRm required to cover rare, but very large, residual imbalances caused by outages,… which can be less flexible. Under the assumption that at least 3 of the existing CCGT units will still be available in the system in 2018, existing FCR and FRRa resources will probably allow to cover the FCR and FRRa needs in all scenarios, although margins are decreasing and become very narrow. Significant investments in both upward and especially downward FRRm are required towards 2018 to cover the simulated reserve needs. The currently high concentration of FCR, FRRa (and in lesser extent FRRm) resources on a single technology (CCGTs, OCGTs) hinders the economically efficient procurement of reserves. Therefore a further diversification of reserve resources towards participation of existing and new power plants, RES (wind, biomass, CHPs) and load is of key importance to facilitate the market and to enable a more efficient procurement of reserves. This requires not only investments in the technical capability, but also might require changes in the regulatory framework, such as smart support schemes for RES and CHP units to participate in the ancillary services market, and changes in the FCR, FRRa and FRRm market design. Page 45 of 56 Ancillary services study - horizon 2018 References [1] CREG decision (B)120621-CDC-1162 issued on the 21th of June 2012: www.creg.be [2] Draft of NC LFC&R published by ENTSOe on 1/2/2013 for public consultation: https://www.entsoe.eu/major-projects/network-code-development/load-frequencycontrol-reserves/ [3] Further information on the Elia balancing mechanism is available on: http://www.elia.be/en/products-and-services/product-sheets#balance [4] Belgian Federal Grid Code: http://www.ejustice.just.fgov.be/cgi_loi/loi_a.pl?language=nl&caller=list&cn=2002 121942&la=n&fromtab=wet&sql=dt='koninklijk%20besluit'&tri=dd+as+rank&rech =1&numero=1 [5] ‘Harnessing Variable Renewables – A guide to the balancing challenge’ by IEA, 2011, 228 pg [6] ENTSOe Final Report on Operational Reserves: https://www.entsoe.eu/major-projects/network-code-development/load-frequencycontrol-reserves/ [7] ENTSOe Operational Handbook Policy 1 on Load-Frequency Control: https://www.entsoe.eu/publications/system-operations-reports/operationhandbook/ [8] CIPU contract: http://www.elia.be/en/products-and-services/~/media/files/Elia/Products-andservices/ProductSheets/S-Ondersteuning-net/S5_F_CIPU_08_07.pdf [9] Based on publically available operational data of BritNed cable: http://energieinfo.tennet.org/Connection/NominatedCapacityTSO.aspx [10] Based on publically available PV operational data of Amprion: http://www.amprion.de [11] Final Framework Guidelines on Electricity Balancing, issued by ACER on the 20 th of September 2012: http://www.acer.europa.eu/Electricity/FG_and_network_codes/Pages/Balancing.as px Page 46 of 56 Ancillary services study - horizon 2018 Annex 1: Extract from CREG decision CREG formulated following request in the decision on the proposal of Elia for the reserve volumes for the year 2013 [1], which was introduced by CREG on the basis of Article 233 of the Federal Grid Code. 40. La détermination des volumes nécessaires à ELIA se heurte de plus en plus à un manque de ressources disponibles sur le territoire belge. Cette constatation se conjugue avec le souci des producteurs de valoriser leurs unités de production, ce qui peut mener dans certains cas à des décisions de fermeture de certaines d’entre elles si leur maintien en activité n’est plus rentable. Au-delà de la sécurité d’approvisionnement vue sous l’angle de l’adéquation, la Belgique se dirige donc vraisemblablement vers des besoins importants en nouveaux moyens de réglage et la couverture de ces besoins ne peut se baser sur des décisions à court terme. Dans une perspective d’anticipation des problèmes liés au déficit en moyens de réglage, il paraît donc nécessaire d’étudier l’évolution à moyen terme des besoins de réglage. Dans cette mesure, la CREG demande à ELIA de réaliser avant la fin de 2012 une étude d’évaluation des besoins en moyens de réglage à un terme de cinq ans (fin 2017-début 2018), et de détermination, pour chaque type de réserve, du volume des ressources que l’on peut dès à présent raisonnablement prévoir comme disponibles à ce terme pour couvrir les besoins de la zone belge. Elle demande également à ELIA de publier sur son site les résultats de cette étude avant la fin de l’année 2012. Elia and CREG informally agreed to postpone this study to Q1 of 2013. Page 47 of 56 Ancillary services study - horizon 2018 Annex 2: FRR dimensioning methodology This section gives a high level explanation of the methodology for the dimensioning of the reserves. The methodology used is in line with the one used for the determination of the reserves volumes for 2012 and 2013, which was approved by CREG resp. in its decisions (B)110519-CDC-1056 and (B)120621-CDC-1162. The dimensioning of the total combined amount of FRRa and FRRm is performed on the basis of the expected future residual system imbalance. The first step therefore is to determine the expected amount of residual system imbalances in 2018. This is performed by combining all the expected residual system imbalance drivers for 2018, which are described in detail in Annex 3. Different assumptions are taken for the evolution of these imbalance drivers towards 2018 for each of the simulated scenarios. The FRR dimensioning methodology is based on the following operational aspects: the combination of FRRa and FRRm is used to cover the quarter hourly expected residual system imbalances; the FRRa is used to cover the fast imbalances, which in this model are simulated by the delta between consecutive quarter hourly expected imbalances (volatility of the residual system imbalance); and the FRRm de-saturates the FRRa to prepare the FRRa for further imbalances and supports the FRRa in case of large imbalances. The total required amount of upward (downward) FRR – i.e. the combined volume of FRRa together with FRRm - is defined as the amount required to cover the expected future quarter hourly negative (positive) residual imbalances during 99,9% of time, thereby allowing a deficit probability of 0,1% (8,7 hours/year) for both positive and negative residual imbalances. The figure below shows this for an arbitrary case. As required by ENTSOe the N-1 incident acts as a strict minimum for the FRR dimensioning. The required amount of FRRa is determined as the amount of FRRa that is required to cover X% of the fast imbalances. These fast imbalances are modelled by the delta between consecutive quarter hourly expected residual imbalances. This delta is also called the quarter hourly volatility of the residual system imbalance. This is shown in the picture below: Page 48 of 56 Ancillary services study - horizon 2018 The value of X%, being the deficit probability, is chosen based on historical observed deficit probabilities for FRRa to cover the delta between the residual system imbalances of consecutive quarter hours together with the observed ACE quality. Based on historical observations, Elia defines the deficit probability in order to have a satisfactory ACE quality, according to ENTSOe prescriptions. The monotonous diagram of the volatility of the residual system imbalances allows to determine the required amount of FRRa. In a last stage the amount of FRRm is defined as the difference between the total required amount of FRR and the amount of FRRa, as shown in the picture below. Due to the uncertainty on the specifications and availability of the future reserve products no link is made between the availability of the reserves products and the deficit probability of covering the imbalances. This means that the resulting volumes are expected to be available during 100% of the time. In the annual calculation of the required reserve volumes, Elia is able to calculate equivalent reserve volumes with an availability of less than 100% of time [1]. Page 49 of 56 Ancillary services study - horizon 2018 Annex 3: Imbalance drivers This annex provides a detailed overview of the different imbalance drives that were used as input to define the required amount of FRRa and FRRm for 2018 –as indicated in Annex 2- along with the different assumptions that were taken in the different simulated scenarios. In order to simulate future expected imbalances, TSOs typically use historical imbalances of the previous year, to which all additional expected imbalances due to future system evolutions, such as the increase in VRE capacity or the integration of an HVDC interconnector, are added. The figure below gives an overview of the different imbalance drivers and shows the way they are combined to become the estimated residual system imbalance for 2018. All imbalance drivers – except the forced outages - are modelled on the basis of known timeseries for 2012, in which case they are combined by summing the time-series as to preserve the effects of mutual correlation. The sum of these imbalance drivers is then transformed to a probability distribution, which is convoluted with the probability distribution of expected imbalances due to forced outages of units and the planned BE – UK HVDC interconnector (NEMO). A convolution doesn’t account for any correlation between the distributions, which is justified in this case as the occurrence of forced outages and the occurrence of other imbalance drivers is uncorrelated. Historical imbalances The model uses the observed system imbalances of 2012 as a baseline, to which additional imbalances due to: incremental (compared to 2012) installed capacity of wind or PV production; future HVDC Interconnectors; forced outages of units; … Page 50 of 56 Ancillary services study - horizon 2018 are added in order to obtain the expected residual system imbalance for 2018. Periods during which forced outages occurred in 2012 were deleted from the quarter hourly time-series of the 2012 residual system imbalance, as a separate statistical model to simulate imbalances caused by outages is incorporated to estimate the required amount of reserves. This is required as outages only occur rarely, so that the events gathered in one year do not necessarily represent all possible combinations of outages. The figure below gives the distribution of the 2012 quarter hourly residual system imbalances (without forced outages): The residual system imbalances (absolute) energy slightly decreased in 2012 – for the first time in four years – compared to the 2011 case. This is assumed to be the result of better incentives given by the new imbalance mechanism, as well as due to increased efforts in monitoring and communication to BRPs. It has to be noted however that in 2012 much more positive than negative system imbalances occurred. This evolution towards positive imbalances is observed since a few years and is caused by the increase of RES units in the system. Continued efforts of BRPs to improve their forecasting quality of load, VRE and better price incentives by the new imbalance mechanisms might eventually even reduce the baseline residual system imbalances. This is taken into account by a ‘correction’ factor for each scenario that is simulated. In the low reserve needs scenario it is expected that, due to better incentives and increased forecast quality of both load, and the VRE already present in the residual system imbalance of 2012, the baseline will decrease further in the future. The quarter hourly time-serie of the residual system imbalances of the baseline is therefore reduced by 5%. In the high scenario it is expected that, although balancing incentives and forecasting quality improves, the baseline will remain constant as insufficient system flexibility is expected to be available to realize the potential of increased forecasting quality. The following table summarizes the assumptions for the baseline in the different scenarios: Scenario Assumed change quarter hourly residual system imbalances of the baseline Low -5% Medium -3% High 0% HVDC interconnector The planned 1000 MW HVDC interconnector between Elia Belgium and National Grid UK is expected to be commissioned in 2017 – 2018. The integration of such an interconnector in the system will introduce some additional residual imbalances into the system because of: Outages: in contrast to the loss of a line or cable within a synchronous system, which has no direct impact on the system operation due to the N-1 security, the Page 51 of 56 Ancillary services study - horizon 2018 loss of an HVDC interconnector between synchronous areas introduces a physical imbalance in the affected areas. The imbalance can be either positive, in case of the loss of the cable in export conditions, or negative in case of the loss of the cable in importing conditions. These imbalances are treated in the probabilistic model for outages, as explained further in this annex. Ramping: in most cases the interconnected TSOs agree on a maximum ramping rate for the HVDC cable. A ramping rate constraint of 100 MW/min is a commonly used value. The ramping is concentrated symmetrically on the change of the quarter hour. It is expected that the ramping of the HVDC interconnector will introduce additional imbalances into the system. The figure below shows the origin of imbalance caused by ramping in case hourly energy blocks are exchanged over the HVDC cable. This is illustrated by the blue line, which shows the ramping in case of an hourly schedule change from 500 MW import to 300 MW export. The simultaneous upward or downward ramping of resources within the control area will partly offset these ramping effects. In order to simulate this behaviour, 2012 hourly operational data of an interconnector between UK and RG CE was used [9]. The imbalances were calculated as the quarter hourly difference between the scheduled energy block and the realized flow due to ramping. Different assumptions are taken for each of the different scenarios for the size imbalances that actually would occur. Scenario Size of residual imbalances due to HVDC ramping [as % of theoretical ramping imbalances] Low 10% Medium 30% High 50% The market design for the HVDC interconnector plays an important role on the system operation. Situations in which a BRP compensates the ramping on an quarter-hourly energy basis might lead to an overshoot (or undershoot) just before or after the ramping period itself (shown by the red line in the figure above). Outages of units and HVDC interconnector The residual imbalances due to outages of large units and HVDC interconnectors are simulated separately because of: their low probability of occurrence: the baseline of 2012 doesn’t take all possible combinations of outages into account; and imbalances due to new planned units or HVDC interconnectors (towards 2018) must be added, while those of power plants that will be decommissioned have to be removed. The dimensioning of reserves is based on residual system imbalances. The imbalance mechanism based on single marginal pricing (since 1/1/2012) incentivises BRPs to restore the balance of their perimeter as soon as possible. This has an impact on the expected residual imbalances due to outages of power plants as shown below. Page 52 of 56 Ancillary services study - horizon 2018 Unit outages The model for the unit outages is a Monte-Carlo time-series simulation. The following probabilities for the outages of the 2018 production park are taken into account: Power plant type Estimated outages/year Nuclear units 1,5 Other power plants 3 For the low and medium reserve needs scenario, due to the expected increase in intraday market liquidity towards 2018, it is assumed that the full imbalance due to the outage will persist for 3 hours in the system, after which the BRP restores 15% of the outage per hour. After 8 hours it is assumed that the full imbalance is dealt with by the BRP, as shown in the figure below. Due to the limited assumed intraday flexibility however in the high reserve needs scenario, it is assumed that in this scenario the full imbalance due to an outage of a unit persists up to 8 hours in the system. HVDC interconnector For the future HVDC interconnector between Belgium and UK, the assumption is made that, in case of an outage, the full imbalance persists for 8 hours in the system. Afterwards the imbalances is covered by intraday market deals or by a third BRP party, depending on the market arrangements to be made18. The model assumes 2,15 outages of NEMO per year, half of which occur in a situation of full import (loss of 1000 MW) and half of them which occur in case of full export (excess of 1000 MW). 2018 estimated residual imbalances due to FOs The figure below shows the estimated residual imbalances due to forced outages of power plants and NEMO for the low and medium reserve needs scenario. Due to the limited amount of ID flexibility available, the imbalances due to FOs are assumed to be larger in the high reserve needs scenario. 18 These assumptions have a high uncertainty, as the future market arrangements for the NEMO interconnector are still unclear. Page 53 of 56 Ancillary services study - horizon 2018 FO NEMO (import) FO power plant FO NEMO (export) PV and wind residual forecast errors The 2018 estimated imbalances due to incremental installed capacity of PV and wind, compared to the 2012 baseline, are modelled by extrapolating the 2012 forecast error which is calculated on the basis of existing 2012 quarter hourly forecasts and measured infeed of wind and PV. Furthermore it is assumed that the forecasting quality will improve in the future, and that, depending on the scenario, BRPs are able to adjust their position accordingly on the intraday market or by accessing the flexibility within their perimeter. PV forecast errors In 2012 Elia started publishing a DA aggregated forecast for the installed PV production output in the Belgian control area. Since the end of March 2013, Elia started also to publish the real-time up-scaled measurement for the actual PV production output as well as an ID solar forecast. For the purpose of this study, the estimated residual forecast error for the PV production is modelled on operational data of a nearby control area [10], for which a high correlation exists for the solar forecast error with the Belgian control area. This reference forecast is an intraday forecast at 8 AM. Following assumptions on the improvement of solar forecasts and the ability of BRPs to adjust their position accordingly in the intraday timeframe were made for the different scenarios. Scenario Improvement of forecast quality Ability to adjust position accordingly in ID Total estimated reduction of forecast error Low 25% 70% -18% Medium 25% 40% -10% High 25% 10% -3% Wind forecast errors The onshore wind forecasts error is modelled on the basis of the Elia day-ahead onshore wind forecast for 2012. Following assumptions were taken for the different scenarios: Page 54 of 56 Ancillary services study - horizon 2018 Scenario Improvement of (ID) forecast quality Ability to adjust position accordingly in ID Total estimated reduction of forecast error Low 40% 70% -28% Medium 40% 40% -16% High 40% 10% -4% The offshore wind forecast error is modelled by combining the day-ahead forecast errors of Elia and the existing offshore wind parks. Following assumptions were taken for the different scenarios. Scenario Improvement of (ID) forecast quality Ability to adjust position accordingly in ID Total estimated reduction of forecast error Low 40% 70% -28% Medium 40% 40% -16% High 40% 10% -4% By doing so it is assumed that the imbalances caused by the cutting-off of offshore wind parks due to storms are reduced accordingly. This is justified by the fact that such events are predictable, in which case preventive measures can be taken. Elia will investigate this further in 2013. PV and wind ramping imbalances The impact of the ramping of VRE within the hour is modelled by taking the difference between the hourly average output and the quarter hourly output. This is shown in the figure below: Whether this effect will create residual quarter hourly imbalances in the system depends of the actual amount of available system flexibility within the hourly timeframe. Currently day-ahead an intraday markets exchange hourly energy blocks, meaning that the flexibility within the hour has to be provided by flexible units or demand side management within the system. Part of these imbalances are also offset by the simultaneous upward/downward ramping of other units and/or load. Page 55 of 56 Ancillary services study - horizon 2018 Due to the massive integration of RES it is expected that the current flexible power plants, able to provide flexibility within the hour timeframe, will be pushed out of the market more often. Given the hourly resolution of the DA and ID market, flexibility will have to be provided by the RES units themselves, by demand side management and by the remaining flexible power plants in the grid. The following assumptions were taken for the amount of imbalances due to ramping of wind and PV within the hour for all the different scenarios: Scenario Size of residual imbalances due to PV and wind ramping within the hour [as % of theoretical ramping imbalances] Low 10% Medium 40% High 70% In case a liquid 15-minutes intraday market will be available in 2018, it is expected that this effect will not introduce significant quarter hourly imbalances in the system. Page 56 of 56
© Copyright 2026 Paperzz