Why cross-border markets for frequency control are necessary

Transcription

1 Why cross-border markets for frequency control are necessary Workshop on Exchange of Balancing Services 28 October 2010 Yann Rebours EDF R&D

2 Focus of this presentation Only frequency control UCTE s primary (local & automatic) UCTE s secondary (centralized & automatic) Focused on the costs of frequency control reserves for a producer In particular, the day-ahead de-optimisation cost Conclusions for the European market designs 2 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

3 Examples of costs for f/p reserves Fixed costs: More complex control systems An over-sized portfolio to provide reserves Variable costs: Increased wear-and-tear maintenance cost Call of additional means start-up cost Lower efficiency production cost De-optimisation of the generation dispatch production cost 3 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

4 Main f/p cost components for a producer Fixed costs ( /year) Investments in extra generating capacities, extra workforce Capacity reservation cost ( /MW/h) De-optimisation cost, opportunity cost Utilisation cost ( /MWh, /call, etc ) More fuel is used, increased wear-and-tear, Very difficult to calculate precisely these costs: Many components Interdependent time horizons Hard to separate energy-only and f/p costs Costs very different from one country to another (energy-mix dependent) 4 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

5 5 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010 The day-ahead de-optimisation cost

6 The day-ahead f/p demand in France In D-1, RTE calculates an inflexible ½-h demand: Primary f/p reserve 700 MW Secondary f/p reserve 500 MW à 900 MW Note: peak load 86 GW R sec France if gradient = if gradient total total gradienttotal = gradient where a = 10 MW b = 150 MW < MW MW consumption max 500, gradient 6 + gradient total al exchanges max zone + b 2 b 6 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

7 The day-ahead f/p demand for a producer RTE dispatches this demand amongst producers: The price is agreed ex-ante The French producers take this constraint into account and may exchange some reserves within France 7 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

8 The D-1 de-optimisation cost Method: easy & standard (but some limitations) Applied to EDF Producer over 879 days with operational data & algorithm (from 09/2004 to 08/2007) De-optimisation cost = Cost D-1 dispatch with reserves Cost D-1 dispatch without reserves G1 G2 G1 G2 G3 energy for loads frequency control non-used capacity without reserves (cheaper dispatch) with reserves (more expensive dispatch) 8 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

9 The method in practice OTESS 1. APOGEE dataset with initial constraints 2. APOGEE dataset with reserve demand = 0 MW APOGEE (EDF s short-term dispatch algorithm) APOGEE dispatching 1 APOGEE dispatching 2 OTESS (Data management tool specifically developed) Other data for correlation Database Web-based interface 9 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

10 Advantages and limits of the method for EDF Advantages of our study vs previous literature: APOGEE takes into account many constraints (dynamics, hydro, thermal ) Actual day-ahead operational data as input However, some limits in our study: If hydro energy is provided, reserve is provided as well (historical constraint) Water usage was not taken into account (days are considered as independent) The precise management of hydro valleys was not considered (only a global constraint for each reservoir) Obviously, manual day-ahead re-dispatch is not modelled Only short-term variable costs are considered Therefore, this method gives a minimum cost (real cost is higher) Further studies has been performed to remove some limits. A publication is expected 10 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

11 1. The D-1 de-optimisation cost is not negligible Up to 7,8 % of the dispatching cost (= hundreds of M /year) The reserve should be remunerated in order to avoid distortion of energy prices An efficient procurement that reveals the true cost is essential 11 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

12 2. The D-1 de-optimisation cost is variable During 54 % of the time, the cost changed by more than 20 % from one day to the following Allowing short-term arbitrage is thus essential Cross-border exchanges may help 12 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

13 3. Reserve MC can be higher than energy MC Reserve marginal cost > Energy marginal cost happens often D-1 cross-border capacity reservation for reserve may make sense Percentage of time when marginal costs of energy are higher than marginal costs of reserves as a function of the time step for all the studied days 13 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

14 4. Synchronous time control & reserve To maintain the average frequency at 50 Hz, the secondary target frequency is changed every day by f c = 0.01 Hz Because of time control, the reserve demand was modified by the French TSO up to January 2009 (±150 MW) P G from primary frequency control (MW) R p Operating points: (a) To delay time (b) Normal (c) To accelerate time 0 f n + f c f n f c f n Frequency (Hz) 14 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

15 4. Synchronous time control costs money This policy cost millions of euros/year in D-1 de-optimisation Does the value of time control justify its cost? [Rebours et al (2009)] What about other countries? f target = Hz (27 days over 01/ /2007) f target = Hz (175 days over 01/ /2007) 15 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

16 Conclusions 16 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

17 Conclusions Frequency control reserves incur various costs for producers: Fixed ( /year), capacity reservation ( /MW/h) & utilisation cost ( /MWh, /call) The allocation of the service cost should recognize this fact The reserve costs may be hard to estimate, but some solutions exist For D-1 de-optimisation cost, results & impacts on the design: The cost is not negligible Payment of the service is compulsory The cost is volatile Short-term arbitrage is essential and cross-border may help Remuneration should have a time-variable price Res. > Enr. MC happens D-1 cross-border capacity reservation may be useful Time control policy costs millions of euros/year for little value Such a policy should be removed (done in January 2009 in France) Therefore, liquid markets for frequency control are necessary Cross-border markets help increase liquidity To avoid cross-border distortions, uniform market designs may help 17 - Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010

18 Acknowledgments & further reading Many thanks to all the colleagues who fuelled this work Most of the material presented & further references can be found in: Y. Rebours, "A Comprehensive Assessment of Markets for Frequency and Voltage Control Ancillary Services", PhD Thesis, Faculty of Engineering and Physical Sciences, The University of Manchester, Manchester, U.K., Available: ebours_phd_2008.pdf Y. Rebours, J. Pestourie, and E. Monnot, "Five Good Reasons to Abandon Synchronous Time Control", presented at the IEEE Powertech, Bucharest, Romania, Why f/p cross-border markets are necessary - Workshop on Exchange of balancing services - 28 October 2010