Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Ancillary Services Market in India


Published on

Published in: Business, Technology
  • Be the first to comment

  • Be the first to like this

Ancillary Services Market in India

  1. 1. Ancillary Services Market in IndiaAF-Mercados EMI
  2. 2. Ancillary Services Voltage Support Service Regulation and Frequency Response Service Energy Imbalance Service Operating Reserve Service Black Start Capability Service In each of the above it is important to specify  Nature of Service  When is the service required?  Source of Service and Supplier Qualification  Procurement Mechanism  Charging Mechanism AF-Mercados EMI 2
  3. 3. Regulation and Frequency Support Service  Regulation and frequency response services are necessary for the continuous balancing of resources (generation and Control Area interchange) with load, and to assist in maintaining scheduled Interconnection frequency at 50 Hz.  The above frequency profile is obtained after considerable load shedding by the states (illustrated in the Next Slide) AF-Mercados EMI 3
  4. 4. Load Shedding during peak hours and off peak hours on08.04.2013 AF-Mercados EMI 4
  5. 5. Considerations  Regulation and Frequency support service is provided by  Primary Control  Secondary Control  Tertiary Control  The states already resort to load shedding, RLDCs are able to “observe” curtailed demand  Tertiary Control can be provided to manage deviations between the “curtailed demand” and generation  Can the Frequency Support Services operate to always match this “curtailed demand” with generation obtained through tertiary frequency control ancillary service?  Or, do we need frequency triggers? AF-Mercados EMI 5
  6. 6. Pre-requisites What quantum of ancillary services should be sought?  Regulation Service requirements need to be determined consistent with reliability standards to be set by NLDC/CEA and approved by CERC.  The Regulation Service requirements need to include locational requirements and consider transmission constraints  Deviation is different in different states and there may be transmission constraints between states  What happens when  Insufficient Regulation service is bid into the market  Scheduled Resources are not available  More than anticipated regulation service is required Regulation Service providers may receive Regulation Service signals directly from the RLDC (even if they are located in the control area of the SLDC).  Receiving regulation signals directly from the NLDC does not eliminate the need to receive signals directly from the SLDC Regulation Response Rate needs to be established to qualify resources for this service (can vary on hourly / seasonal basis)  Are service allocation principles based only on costs (bids) or also Response Rates? AF-Mercados EMI 6
  7. 7. Mechanisms for Performance Tracking If the resource providing FS service deviates from schedule – payment mechanism needs to be linked to performance AF-Mercados EMI 7
  8. 8. Procurement Mechanism Through Power Exchanges based on supply offers  Offer parameters – price (Rs/MW), Response Rate, location, MW  Joint management of Congestion and FS would require SCUC with network  How is the demand curve determined?  Demand curve would need to be considered inelastic initially (till we have adequate supplies)  Since procurement would be “location-based” (markets balkanized by transmission constraints), state-specific considerations might be required Two Settlement  Day Ahead  Real Time (since capacity requirements and Response Rate requirements may change close to real time) Regulation to prevent abuse of Market Power  Price Cap / moving yardstick (discussed later)AF-Mercados EMI 8
  9. 9. Charging Mechanism In NYISO, LSEs / Generators (who do not provide FS service and do not follow RTD base points sufficiently accurately) in proportion of their load/generation Alternatively, more efficient Shapley pricing and computationally more efficient Aumann-Shapley pricing mechanisms could be used  These have extensively been applied in allocation of transmission costs AF-Mercados EMI 9
  11. 11. How do the ISOs / RTOs compensate the Generators forReactive Power?  Most make capacity payments according to compensate the allocated revenue requirements  Some pay the “opportunity cost” of reactive power when the generators need to back down real power output  Some impose penalties on generators for failing to provide reactive power 11 AF-Mercados EMI
  12. 12. International Experience (Taken from Richard O’Neill) In England and Wales, a generator can accept a  default payment of ~ $2.40/Mvarh leading or lagging, or  it may offer contracts with a minimum term of one year. In Australian ISO, generators and synchronous condensers.  receive an availability payment,  an enabling payment when dispatched and  opportunity costs from forgone sales of real power. In India, the regulator imposes a 10 paise/kVArh price on reactive power when the 1.03 < voltage < .97 In the Netherlands, generators are  contracted are paid for reactive power capability  no additional payment is made when it is supplied. In Sweden reactive power is supplied by generators on a mandatory basis, and there is no compensation. 12 AF-Mercados EMI
  13. 13. Spot Markets for Reactive PowerFERC recognizes that real time prices could be determined in the market through auctions.  the reactive power prices could either be calculated directly or  derived from the implicit opportunity costs associated with real power prices and supplier’s real power energy bids. The mechanics of price determination in each of these approaches is:  Under the direct pricing approach, reactive power sellers would submit price bids for supplying specific amounts of reactive power and the reactive power price would be the highest accepted price bid.  Under the derived approach, reactive power suppliers would submit price bids for supplying real power as well as information indicating the trade-off between supplying various amounts of real and reactive power 13 AF-Mercados EMI
  14. 14. Market Power in Reactive Power Markets Simulation and experimentation are needed to understand the effects of alternative auction market designs before such a spot market is implemented. 14 AF-Mercados EMI
  15. 15. Objectives of Simulations The objectives of the simulations done (on CIGRE 32 Bus ssystem) are two-fold:  To formulate and simulate strategic behavior of players (System Operator (SO) and the GenCos),  The results are intended to suggest mechanisms for addressing market power concerns of the regulator.  To study the “potency” of price cap regulation in alleviating abuse of market power, and suggest an alternative regulatory mechanism.  A comparison of the price cap regulation and the proposed alternative regulatory mechanism is drawn in terms of their respective abilities to produce “production efficiency” and “allocative efficiency” at the same level as a pure competitive benchmark case. AF-Mercados EMI 15
  16. 16. The Analysis What is the impact of strategic behaviour of players in the reactive power market on reactive power dispatch? How do the strategically behaving GenCos respond to price- cap regulation? Does ownership of a generator / synchronous condenser by the Public Sector (Government owned) help to mitigate market power? Does the suggested regulatory mechanism induce efficiency? 16AF-Mercados EMI
  17. 17. The Game: Formulation Multi-leader follower game  Multiple dominant players – GenCos  One follower – the SO Response of SO is constrained to be identical for each leader GenCos bid different quantities of reactive power at different prices to maximize profits  Supply function competition SO dispatches the system given these bids so as to minimize the cost of reactive power procurement 17AF-Mercados EMI
  18. 18. Reactive Power Management Scheme Day ahead real power markets clear first. The reactive power market clears in real time. The generators know that they can be called upon to generate reactive power, which might require them to change their real power dispatch.  This may alter their expected cash flow in the real power market. Hence they bid a supply curve for reactive power. The system operator (SO) minimizes the cost of procurement of reactive power and dispatches reactive power subject to the security constraints All reactive power suppliers (generators) in a geographic area get the same price 18AF-Mercados EMI
  19. 19. Solution Method (1)  The SO’s KKT conditions are parameterized in strategic variables k  The KKT conditions of the SO’s problem are concatenated with the constraints of the GenCo’s problem  Each GenCo’s problem is then a Mathematical Problem with Complementarity Constraints (MPCC)  The Equilibrium problem among the above MPCCs represents a “generalized Nash game” and it could have zero or multiple Nash equilibria  Since SO’s problem is non-convex, the solution to the KKT conditions may lead to a saddle point or a local maximum… practical way to overcome this is to try with different initial point Cont… 19 AF-Mercados EMI
  20. 20. Solution Method (2) GenCos are assumed to compete with each other in terms of their Supply Functions Hence to find Nash Equilibrium (Equilibria), KKT conditions of all the GenCos need to be solved simultaneously This is actually a non square Non-linear Complementarity Problem This makes these problems harder to solve as compared to standard Nash game Cont… 20 AF-Mercados EMI
  21. 21. Solution Methods (3) Leyffer and Munson (2005) have proposed a NLP formulation which aims to avoid this difficulty by minimizing the complementarity constraints The constraints do not include any complementarity conditions It is shown in Leyffer and Munson that local solution to above problem with the objective function value = 0, is a strongly stationary point of the multi-leader follower game In all the cases presented in this paper, the value of the objective function was less than 10-9  This compares well with the only other similar model (Bautista, Anjos, Vanelli, IEEE Trans. on Power Systems 2007), where the objective function value reported is 10-4 21AF-Mercados EMI
  22. 22. How do the GenCos and the System Operator respond to the reactive power management scheme proposed above? 22AF-Mercados EMI
  23. 23. The three cases  Case A: Competitive Setting  Case B: Oligopolistic Setting: Supply Function Equilibrium (SFE) with Price Cap  Case C: SFE with a Price Cap and Government-owned GenCo 23AF-Mercados EMI
  24. 24. 6 Bus Example: Case A 2 3 Q=0.539 6 Q=0.449 5 Q=0.749 1 = $0.399 pu 4 Payment for Reactive Power = 0.693 $/hrAF-Mercados EMI
  25. 25. 6 Bus Example: Case BGenCo 1 owns generators at Nodes 1 and 3, GenCo 2 owns generator atNode 2 2 3 Q=0.539 6 Q=0.449 5 Q=0.749 1 1 = 9.759 $/hr 2 = 3.403 $/hr = $7.981 pu 4 Payment for Reactive All Values in pu Power = 13.855 $/hr AF-Mercados EMI
  26. 26. 6 Bus Example: Case CGenCo 1 owns generators at Nodes 1 and 3, GenCo 2 owns generator atNode 2, SO puts up a 100 MVAr Synchronous Condenser at Node 1 2 3 Q=0.067 k = 20.00 k = 12.11 6 Q=0.093 k = 2.12 5 Q=0.883 1 1= 0.532 $/hr Q=1.00 2= 1.000 $/hr K=2 = $1.000 pu 4 Payment for Reactive All values in pu Power = 2.043 $/hr AF-Mercados EMI
  27. 27. Nordic 32-Bus System Similar results are obtained here However, in the earlier cases the placement of the SO-Owned generator lead to a deviation in the voltage profile and real power generation from Case A  The placement on Bus no. 4072 in this case was such that not only were the voltages closer to those in Case A  But real power dispatch and voltage angles remained unchangedWhy is this result important?AF-Mercados EMI
  28. 28. Production Efficiency and Allocative Efficiency The real power dispatch is 11331.1 MW in Case A, 11330.2 MW in Case B and 11330.9 MW in the Case C  Also the state variables (V and ) are very close to the competitive case. The effectiveness of a regulatory mechanism is to be measured in terms of its ability to mimic conditions of pure competition. Hence it is demonstrated that prudent application of the alternative regulatory mechanism leads to the same production efficiency as the competitive case. The allocative efficiency is however compromised and leads the GenCos to charge a higher price than that under competitive conditions (Case A). The outcome of the alternative mechanism (Case C) is however shown to be better than the uniform price cap mechanism (Case B) in terms of both production and allocative efficiency.‘prudent’ here is used in terms of the selection of the optimal site and capacity of Government-owned generator AF-Mercados EMI
  29. 29. Conclusions The Simulation investigates the problem of market power in real time spot reactive power markets We model the equilibrium which emerges from the strategic interaction between GenCos using the supply function equilibrium framework. When applied prudently, the proposed regulatory mechanism is shown to incentivize the competing GenCos to lower their bids and hence reduce the procurement cost of reactive power. This mechanism of regulation is non-intrusive and yet is shown to mimic the outcome of a competitive market better than a plain price cap regulation.AF-Mercados EMI
  30. 30. THANK YOUAF-Mercados EMI 30
  31. 31. The SO’s Problem Maximize  Q fi J= f i Subject to  Pfi  PDi   Vi V j Gij cos(i   j )  Bij sin(i   j )  i N iP f j 1   N Q f fi  QDi   Vi V j Gij sin(i   j )  Bij cos(i   j ) i j 1 iQ Vi  Vi min i  imin Vi  Vi max i  imax Q fi  Qmin i f  fi min Q fi  Q max fi i f  fi max fi  Pfi   Pfi min i f  min fi Pfi  Pfi max i f  max fi k fi (a fi  b fiQ fi )   i f  fi   0 31 AF-Mercados EMI
  32. 32. The GenCo’s Problem Max. R f   Q fi   (a fi  b fiQ fi )Q fi i i Subject to k fi  k min fi  0 i  min fi k max fi  k fi  0  max fi 32AF-Mercados EMI
  33. 33. Solution Method …Contd  Let K f  k fi , fi  H f  denote the bidding strategy of the generating company f.  The KKT conditions of SO are parameterized in   terms of K f , f  1, 2,...., F  The KKT conditions are necessary for optimality of the ISO’s optimization problem.  Since the SO’s problem is not convex, the solution to the KKT conditions may lead to a saddle point or a local maximum.AF-Mercados EMI
  34. 34.  Let y denote the vector of all decision variables and lagrange multipliers of the ISO’s problem P , Q , V ,  ,  ,   ,  i ,  i ,  fi ,  fi   Q P min max min max   y , y  O 1 y fi fi i i i i ,  ,  ,  ,     min max fi fi fi y  Pfi , Q fi ,Vi ,i ,  ,  o i Q i P  y   1 i min , i max , min fi , max fi ,  fi ,  min fi , max fi , AF-Mercados EMI
  35. 35.  Let the following represent SO’s Equality, Inequality and Complementarity KKT conditions z ( K1 , K 2 , K3 ,..., K F , y)  0 E I z ( K1 , K2 , K3 ,..., K F , y)  0 z ( K1 , K2 , K3 ,..., K F , y)  0 CAF-Mercados EMI
  36. 36.  The GenCo f’s problem now is Max. R f   Q fi   (a fi  b fiQ fi )Q fi Subject to i i k fi  k min  0 i  min fi fi k max  k fi  0i  max fi fi E z ( K1 , K2 , K3 ,..., K F , y)  0 wE (dual vector ) f I z ( K1 , K2 , K3 ,..., K F , y)  0 wIf (dual vector ) C z ( K1 , K2 , K3 ,..., K F , y)  0 wC (dual vector ) fAF-Mercados EMI
  37. 37. Minimize C pen   Subject to  R f z I z  T R f z I I z E E z C C I z E C  o .w f  o .w  o .w  0  1  1 .w f  1 .wE  1 .wC  y1   y y y y  y o y y y R f z I I z E E z C C T  1 .w f  1 .w  1 .w  0  z ( K , K , K ,..., K , y)  wI  I y 1 y y y   1 2 3 F  f  R f z f I z f E z f C  .w f  .w f  .w f   f   f 0 I E C min max K f K f K f K f  min .( K f  K min )  f f K f  K min  0 f K max  K f  0  max f .( K max f  Kf ) fAF-Mercados EMI
  38. 38. Two-Bus Example: Case A Q11  0.881 pu Q22  0.528 puGenCo 1 GenCo 2 1 2 System Flow1 to Flow2 to Marginal Price: PD=0.7, 2: 0.181 1: - PD=0.7, QD=0.7 0.172 pu QD=0.7 0.533*0.881 = pu $0.469 pu 1 2 Generator Nodes Pmin 0.450 0.850 Pmax 0.550 0.950 Qmax 1.000 1.000 bfi 0.533 0.889 afi 0 0 38 AF-Mercados EMI
  39. 39. Two-Bus Example: Case B Q11  0.881 pu Q22  0.528 pu GenCo 2 GenCo 1 2 1 PD=0.7, PD=0.7, QD=0.7 QD=0.7 Dispatch Remains the same, however, the GenCos jack up their bids to the maximum possible level (k = 20) The System Marginal Price ($/pu) = 9.393 If either GenCo deviates from this outcome they lose their profitsAF-Mercados EMI
  40. 40. Two-Bus Example: Case C Q11  0.365 pu Q22  0.112 pu GenCo 2 Qso,1  1.000 pu GenCo 1 2 1 PD=0.7, PD=0.7, QD=0.7 QD=0.7 k11  5.477 k22  10.738 kso,1  2.000 Uniform System Marginal Price = $1.066 pu Solution does not change with initial valuesAF-Mercados EMI