Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration

Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration S. Martin1,2 Universidad de M´alaga M´alaga, Spain ISMP 15 July 2015 1UMA, Universidad de M´alaga, M´alaga, Spain. 2These notes are based on joint work with Yves Smeers (CORE, UCL, Belgium) and J. A. Aguado (UMA). Errors and shortcomings in this presentation are mine. Outline 1Introduction 3 2Methodology 9 Market Description 10 Ramping Products 15 Mothballing Process 18 Uncertainty Modeling 20 3Mathematical Approach 21 4Case Study 24 Results 27 5Conclusion 33 6Thank You 34 Introduction 1. Introduction Context European power market is currently decommissioning and mothballing considerable conventional generation capacity. Subsidized zero marginal cost renewable units increase their share in the generation mix and decrease electricity prices. This makes conventional plants operating in this distorted “energy only” markets unable to recover their Fixed Operating and Maintenance costs (FOM). The plants taken off line are those with higher FOM, which in many cases are recent highly efficient and flexible Combined Cycle Gas Turbines (CCGT) and Open Cycle Gas Turbines (OCGT) units. The system is thus losing a significant amount of the flexible capability that is, or will be, necessary for dealing with variability and volatility of renewable generation. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 3/33 1. Introduction Overall Question The market does not reveal the need for that flexibility, whether because of agents’ myopia or policy distortions. Will the decommissioning continue without concern for needed flexibility service as long as existing capacity is sufficient to cover demand? Could the introduction of some additional flexibility products, at some stage, stop this trend? REMARK: The proposed model is intended for exploring operation bounds and links among parameters, and not to get precise operation values. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 4/33 1. Introduction Regarding Variability from Wind Generation The additional reserve requirement due to renewable resources integration is a contentious subject, that has been studied by many authors: Doherty et al 2005, Smith et al 2007, Tuohy et al 2009, Ortega-Vazquez et al 2009, Papavasiliou et al 2011. General consensus: an increase in variability leads to an increase in the required operational flexibility, in particular the ramping capability. This is true despite the fact that under certain conditions the flexibility already required for contingency and load following could suffice to cover the uncertainty due to forecasting error of intermittent renewable generation. Capacity for ramping flexibility adds to the one already necessary for frequency support to accommodate large fractions of wind power. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 5/33 1. Introduction Literature Review on Flexible Ramp Capability Products Focus on proposals at Midcontinent Independent System Operator (MISO) and California Independent System Operator (CAISO). Definition: N. Navid and G. Rosenwald, “Market solutions for managing ramp flexibility with high penetration of renewable resource,” Sustainable Energy, IEEE Transactions on, vol. 3, no. 4, pp. 784–790, Oct 2012. L. Xu and D. Thretheway, “Flexible Ramping Products Incorporating FMM and EIM. Revised Straw Proposal,” CAISO Market Analysis and Development & Market and Infrastructure Policy, Tech. Rep., 13 Aug. 2014. Discussion: B. Wang and B. F. Hobbs, “A flexible ramping product: can it help realtime dispatch markets approach the stochastic dispatch ideal?” Electric Power Systems Research, vol. 109, no. 0, pp. 128 – 140, 2014. J. Ryan, E. Ela, D. Flynn, and M. O’Malley, “Variable generation, reserves, flexibility and policy interactions,” in System Sciences (HICSS), 2014 47th Hawaii International Conference on, Jan 2014. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 6/33 1. Introduction Focus of this Work Observation: Perverse effect, penetration of variable generation increases the need for ramping capability at the same time as it lowers electricity prices to levels incompatible with the remuneration of the conventional plants that provide that flexibility. Questions: We analyze the mothballing process and study how it is affected by a price cap implemented in the energy only market. The question we address is whether we need both energy and ramping product markets or whether a sole price cap is relevant. We test the robustness of the response to this question by verifying how it is affected by certain features of the market such as feed-in premium to wind, wind forecast, reserve requirement (estimation of the wind forecast error) and the agents risk aversion. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 7/33 Methodology 2. Methodology Ramping Products MISO’s Proposal (II) The requirement for ramping products at period tis determined by the expected variability of the net demand at period t+ 2 within a certain confidence level. Committed ramp capability at period tmust be sufficient to allow the system to go from the demand dt, at period t, to any value in the variability range [dt+2 −Kdσd,t+2,dt+2 +Kuσd,t+2] at period t+ 2. dt+2 is the expected demand at period t+ 2, and σd,t+2 is the forecast of standard deviation for demand at period t+ 2. Ku,Kdare constants that depend on the confidence level. Values Ku=Kd= 2.5 are suggested by Nivad. They correspond to a confidence level of approximately 90%, assuming a normal distribution for demand. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 15/33 2. Methodology Ramping Products Ramping Products in Our Model Coincidences with MISO proposal: Only dispatchable generators can supply ramping products. Ramping products are committed in day-ahead and deployed in real time. Required amount of ramping products. Differences with MISO proposal: In our model the energy from these products is remunerated at opportunity cost in real time, this differs from MISO’s initial proposal that supposes that they are priced at day-ahead cost. In our model, capacity reserved for ramping products is remunerated at opportunity cost as implicit in the optimization problem, instead of using a demand curve as proposed at MISO. In short, we assume the pricing of both energy and capacity in ramping products to be at opportunity cost. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 16/33 2. Methodology Mothballing Process Mothballing: Description and Assumptions Assumptions: Progressive retirement of the conventional plants. Criterion: Dismantling occurs when the margin that they make on energy and services is lower than their fixed operation and maintenance cost (FOM). Dismantling stops when all active plants cover their FOM. The margins of the dispatchable units and their FOM are calculated over a period of one year using only four representative days. Wind is represented by four wind days (24 periods of one hour in each day). These patterns come from the clustering of historical data for wind in the Spanish System in 2012. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 17/33 2. Methodology Mothballing Process Mothballing Algorithm Mothballing is modeled as an iterative process summarized as follows: 1Start with all dispatchable generators. 2Solve the optimization problem and calculate the margins for all dispatchable generators in the system, except for the back-up generator, that is never mothballed. 3The dispatchable generator with the lowest negative value of margin is mothballed. 4Solve the optimization problem with the remaining generators in the system and recalculate the margin for each dispatchable generator still in the system. 5If the margin is greater of equal than zero for all the dispatchable generators in the system, the mothballing process stops at this point; in other case we go to step 3). S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 18/33 2. Methodology Uncertainty Modeling Uncertainty Modeling We consider only uncertainty from wind generation. A whole year is represented by only four representative days, based on historical data for the Spanish system in 2012 (from the Spanish TSO). Each day consists of 24 values equal to the expected wind for each hour in that day. In order to take into account the wind forecast error, 12 scenarios are considered for each hour, assuming the values in an hour tfor a day ξ fit a beta distribution (Bofinger et al 2002, Fabri et al 2005), with average µξ,tand standard deviation σξ,t= 0.2µξ,t+ 0.02 in per unit values (Ortega-Vazquez et al, 2009). The total number of scenarios is 4×24×12, 4 wind days, 24 periods per day and 12 scenarios for each period. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 19/33 Mathematical Approach 3. Mathematical Approach Set of Constraints Capacity bounds for dispatchable generators. Capacity bounds for wind turbines. Balancing equations. Ramping constraints. Reserve requirement constraints. Ramping product requirement constraints. Conditional Value at Risk Constraints. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 21/33 3. Mathematical Approach Objective Function Consumers welfare (in case of price responsive demand). Incomes from Feed-in Premium to wind generation. Generation cost of dispatchable generators. Expected value of balancing cost. CVaR of the balancing cost. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 22/33 Case Study 4. Case Study Summary of the Test Cases Table : Summary of the test cases Demand Backup gen. (e/MWh) Ramping products Price Responsive - Fixed demand 100, 300, 563.8, 1000 No ramping products Price Responsive - Fixed demand 100, 300, 563.8, 1000 S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 24/33 4. Case Study Results Results for Several Configurations (II) Price cap my= 0.15 (wind), mx= 0.02 (dispatch. gen.) (e/MWh) Case 1 Case 2 Case 3 Case 4 µξ(%) expec. wind 5.86 61.17 22.20 22.20 22.20 22.20 22.20 22.20 ρ+(e/MWh) prem. 30.00 30.00 0.00 80.00 30.00 30.00 0.00 80.00 λrisk aversion 0.40 0.40 0.40 0.40 0.00 1.00 0.40 0.40 Energy demand in day-ahead (GWh) FD-RP 28.75 28.75 28.75 28.75 28.75 28.75 28.75 28.75 PR-RP 26.33 29.45 27.34 27.34 27.34 27.24 26.88 26.81 PR-NRP 28.58 27.19 28.55 28.73 28.74 28.41 28.57 28.60 Total profit (Me/h) (wind + dispatchable) PR-RP - 0.71 0.49 0.72 0.64 0.69 0.68 0.76 0.72 PR-NRP - 0.41 0.65 0.51 0.47 0.47 0.51 0.57 0.32 FD-RP 100.00 1.41 0.43 1.41 1.33 1.32 1.48 1.46 1.45 FD-NRP 100.00 1.39 0.71 1.37 1.34 1.31 1.44 1.41 1.16 FD-RP 300.00 5.02 0.43 4.54 4.47 4.26 4.92 4.81 4.80 FD-NRP 300.00 4.98 0.71 4.54 4.50 4.37 4.78 4.47 4.22 FD-RP 1000.00 1.96 0.43 0.55 0.47 0.52 0.45 0.54 0.53 FD-NRP 1000.00 1.92 0.71 0.51 0.47 0.47 0.51 0.57 0.31 FD-RP 563.79 1.26 0.43 0.55 0.47 0.52 0.49 0.54 0.53 FD-NRP 563.79 1.23 0.71 0.51 0.47 0.47 0.51 0.57 0.31 FD = fixed demand, RP = with ramping products, PR = price responsive, NRP = No ramping products. ∗For Case 4, my= 0.60, mx= 0.02. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 31/33 5. Conclusion Conclusions Price cap has a high impact on dismantling process. Ramping products have almost no impact on dismantling, but they have a significant impact on lowering the energy prices in the case of fixed demand. Answer to the initial questions: Regarding the dismantling process, a sole price cap is relevant. The results are robust respect to changes in: feed-in premium to wind, wind forecast, reserve requirement, estimation of the wind forecast error, and the agents risk aversion. S. Martin Effect of Ramping Requirement and Price Cap on Energy Price in a System with High Wind Penetration 32/33 Thank You for Your Attention!