1 / 42

J. McCalley

J. McCalley. Wind, Markets, and Capacity. Outline. Basics of electricity markets Wind and markets Dispatchable intermittent resources (DIR). 2. Summary of power balance control levels. We are addressing inclusion of wind in real-time and day-ahead electricity markets. ….

donat
Download Presentation

J. McCalley

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. J. McCalley Wind, Markets, and Capacity

  2. Outline Basics of electricity markets Wind and markets Dispatchable intermittent resources (DIR) 2

  3. Summary of power balance control levels We are addressing inclusion of wind in real-time and day-ahead electricity markets …

  4. Basics of electricity markets • Locational marginal prices (LMPs) • Markets compute the LMPs via an internet-based double auction that maximizes participant benefits • There are 2 separate settlement processes. 4

  5. Locational marginal prices • One for each bus in the network. • Three components – see above. • If the network is lossless, and the transmission capacity is infinite, then all buses have the same LMP. 5

  6. MISO and PJM balancing areas 6

  7. RT LMPs in the MISO and PJM balancing areas 7:20 am (CST) 9/8/2011 7 MISO - PJM Interconnection Joint and Common Market Web site, located at www.miso-pjm.com/markets/contour-map.html.

  8. RT LMPs in the MISO and PJM balancing areas 7:40 am (CST) 9/8/2011 8 MISO - PJM Interconnection Joint and Common Market Web site, located at www.miso-pjm.com/markets/contour-map.html.

  9. Average annual locational marginal prices 9

  10. Market clearing price L. Tesfatsion, “Auction Basics for Wholesale Power Markets: Objectives and Pricing Rules,” Proceedings of the 2009 IEEE Power and Energy Society General Meeting, July, 2009. Computed as the price where the supply schedule intersects the demand schedule. SUPPLY Price ($/MWhr) DEMAND Quantity (MWhr) 10

  11. Market clearing price L. Tesfatsion, “Auction Basics for Wholesale Power Markets: Objectives and Pricing Rules,” Proceedings of the 2009 IEEE Power and Energy Society General Meeting, July, 2009. Computed as the price where the supply curve intersects the demand curve. SUPPLY Price ($/MWhr) DEMAND Quantity (MWhr) 11

  12. Locational marginal prices 12

  13. CAISO market design Schedules entire “next-day” 24hr period. Schedules interchange for entire “next-day” 24hr period, starting at current hour, optimizing one hour at a time (1 value per hr) Computes dispatch every 5 minutes. 13

  14. Electricity “two settlement” markets Internet system Energy & reserve offers from gens Which gens get committed, at roughly what levels for next 24 hours, and settlement Day-Ahead Market (every day) Energy bids from loads Generates 100 mw; paid $100. Energy offers from gens Internet system Generation levels for next 5 minutes and settlement for deviations from day-ahead market Real-Time Market (every 5 minutes) Energy bids from loads Generates 99 mw; pays $1. 14

  15. DAY-AHEAD MARKET (DAM) Reliability Assessment Commitment Process (RAC) REAL-TIME MARKET (RTM) The Day-Ahead Energy and Operating Reserve Market is a financially binding market that clears energy, reg reserve, spin reserve & supp reserve hourly. The Real-Time Energy and Operating Reserve Market is a financially and physically binding market that clears energy, reg reserve, spin reserve and supp reserve every 5 minutes. Two markets and a process SC-SCUC commits resources, schedules regulating reserves on committed resources and/or releases emergency operating ranges on resources. SC-SCED is used in Real-Time Energy/Operating Reserve Market to dispatch & price energy, regulating reserve, spinning reserve and supplemental reserve on a 5-minute basis. SC-SCED is used to clear/price energy, regulating reserve, spinning reserve and supplemental reserve on hourly basis. Ref: M. Tackett, Experience with Implementing Simultaneous Co-optimization In The Midwest ISO Energy & Operating Reserve Markets, IEEE PES General Meeting, 2009.

  16. Note the difference between use of SC-SCED for DAM and RTM. DAM SCUC gives a 24 hour solution Two markets - comments 1 2 3 4 5 6 7 8 9 16 17 18 19 20 21 22 23 24 SCED gives one solution per hour RTM Real-time conditions (determined by SCUC and RAC) 0 5 10 15 20 25 30 35 40 45 50 55 60 5 10 15 20 25 30… SCED gives one solution every 5 minutes

  17. They are both tools to solve optimization problems. But different optimization problems. Here are some observations. SCUC Objective SCED Objective Simplified versions of SCED and SCUC • Decision variables are Pgk • Objective & constraints are linear • Pgk are continuous valued • It is a linear program (LP). • It is a convex programming problem. • It is solved by simplex, very efficient. • For a single time period (1 hr or 5 min). • It provides LMPs. • Decision variables are zit, git, yit, xit • Objective & constraints are linear • zit, yit, xit are discrete, git is continuous • It is a mixed integer linear program (MIP). • It is a non-convex programming problem. • It is solved by branch and bound. • For multiple time periods (2-24 hrs or more) • It does not provide LMPs.

  18. Day-ahead LMPs in ISO-NE balancing areas For hour ending 11:00 am (EST) 9/8/2011 New England ISO website, at http://www.iso-ne.com/portal/jsp/lmpmap/Index.jsp. 18

  19. RT LMPs in the ISO-NE balancing areas 10:25 am (EST) 9/8/2011 New England ISO website, at http://www.iso-ne.com/portal/jsp/lmpmap/Index.jsp. 19

  20. RTAncillary service prices in ISO-NE bal areas TMSR=10min spinning rsrv TMNSR=10min non-spinning rsrv TMOR=30min operating rsrv Regulation clearing price is $5.11/MW. Load Zones: Connecticut (CT), Southwest CT (SWCT), Northeast Massachusetts/Boston (NEMABSTN) 10:25 am (EST) 9/8/2011 New England ISO website, at http://www.iso-ne.com/portal/jsp/lmpmap/Index.jsp. 20

  21. How does wind participate in markets? Wind is price-taker. Demand schedule without wind Point X Price ($/MWhr) Demand schedule with wind Point Y Supply schedule Quantity (MWhr) • “Old” approach #1 • Participates in day-ahead energy • Does not participate in AS or RT • Wind generates what it can • No deviation penalties • Paid based on computed LMP without wind, Point X below • Marginal unit backed off • Assumes wind satisfies highest “willing to pay” load. • “Old” approach #2. • Participates in day-ahead energy • Does not participate in AS or RT • Wind generates what it can • No deviation penalties • Paid based on computed LMP with wind, Point Y below • Marginal unit backed off • Assumes wind satisfies lowest “willing to pay” load 21

  22. How does wind participate in markets? • “New” Midwest ISO approach: • Dispatchable intermittent resource (DIR) • Participates in day-ahead energy • Makes offer into RT market like any other generator. But one unique DIR feature: • Instead of capacity max offered in by other generation resources, the forecasted wind MW is used as the operation capacity maximum; • Units are expected to follow the dispatch signal; • Units missing “schedule band” of 8% on either side dispatch instruction for four consecutive 5-min periods are penalized. • What are implications? Midwest ISO Market Subcom, “Dispatchable Intermittent Resource Implementation Guide,” March 1, 2011, at www.midwestiso.org/Library/Repository/Meeting%20Material/Stakeholder/MSC/2011/20110301/20110301%20MSC%20Item%2012a%20DIR%20Implementation%20Update.pdf. 22

  23. How does wind participate in markets? • What are implications? •  Wind is dispatchable! Forecasting is key! • DIRs are expected to provide rolling forecast of 12 five-minute periods for the Forecast Maximum Limit. • If forecast not submitted in time, MISO forecast is used. • Each 5 minute dispatch optimization uses Forecast Maximum Limit based on the following order • 1. Participant submitted Forecast for the interval • Must be less than or equal to the Feasibility Limit • Must have been submitted less than 30 minutes ago • 2.MISO Forecast • Must be less than or equal to the Feasibility Limit • Must have been created less than 30 minutes ago • 3.State Estimator 23

  24. How does wind participate in markets? Midwest ISO, “Dispatchable Intermittent Resource Design,” July 2010, available at https://www.midwestiso.org/Library/Repository/Meeting%20Material/Stakeholder/RSC/2010/20100726-27/20100726-27%20RSC%20Item%2019b%20Dispatchable%20Intermittent.pdf. 24

  25. Midwest ISO’s wind forecasting accuracy? 25

  26. Why is DIR beneficial? • DIRs are more likely to reduce output when LMP is negative because dispatch will instruct them to reduce; there are penalties for not following dispatch. • Inclusion of the DIRs in the RT dispatch will give SCED more flexibility to manage constraints. Therefore, there will be fewer manual curtailments: • Benefits wind for increased MWhrs produced • Benefits to system because wind offers low and therefore affects all time periods some (has very large effect during peak periods) – see next slide. 26

  27. Why is DIR beneficial? Difference in prices with (solid) and without (dashed) wind. Slanted lines are demand curves for night, day, and peak. Without wind, prices are slightly higher at night, significantly higher during the day, and much higher during the peak. “Wind energy and Electricity Prices: Exploring the “merit order effect”,” a literature review by Poyry for the European Wind Energy Association, April , 2010., available at www.ewea.org/fileadmin/ewea_documents/documents/publications/reports/MeritOrder.pdf. 27

  28. Generation Expansion Planning A simple statement of the GEP problem is as follows. • where • I(t) is total investment costs at year t • S(t) is total salvage value of retired plants at year t (and for all plants still in operation at year T). • F(t) is total fuel costs in year t. • M(t) is total maintenance costs in year t. • O(t) is the cost associated with outages. • and the overbars in the objective function indicate values must be present-worth. 28

  29. Generation Expansion Planning An alternative formulation of the GEP problem removes outage costs from the objective function and then constrains an index reflecting reliability: loss-of-load probability, LOLP, loss of load expectation, LOLE, or expected unserved energy EUE).. • where • I(t) is total investment costs at year t • S(t) is total salvage value of retired plants at year t (and for all plants still in operation at year T). • F(t) is total fuel costs in year t. • M(t) is total maintenance costs in year t. • LOLE(t) is the expected duration of time, in number of days, the system would be in an interrupted state. A standard threshold for LOLE is 1 day/10yrs or 0.1 days/yr. • and the overbars in the objective function indicate values must be present-worth. The EPRI program EGEAS is a well-known program that works like this. 29

  30. Capacity Value • The LOLE constraint is a function of three attributes: • Capacity of each unit • Availability of each unit (% of time it is available or 100-% of time it is forced out) • Load • The capacity value of a generation resource is the contribution that it makes to generation system adequacy, i.e., to satisfying the LOLE constraint. • For standard generators, this value is NOT the unit’s capacity because the unit may be forced out of service. Forced outage rates (FORs) of conventional units range between 2 and 20% giving availability of 80-98%, whereas wind energy is available at varying levels that average between 30-45%. • For windfarms, this value is NOT the windfarms capacity because the windfarm may be forced out of service and because the wind resource is rarely sufficient for providing windfarm capacity. • Capacity credit is used to identify the percentage of a windfarm’s capacity which should be identified for reliability calculations at peak load. For example, MISO was using 12.9% capacity credit for wind. This indicates the reduced availability at peak load. 30

  31. Capacity Value There is another issue why wind is considered to be less available than conventional generation. For conventional generation, the outage of one plant is independent of the outage of another plant (except in certain cases related to cascading which we will not consider here). On the other hand, if the wind speed decreases significantly at windfarm A, then it is also very likely to decrease significantly at a nearby windfarm B. In other words, wind capacity experiences correlation between plants; conventional capacity does not (and this “geodiversity” is one reason why larger boundaries for control areas can be more effective than smaller boundaries). 31

  32. Capacity Value An approach to account for low capacity factor and geodiversity is to set the capacity for variable generation equal to the amount of load that could be added without changing the risk of a shortage in generation capacity at peak load, as measured by loss of load expectation (LOLE) or loss of load probability (LOLP). This is referred to as the effective load carrying capability, or ELCC. This concept is illustrated below [*]. Horizontal line is LOLE=0.1. day is achieved for peak load less than or equal to 10000 MW. Addition of new gen moves LOLE function. If we required that the load remain the same, the LOLE would go down (get better) to about 0.09. Assume we want to maintain same LOLE value of 0.1we may grow load by 400 MW! This 400 MW load growth is called ELCC of new gen. ELCC of an additional gen will only be equal to capacity of that additional gen if additional gen is dispatchable & 100% reliable. LOLE functions with & without new gen. [*] “Capacity Value of Wind Power,” Task Force on the Capacity Value of Wind Power, IEEE Power and Energy Society, Andrew Keane, Member, IEEE, Michael Milligan, Member, IEEE, Chris J. Dent, Member, IEEE, Bernhard Hasche, Claudine D’Annunzio, Ken Dragoon, Hannele Holttinen, Nader Samaan, Lennart Söder, and Mark O’Malley, IEEE Trans on Power Systems, Vol. 26, Is 2, 2011, pp. 564-572. 32

  33. Capacity Value Ref [*] provides a three-step method for computing ELCC, which depends on development of the capacity outage probability table (COPT). The description is lifted verbatim as follows: “The COPT of the power system is used in conjunction with the hourly load time series to compute the hourly LOLPs without the presence of the wind plant. The annual LOLE is then calculated. The LOLE should meet the predetermined reliability target for that period. If it does not match, the loads can be adjusted, if desired, so that the target reliability level is achieved. The time series for the wind plant power output is treated as negative load and is combined with the load time series, resulting in a load time series net of wind power. In the same manner as step 1, the LOLE is calculated. It will now be lower (and therefore better) than the target LOLE in the first step. The load data is then increased by a constant across all hours using an iterative process, and the LOLE recalculated at each step until the target LOLE is reached. The increase in peak load (sum of) that achieves the reliability target is the ELCC or capacity value of the wind plant.” What is a capacity outage probability table (COPT)? How to do the following: “The COPT of the power system is used in conjunction with the hourly load time series to compute the hourly LOLPs” ? How to compute the COPT? [*] “Capacity Value of Wind Power,” Task Force on the Capacity Value of Wind Power, IEEE Power and Energy Society, Andrew Keane, Member, IEEE, Michael Milligan, Member, IEEE, Chris J. Dent, Member, IEEE, Bernhard Hasche, Claudine D’Annunzio, Ken Dragoon, Hannele Holttinen, Nader Samaan, Lennart Söder, and Mark O’Malley, IEEE Trans on Power Systems, Vol. 26, Is 2, 2011, pp. 564-572. 33

  34. What is a COPT? • The capacity outage table is a table of generation outage states and their associated probabilities. A very simple COPT is provided below for a power system that has only one generation that may either be • up, generating C (with capacity outage 0) at probability A, or • down, generating 0 (with capacity outage C) at probability U. If we had a power system comprised of 2 identical units, the COPT would be as below. 34

  35. What is a COPT? Consider a system with two 3 MW units and one 5 MW unit, all of which have forced outage rates (FOR) of U=0.02. There is a procedure which will allow us to construct the COPT, and it results in the table below. • This table tells us that over a given time interval, the probability that the system will have a capacity outage: • of 0 MW is 0.941192; • of 3 MW is 0.038416; • of 5 MW is 0.019208; • of 6 MW is 0.000392; • of 8 MW is 0.000784; • of 11 MW is 0.000008. The procedure for obtaining the COPT is easily extended to consider any number of units with any forced outage rates, identical or not. But before considering that procedure, let’s answer the second question. 35

  36. COTP & hourly load time series for hourly LOLP Question 2: How to do the following: “The COPT of the power system is used in conjunction with the hourly load time series to compute the hourly LOLPs” ? Observe we will have loss of load if the load exceeds the generation capacity. The generation capacity will be the installed capacity, call it IC, less the capacity that is outaged, call it Ck (corresponding to capacity outage state k). Thus, we see loss of load if d>IC-Ck So we would like to obtain Pr(d>IC-Ck). We can view this another way, by observing the criterion for loss of load is Ck>IC-d and so we desire Pr(Ck>IC-d). 36

  37. COTP & hourly load time series for hourly LOLP Question 2: How to do the following: “The COPT of the power system is used in conjunction with the hourly load time series to compute the hourly LOLPs” ? In our above example, we observe that our capacity is 11 MW. Let’s assume that the load is 5 MW. Then we may obtain Pr(Ck>11-5) =Pr(Ck>6) = Pr(Ck=8)+Pr(Ck=11) = 0.000784+0.000008 =0.000792. 37

  38. COTP & hourly load time series for hourly LOLP Denote the capacity outage as a random variable Y. Observe that the probabilities given by the COPT characterize a probability mass function (which is the discrete version of a probability density function); we will define it as fY(y), as indicated below. We can then define a cumulative probability function FY(y) according to: 38

  39. COTP & hourly load time series for hourly LOLP Note that IC-d is reserve! So FY(y)=P(Y>y) =P(CapOutage>Reserve) =LOLP At points of discontinuities, we should use the lower probability for the LOLP, i.e., LOLP(d=8)=Pr(Y>11-8) =Pr(Y>3)=0.0204 Once we have this table, given a certain load d, we may compute the LOLP during the desired time interval, as the probability that the capacity outage exceeds IC-d, which is just FY(y=IC-d), that is, LOLP(d)=FY(y=IC-d). Recalling IC=11, with a load of 5 MW, we obtain LOLP(5)=FY(y=6)=0.000792. 39

  40. COTP & hourly load time series for hourly LOLP Load time series To include the effect of wind, just use a net load time series instead of a load time series. 40

  41. COTP & hourly load time series for hourly LOLP Now we can obtain the LOLE as the expected amount of time during the year that load will be interrupted. This will be In this case, ∆T(t)=1hr, and so our LOLE expression is For our 10 hour period given in Table 9, the contribution to LOLE is 0.1814 hrs, which is 10.884 minutes. If our 10 hours is representative of the rest of the year, then it means our annual LOLE would be (8760/10)*10.884/60=158.91 That is, our annual LOLE would be 159 hours, or 6.625 days. This would be significantly above our 0.1 day in 10 years! 41

  42. Capacity Value We can then adjust loads and recompute, with and without wind, in order to obtain the below curves, and from them, the ELCC of wind. LOLE functions with & without new gen. [*] “Capacity Value of Wind Power,” Task Force on the Capacity Value of Wind Power, IEEE Power and Energy Society, Andrew Keane, Member, IEEE, Michael Milligan, Member, IEEE, Chris J. Dent, Member, IEEE, Bernhard Hasche, Claudine D’Annunzio, Ken Dragoon, Hannele Holttinen, Nader Samaan, Lennart Söder, and Mark O’Malley, IEEE Trans on Power Systems, Vol. 26, Is 2, 2011, pp. 564-572. 42

More Related