ISSUES/LOCATIONS

Documents Home
View PDF, DOC, PPT, and XLS files on line
RSS

Add NWW documents to your site (click here)

Sign up for daily updates

Keep Wind Watch online and independent!

Donate $10

Donate $5

News Watch

Selected Documents

Research Links

Alerts

Press Releases

FAQs

Publications & Products

Photos & Graphics

Videos

Allied Groups

Wind and Energy Markets: A Case Study of Texas  

Author:  | Economics, Emissions, Grid

Abstract

Many jurisdictions worldwide are greatly increasing the amount of wind production, with the expectation that increasing renewables will cost-effectively reduce greenhouse emissions. This paper discusses the interaction of increasing wind, transmission constraints, renewable credits, wind and demand correlation, intermittency, carbon prices, and electricity market prices using the particular example of the Electric Reliability Council of Texas (ERCOT) market.

I. Introduction

There has been a great increase in the amount of renewables, particularly wind power, in the United States, with highest amounts of wind generation in the states of Texas, Iowa, California, Oregon, Washington, and Oklahoma. Wind power has also grown rapidly in other countries, including those of Europe and also in Australia. There are undoubtedly several policy drivers for this global expansion of wind; however, the most widespread is to respond to the need to reduce greenhouse gas emissions. That is, the implicit assumption is that encouraging wind is a cost-effective way to reduce greenhouse emissions. This paper explores the interaction of various issues to try to understand the implications of the policy to increase wind in the United States. In particular, this paper examines the interaction of increasing wind, transmission constraints, renewable credits, wind and demand correlation, intermittency, carbon prices, and electricity market prices using the particular example of the Electric Reliability Council of Texas (ERCOT) market.

The rest of this paper is organized as follows. Section II begins with a description of offer-based economic dispatch in US markets, focusing on the particular context of real-time markets using a very simple example. Section III expands on the example by introducing transmission limitations. Section IV discusses the effect of US production tax credits and renewable energy credits. Section V considers transmission price risk and transmission construction. Section VI focuses on wind and demand correlation and intermittency. Section VII puts together the various costs and estimates the cost of reducing carbon emissions with wind. Section VIII concludes.

II. Offer-based economic dispatch

In so-called “organized” markets in United States, the process of offer-based economic dispatch by the Independent System Operator (ISO) determines the choice of which generators meet the electric demand. As a stylized example of this process, generators offer to sell:

  • energy, and
  • reserves and other Ancillary Services (AS),

and the ISO selects the offers to meet demand. Typically, there are at least two timeframes for this process:

  • “day-ahead,” technically a short-term forward market to put in place obligations for tomorrow, based on anticipation, and
  • “real-time,” using short-term (several minutes) forecasts of actual demand and other system conditions.

This paper will focus on the real-time energy market, ignoring the day-ahead market and ignoring the AS market since:

  • The real-time market will illustrate the main issues,
  • ERCOT does not currently have a day-ahead market, and
  • wind generators are unlikely to offer reserves and may not participate in the day-ahead market.

However, the implications of the need for AS to support wind production will be discussed briefly.

In the process of offer-based economic dispatch, an offer by a generator is a specification of the minimum price it is willing to receive to produce versus the quantity of production. An offer applies for a particular hour or range of hours. To simplify our discussion, we will consider a particular hour, ignoring intra-hour variations, and will consider a particular type of offer, namely a “block” offer. Specification of a block offer requires a quantity and a price and can be interpreted as an offer:

  • to generate at any level up to maximum power in the block in MW,
  • for remuneration at least equal to the nominated price in $/MWh.

Figure 1. Block offer.

Figure 1 shows an example block offer for 150 MW at a price of $50/MWh. The ISO receives offers from the various generators in the market and then selects amongst the offers to meet its short-term forecast of demand based on the offer prices. As a general principle, the ISO uses an offer with a lower offer price in preference to a higher offer price. Examples of such markets are the “organized markets” of the Northeast U.S. (PJM, ISO-NE, NYISO), Midwest, California, Southwest Power Pool (SPP), and Texas (ERCOT). Other markets throughout the world, such as the Spanish and the Australian markets, are broadly similar but have some particular differences.

A significant issue for market participants is the model of price formation; that is, the rule under which prices are set. In these markets, roughly speaking, the highest accepted offer price or, equivalently, the offer price that would serve an additional MW of demand, sets the price for all energy sold. This description is somewhat loose and a more careful definition is needed if, for example, there are insufficient offers to meet demand, the demand is at a jump in prices between block, and in the case of limiting transmission constraints (“congestion”).

To illustrate the operation of a real-time market that includes wind resources, we will consider a very simple example system. The transmission in this system consists of just two lines joining three “buses,” M, W, and N, which are the points of interconnection between generators, transmission lines, and substations that serve demand. This topology simplifies the situation compared to reality, but is useful as a start.

Although some market designs do not allow or do not require wind to make explicit offers, we will assume that wind generators offer into the market. Wind (at buses M and W) and thermal (at buses W and N) submit block offers into the real-time market to meet 1,500 MW of demand (at N). To distinguish the various generators, the wind farms at M will be called the “green” wind farms, the wind farm at W will be called the “red” wind farm, the thermal generator at W will be called the “white” generator, and the thermal generator at N will be called the “gray” generator. The offer prices are illustrated in Figure 2: wind offers at the lowest price of $20/MWh, then the “white” thermal at $50/MWh, and then the “gray” thermal at $100/MWh. We will first consider unlimited transmission in this section and then consider limited transmission in Section III.

Figure 2. Example with unlimited transmission, 1500 MW demand at N, and block offers.

To meet demand on the basis of using lower offer priced generation in preference to higher offer priced generation, all 200 MW of available wind generation, 1,000 MW of “white” thermal, and 300 MW of “gray” thermal generation are required. Amongst the dispatched generation, the highest accepted offer price was $100/MWh from the “gray” thermal generator at bus N. Note that to serve an additional MW of demand at any bus it would be necessary to use an additional MW of “gray” generation. The “green” and “red” wind and the “white” thermal generator are all fully dispatched. The situation is illustrated in Figure 3, and to summarize, the price paid to all generators and paid by demand is $100/MWh.

Figure 3. Dispatch and prices for 1500 MW demand, unlimited transmission capacity.

III. Transmission limitations

In practice, limitations on transmission capacity can limit the flexibility of the ISO to dispatch from the lowest priced offers. If the limited capacity of transmission prevents the use of an offer with a lower price then the highest accepted offer can be thought of as varying with the location of the bus. So-called “nodal” or “locational marginal prices” reflect this variation. Roughly speaking, the price at each bus is based on the offer price to meet an additional MW of demand at that bus. In the ERCOT market until December 2010 and in the Australian market, there is currently a coarser “zonal” representation of transmission. For example, the zones in the ERCOT zonal market are illustrated in Figure 4. Peak load in ERCOT is around 63 GW, with generation capacity around 73 GW.

From Figure 4, it is clear that the wind is predominantly in the West zone and that the export capability from the West zone, which varies but is on the order of 2.5 GW, is significantly less than the wind generation capacity. These stylized facts will be represented into the example to qualitatively understand the effects of transmission constraints.

Figure 4. ERCOT zones.

In particular, we will now assume that there is only 100 MW of transmission capability from bus M to bus W and only 1,000 MW of transmission capability from bus W to bus N. Optimal dispatch now involves only 100 MW of production in total from the three “green” wind farms, 50 MW from the “red” wind farm, 850 MW from the “white” thermal generation, and 500 MW from the “gray” thermal generation. The presence of transmission constraints has necessitated less use of low offer price resources and more use of higher offer price resources.

Again, the highest accepted offer price was $100/MWh from the “gray” thermal generator at bus N, but this no longer determines the price at the other buses because of the transmission limitations. While the “red” wind fully was dispatched at bus W, the “white” thermal generator at bus W was not fully dispatched. Moreover, the “green” wind at bus M was not fully dispatched.

Figure 5. Dispatch and prices with limited transmission.

To translate these observations about dispatch into the LMPs, denote the LMPs at buses N, W, and M, by LMPN, LMPW, and LMPM, respectively. Now note that:

  • To meet an additional MW of demand at N would dispatch an additional MW of $100/MWh “gray” thermal generation, so LMPN = $100/MWh at N,
  • To meet an additional MW of demand at W would dispatch an additional MW of $50/MWh “white” thermal generation, so LMPW = $50/MWh at W,
  • To meet an additional MW of demand at M would dispatch an additional MW of $20/MWh “green” wind generation, so LMPM = $20/MWh at M.

Because of these prices, the “green” wind is paid $20/MWh and the “red” wind is paid $50/MWh. These outcomes are shown in Figure 5.

IV. The effect of production tax credits and renewable energy credits

US Federal production tax credits (PTCs) and state renewable energy credits (RECs) are subsidies to renewable generation that only accrue when the renewable resource is actually generating. That is, they are paid on a per MWh basis as a subsidy “outside” the market. Internationally, there are various mechanisms to provide subsidies to wind and other renewables. For example, there is a somewhat different mechanism in Spain. Moreover, new wind development in the U.S. is likely to elect to receive investment tax credits (ITC), rather than PTCs.

Since the “green” wind farms are not fully dispatched, they are foregoing some of the subsidies. In general, to get preference in the ISO dispatch process, the wind farm must reduce its offer price. For example, if one of the “green” wind farms at M dropped its offer below $20/MWh then the lowest price offer would be fully dispatched. However, the other “green” wind farms could also reduce their offer prices and the question arises as to what offer prices would result. In general, estimating the resulting equilibrium price requires a model of competitive interaction, which has a host of assumptions. Nevertheless, we will estimate a bound on the resulting nodal price at M, LMPM.

In particular, suppose that the total value of PTCs and RECs is $35/MWh and that the variable operation and maintenance costs of the wind farm are on the order of $5/MWh. If a quantity q is sold by a wind farm at the price LMPM then the operating profit will be:

(LMPM − $5/MWh + $35/MWh) × q.

The operation profit is only positive if LMPM > $5/MWh − $35/MWh. As noted, with limited transmission, LMPM at M is set by the highest accepted wind offer at M. Consequently, if there is intense competition, wind farms may undercut each other, decreasing the highest accepted offer price. That is, the price to the wind farms, LMPM, could drop as low as minus $30/MWh.

Although negative prices seem counter-intuitive from the context of typical commodity markets, the inability to dispose of electricity without cost means that negative prices are possible. Moreover, such prices have been observed in the ERCOT balancing market in the West Zone, as shown in Figure 6 for a particular day in 2009. These negative prices can be interpreted as a transfer from US taxpayers to the market for taking wind power at unfavorable locations. Instead of the subsidies supporting the capital expenditures to develop wind, the subsidies are compensating the market for unfavorable production locations. This circumstance occurred for over 1,200 hours in total in 2008, involving 4,894 15-minute intervals. There were 3,069 negative price intervals in 2009, but only 76 negative price intervals in 2006 and 338 in 2007. As of May 31, 2010, 2,413 such intervals have occurred in 2010. [http://www.ercot.com/mktinfo/prices/mcpe]

Figure 6. Prices in ERCOT balancing market.

This situation is not restricted to the ERCOT market. Although details about subsidy mechanisms to renewables and market operation are different in the Australian market, there is analogous experience in South Australia and Victoria “zones” of the Australian market: South Australia is rich in wind resources, but has low peak demand, whereas Victoria has much higher demand. There is limited transmission capability from South Australia to Victoria. As shown in Figure 7, prices have also been negative in South Australia due to wind.

Figure 7. Prices in the Australian market.

V. Transmission price risk and transmission construction

Differences in the zonal (or nodal) prices represent the (short-term) opportunity cost to transmit power from one location to another in the constrained system. When the transmission constraints bind, the opportunity cost (and therefore transmission price) can be high. For example, the opportunity cost can be as high as $40/MWh or more from the West Zone to demand centers in ERCOT, and higher between South Australia and Victoria in Australia.

The risk of high transmission prices can be hedged by financial instruments issued by the ISO. However, the purchase price for financial instruments should reflect the average expected values of the prices being hedged, so market participants are still exposed to the opportunity cost.

In the longer-term, investment in transmission can increase the capacity to transmit power and therefore reduce the short-term transmission prices. In principle, socially optimal investment to bring energy from remote generation resources would trade-off the cost of new transmission (and new wind generation) against the production cost savings (possibly including the cost of greenhouse emissions) and capital deferment of thermal generation. In practice, production cost savings can only be roughly estimated from offers, and transmission planning may be driven by many goals.

Moreover, on-shore wind is typically far from demand in the U.S. and Australia and transmission constraints often limit transfers from wind to demand centers, as in West Zone wind in ERCOT and South Australian wind in Australia. Therefore, transmission capacity increases require more investment for wind than for thermal generation. In the context of ERCOT, the “competitive renewable energy zones” involve about US$5 billion in transmission investment for an increase in capacity of 11 GW from West. Given an optimistic 40% capacity factor for wind and typical rates of return on regulated investments such as transmission, the transmission for wind in ERCOT can be expected to cost approximately US$20/MWh on average.

It is expected that the ERCOT transmission upgrades will be completed over the next few years and more wind generation will be built, resulting in much more wind production overall. However, West Texas wind is anti-correlated with ERCOT demand since the wind tends to blow more in winter, spring, and autumn than in summer and more during off-peak hours than on-peak, whereas peak demand in ERCOT is driven by air-conditioning, particularly in summer. This is the typical case for on-shore wind in the U.S., but it should be noted that off-shore wind and solar have better correlation with demand.

The left panel of Figure 8 shows a load-duration curve that illustrates the distribution of load over a year. The right panel shows the corresponding distribution of the difference between load and wind, which is called the net load. As can be seen in Figure 8, the peak of the net load is typically nearly as high as the peak of the load, representing the fact that the wind tends not to blow on-peak. On the other hand, the minimum of the net load-duration curve is considerably lower than the minimum of the load-duration curve.

Figure 8. Load-duration and net load-duration.

Off-peak wind production tends to decrease the need for thermal generation off-peak. Therefore, if there is intense competition off-peak, prices may be set negative by wind even without transmission constraints. This concurs with recent experience in the ERCOT balancing market as shown in Figure 9, where prices during the early hours of a day between 2am and 5:30am were negative throughout ERCOT, not just in the West Zone.

These occurrences of negative prices throughout ERCOT represent a transfer from taxpayers to the market for taking wind power at unfavorable times. A serious policy implication is the potential that additional wind at these times may actually increase fossil fuel use and increase emissions, negating the fundamental policy goal underlying increased wind power. Negative prices throughout ERCOT occurred for over 30 hours in 2008. With increasing wind generation and increased transmission capability, the phenomenon of negative prices throughout ERCOT can be expected to become much more common.

Figure 9. Negative prices throughout ERCOT in early morning.

VI. Wind and demand correlation and intermittency

The negative prices result, in part, from the difficulty of forecasting when there will be significant wind and responding to that by reducing or “de-committing” the thermal generation. If off-peak wind can be anticipated in the forecast, centralized “unit commitment” could reduce wind curtailment by de-committing thermal. The Spanish and Australian markets and current ERCOT market do not have centralized unit commitment, but the ERCOT nodal market will have centralized unit commitment. It might also be better to spill more wind under some circumstances. In the longer-term, the thermal generation portfolio might adapt to “peakier” net load by increasing fraction of peaker and cycling capacity. This is illustrated in Figure 8 by the division into baseload, and peaking and cycling.

The intermittency of wind exacerbates this situation, since electricity demand and supply must be matched essentially continuously. This matching is achieved at various timescales:

  • Short-term, by adjustment of generation resources in response to system frequency, “governor action” and “regulation,”
  • Medium-term, through offer-based economic dispatch of resources to match average demand over 15 or 60 minute periods in organized markets and to acquire reserves.

In particular, meeting demand involves more than just the load-duration and net load-duration issues illustrated in Figure 8. Historically:

  • demand for energy has been considered mostly uncontrollable (but somewhat predictable), while
  • generation has been controllable (and mostly predictable).

However, wind generation is intermittent and uncontrollable at various timescales. A way to think about this characteristic is to consider wind to be “negative demand.” Moreover, integration of wind involves more than net load-duration issues. Intermittency of wind imposes requirements for additional ancillary services:

  • Short-term, increased regulation,
  • Medium-term, increased reserves and utilization of thermal resources with ramping capability,
  • Longer-term (as regulation, reserve, and ramping capabilities of existing thermal generation portfolio become fully utilized), additional flexible thermal resources, storage, or controllable demand.

Furthermore, increasing penetration of wind means less controllable generation resources may be on-line to provide ancillary services. The thermal resources that are on-line will operate at lower fractions of capacity; will be required to ramp more; and will operate more sporadically, with possibly worsened efficiencies and emissions, a larger range of prices from off- to on-peak in energy-only markets, and an even greater trend away from baseload to peaker than based on net load-duration characteristics alone.

Various US studies have estimated the “wind integration” AS costs, with estimates varying from a few to around five US$/MWh. This variation in estimates reflects:

  • Variation in particulars of systems,
  • Lack of standardization in estimating costs, and
  • Lack of representation of intermittency in standard generation analysis tools.

As a proxy upper bound to energy-related AS costs, we will consider the capital cost of lead-acid battery based energy storage, divided by the number of cycles of energy over its lifetime, which results in a cost of around US$50/MWh.

This proxy cost cannot be directly translated into an AS cost for wind. For example requirements for increased resources due to intermittency can be reduced by deliberately spilling wind to:

  • Operate at below wind capability to enable contribution of “inertia” and regulation,
  • Ramp from one power level to another at limited rate.

However, since wind turbine costs are primarily capital, spilling and limited ramps will increase the cost of wind power. Nevertheless, aggressive portfolio standards in the 20% to 30% range for energy will almost certainly involve significant changes in operations of both wind and thermal to cope with intermittency. For example, assuming all renewables are wind and:

  • 30% renewable portfolio standard by energy,
  • 40% wind capacity factor (ratio of average production to wind capacity),
  • 55% load factor (ratio of average to peak demand), and
  • ignoring curtailment,

wind capacity would be 41% of peak demand and would exceed minimum demand.

In the particular case of ERCOT, since ERCOT peak demand is about 63 GW, a 30% renewable portfolio standard for energy would require around 26 GW of wind capacity. However, even with 9 GW of wind capacity today in ERCOT, prices are occasionally negative during off-peak in spring in ERCOT, with minimum demand around 25 GW. With 26 GW of wind, there would need to be major changes in: operations; portfolio of generation; storage; and demand. There are multiple possible changes to accommodate intermittency, including:

  • Increased reserves,
  • Relatively more agile peaking and cycling generation,
  • Wind spillage, provision of inertia and regulation,
  • Compressed-air energy storage,
  • Controlled charging of millions of plug-in hybrid vehicles (PHEVs),
  • Using off-peak coal generation to power carbon dioxide separation and sequestration.

It is therefore difficult to estimate the capital and operating cost of an optimal portfolio of changes to cope with intermittency. However, as a rough ballpark proxy for energy-related AS cost due to intermittency we will suppose that lead-acid battery storage for 20% of wind energy production would compensate for intermittency. That is, intermittency would add 20% times US$50/MWh = US$10/MWh to cost of wind. This ballpark compares to estimates of up to US$5/MWh from integration studies.

VII. Cost of reducing carbon emissions with wind

This section will combine the various estimates of utilizing wind and estimate the cost of using wind to reduce carbon dioxide emissions. Some caution is in order in using these estimates, since, for example:

  • ERCOT charges most costs of transmission construction to demand.
  • North American markets generally charge all AS costs to demand, regardless of cause.

Nevertheless, we will add the wind-related transmission and wind-related AS costs to the cost of wind power. To be clear, this suggests the need for care when comparing to similar figures for other generation assets, particularly given other subsidies in the electricity sector and the fact that transmission construction costs and AS prices are not directly reflected in wholesale prices for energy.

The typical unsubsidized cost of wind energy is around US$80/MWh. As discussed previously, we estimate:

  • US$20/MWh incremental transmission for wind in ERCOT,
  • US$5/MWh to US$10/MWh proxy to cost of intermittency.

This results in a total of about US$105/MWh to US$110/MWh for wind, and should be compared to an average balancing energy market price in ERCOT of around US$50/MWh to $60/MWh. That is, wind adds about US$50/MWh to costs. The total annual ERCOT retail energy sales are around 3 times 108 MWh, retail bill around US$30 billion. Therefore, to achieve 30% renewable energy from wind would increase the retail bill by very roughly 0.3 times 3 times 108 MWh times $50/MWh or about US$4.5 billion.

To put this cost in the context of expectations for reducing carbon dioxide emissions, the US Congressional Budget Office estimates $15 per metric ton of CO2 emissions ($13-14 per US ton) as an initial price under House Bill 2454, with ceilings discussed at $30 to $35/US ton. Assuming a 10,000 Btu/kWh heat rate, a little over 1US ton of CO2 is produced per MWh of coal-fired electricity production, less for gas. That is, around at most $15 to $35 of CO2 is produced per MWh, given House Bill 2454 valuations.

We first observe that the negative ERCOT prices for wind suggests that new wind may not be decreasing greenhouse emissions. However, even assuming that wind displaces fossil emissions, it is not “worthwhile” for reducing greenhouse emissions at the initial or ceiling CO2 price.

VIII. Conclusion

This paper has considered offer-based economic dispatch in US markets, focusing on the real-time market using a simple example with wind that incorporated transmission limitations. The effect of production tax credits and renewable energy credits was considered, and the implications of transmission price risk, wind and demand correlation, and intermittency were considered to develop cost estimates for using wind to reduce carbon dioxide emissions.

Ross Baldick, Department of Electrical and Computer Engineering, University of Texas at Austin, baldick@ece.utexas.edu

This article is the work of the author(s) indicated. Any opinions expressed in it are not necessarily those of National Wind Watch.

Wind Watch relies entirely
on User Funding
Donate $5 PayPal Donate

Share:

Get the Facts Follow Wind Watch on Twitter

Wind Watch on Facebook

Share

CONTACT DONATE PRIVACY ABOUT SEARCH
© National Wind Watch, Inc.
Use of copyrighted material adheres to Fair Use.
"Wind Watch" is a registered trademark.
Share

Wind Watch on Facebook

Follow Wind Watch on Twitter