Renewable portfolio standards bring volatility to Mid-Columbia markets.
Robert McCullough is managing partner at McCullough Research in Portland, Ore.
In 2006 and 2007, California, Oregon, and Washington enacted renewable portfolio standards (RPS) designed to mandate a high percentage of renewables over the following 20 years.1 The three initiatives have been highly successful and have increased the quantity of renewable resources – primarily wind – throughout the West.
The increasing number of intermittent and non-dispatchable resources has caused a variety of unintended consequences. A conflict between Bonneville Power Administration (BPA, the federal utility whose control area contains the majority of wind developments) and wind producers has reached the Federal Energy Regulatory Commission in Washington, D.C., and is the subject of a rate proceeding at the BPA.2
While thermal generation is scheduled by hour, day, and month, renewable projects often are subject to weather and climate considerations. The generation of these resources can vary wildly – often from minute to minute. Hydroelectric resources traditionally are viewed as less firm than thermal and nuclear generation, because run-off from winter snows is highly variable. But wind and solar also are variable resources, and unlike hydro they generally aren’t dispatchable. Wind, the current renewable of choice, is especially intermittent, with periods of zero generation interspersed with periods of high generation. In a perfect world, the extra generation from these renewables could be stored easily against later need. The world isn’t perfect, however. For example, the Pacific Northwest’s quasi-battery, the reservoirs along the Columbia and Snake Rivers, often face environmental constraints and can’t be used to store intermittent generation.
BPA currently lists 41 wind projects with a combined nameplate capacity of 4,711 MW (see Figure 1).3
The recent Western Electricity Coordinating Council (WECC) plan indicates massive increases are coming in wind and other renewables in the current decade (18,000 MW in wind alone) in the Western Interconnect.4 The majority of the region’s wind resources are forecasted for Oregon and Washington (see Figure 2).
The eastern desert counties of Oregon and Washington have seen an enormous growth in wind generation. There are a number of reasons why this is such a good location. The wind resources are plentiful and NIMBY (not in my back yard) concerns are rare, given the area’s low population.
One of the unforeseen consequences is that off-peak prices in the large and liquid Mid-Columbia wholesale market went negative for almost one-sixth of the time in 2011 and 2012 (see Figure 3).5
The location of the wind farms in eastern Oregon and Washington generally surrounds the massive hydroelectric resources along the Columbia River. The Mid-Columbia Bus surrounding the major dams has created one of the largest electric markets in the world. Prices at Mid-Columbia are reported in the energy media, at FERC, and at the Energy Information Administration, and are the basis of industrial and resource contracts throughout the Pacific Northwest.
The market is also highly transparent, since there’s no third-party administrative agency that can restrict market information. To recapitulate Paul Samuelson’s classic definition of a free market, there are many buyers and many sellers, and exit and entry are free.6 Unlike the administered markets in California, the market is a traditional open outcry market where any market participant can make a bilateral contract with any other. The presence of a large open outcry market adjacent to the wind farm areas in eastern Oregon and Washington makes it an excellent test bed for the changes we can expect in other areas as renewables become a larger force in market pricing (see Figure 4).
Dow Jones publishes an index for on-peak and off-peak prices at the Mid-Columbia market. Not surprisingly, the correlation between price and quantity is very high. Figure 5-A shows the prices and quantities during off-peak hours in this market from 2008 through the present. And prices at Mid-Columbia are strongly linked with renewables in statistical relationship (see Figure 5-B).7
Although we expect hydroelectric generation to stay roughly constant in years to come, we expect wind generation in the Pacific Northwest will expand sharply. If the hydro flows experienced in 2012 were to occur in 2025, we could expect a markedly greater number of days when off-peak prices would be negative.
The wind additions roughly doubling current capacity are largely mandated in the RPS adopted in Oregon and Washington. Even without comparable investments in California, we would expect the increasing numbers of intermittent renewable resources to significantly lower market prices and increase volatility in the wholesale spot markets (see Figure 6).
Volatility and Risk
Modeling prices requires more sophistication than simply holding hydroelectric generation constant and doubling wind generation. Scaling wind resources from previous years up to their 2013 levels indicates that negative off-peak prices also would have occurred in 2009. The significance of 2009 is that while 2011 and 2012 were relatively wet years, 2009 was relatively dry, with January through July runoff at The Dalles at less than 90 percent of the 80-year average.
To gain a better sense of prices under the California, Oregon, and Washington RPS, we used the base assumptions from the well-respected AURORA XMP model with two major changes.8 First, we updated the natural gas forecast using the Energy Information Administration’s 2013 Early Release Annual Energy Outlook.9 We also rebuilt the WECC to meet the requirements of the Washington, Oregon, and California renewable mandates. Since both wind and hydro are stochastic variables, we ran 20,000 games in a Monte Carlo model based on RPS resources.10 The results indicate that the current instability in off-peak prices isn’t simply an outlier (see Figure 7).
Even assuming an optimal build-out of resources subject to the three RPSs, the additional resources mandated by RPS schedules will increase price volatility and lower expected price across the next 30 years.
As always in a long-run forecast, changes in technology – e.g., electric vehicles and advanced energy storage technologies – will change the result considerably. In fact, the frequent availability of low-cost off-peak power –
and even negative energy prices – and the associated large differences with on-peak prices, will provide incentives for these changes. The low prices will also change power purchasing strategies for industries and utilities. While the risk of energy purchasing likely will increase, the advantages of taking the risk also will increase dramatically.
The present RPS standards will accentuate price instability in the Mid-Columbia market. Negative spot prices will be a feature of our market in the years ahead as oversupply conditions expand with additional mandated renewable resources. It’s likely to make non-dispatchable base load resources less competitive. It’s also likely to make contractual resources, where third parties take the volatility risk, very competitive for industry and traditional utilities.
1. The California renewable portfolio standard (RPS) was enacted in 2002 under Senate Bill 1078, accelerated in 2006 under Senate Bill 107, and expanded in 2011 under Senate Bill 2. The California RPS program requires investor-owned utilities, electric service providers, and community choice aggregators to increase procurement from eligible renewable energy resources to 33 percent of retail sales by 2020. In Oregon, Senate Bill 838 enacted in April 6, 2007, required the state’s Department of Energy to create an RPS under which electric utilities must derive 25 percent of annual retail electricity sales from renewable energy resources by calendar year 2025. In Washington, Initiative 937, a successful ballot initiative in November 2006, required large utilities to obtain 15 percent of their electricity from new renewable resources, such as solar and wind, by 2020, and to undertake cost-effective energy conservation.
5. In 2012 a good water year and the rapidly increasing level of renewables – primarily
wind – contributed to an unusual situation: off-peak energy prices in the Mid-Columbia market fell below zero on 65 days. This is the second year when negative prices were so
significant. The first year, 2011, had negative off-peak prices on 62 days.
7. The statistical relationship using ordinary least squares indicates a degree of heteroskedasticity. The use of Cochrane-Orcutt corrects for the inefficiency of ordinary least squares in the presence of heteroskedasticity. The use of just three explanatory variables – hydro generation, wind generation, and gas prices – is meant only to illustrate the relationship of these variables to spot prices. This is hardly a complete model of the Mid-Columbia market.
10. Monte Carlo modeling is an approach to forecasting where the model is run once for each pick of a set of random variables. The name references testing roulette strategies by spinning the roulette wheel many times to see what the expected outcome is. Each spin of the wheel is referred to as one “game.” In this case, we ran the WECC 20,000 times to get expected values across hydro and wind picks. We used the normal distribution for hydroelectric generation. We derived a uniform distribution for wind based on BPA’s experience over the past five years.