Highlights:

  1. Continued solar growth could lead to significant decreases in wholesale electricity prices during most peak hours.
  2. High solar penetrations could also impact power prices when solar does not operate, affect the frequency and severity of electricity price spikes, and impact natural gas price volatility.
  3. While lower wholesale prices could impact solar’s growth they will also hurt other energy sources, particularly coal and nuclear but also natural gas and energy efficiency.
  4. Ultimately, the price effects of solar have significant but uncertain ramifications for environmental goals, energy prices, and, ultimately, electricity market design.

Power Markets Due for Massive Changes

During the last twenty years, the majority of the U.S. electricity system has restructured from the traditional vertically integrated model to competitive wholesale markets. The defining characteristics of these markets, competitive daily energy markets and dispatch, is about to collide with the rapid increase in solar generation, with uncertain consequences.

As an intermittent resource, solar only generates during daylight hours when electricity prices are highest. Solar has high capital costs but near zero operating costs – solar will always produce when it is able. This drives down power prices when solar operates, displacing other, more expensive forms of generation.

Known as the merit-order effect, this market consequence of increasing solar generation is relatively well recognized – it results directly from the merit order dispatch of competitive wholesale markets. By reducing prices for hours when solar operates, solar could eventually reduce its own economic competitiveness, a phenomenon known as solar value deflation.  In the last year, there have been many articles focused on these effects:

Almost all of these articles have focused on how solar (or wind) hurt its own economic competiveness at high penetration levels. However, they do not address the massive economic disruption that solar could cause for traditional energy sources through wholesale power markets.

This article more broadly examines the potential wholesale market impacts of solar PV at high penetrations. It examines four ways solar can impact prevailing prices: the merit order effect, increased cycling costs, impacting electricity price spikes, and by changing natural gas pricing dynamics.

This is part 3 of a 3-part series on the future of solar. Part 1 described how solar’s continued growth, cost reductions, and tax credit renewal could drive solar to 10% of U.S. generation by 2025. Part 2 discussed the challenges facing states as they reform net metering in search of a successor policy.

Solar to Dramatically Impact Wholesale Power Prices

In the U.S., there are seven competitive wholesale electricity markets which operate the grid and constitute more than 2/3 of national electricity generation. While the specific markets have different rules and designs, they all use two daily energy markets when dispatching resources to operate the grid: day-ahead and real-time energy markets.

The purpose of these markets is to procure sufficient electricity supply to meet demand at the lowest possible price. In practice, this means that power plants are dispatched on the basis of lowest short-run marginal costs. If a power plant has low marginal costs, like nuclear or solar, the plant will almost always be dispatched. Conversely, if a power plant has high marginal costs, like coal or natural gas, it will only be dispatched in hours when its dispatch cost is lower than or at the market price for electricity.

Example of Dispatch Curve

Hypothetical dispatch curve indicates that renewables and nuclear have lowest dispatch costs while natural gas and coal are higher. Demand levels can vary significantly, leading to different price outcomes

Source: EIA

There are four primary ways that solar can impact prices in wholesale power markets:

  1. Reducing prices during daylight hours through the merit-order effect (shifting the dispatch curve to the right)
  2. Increasing prices through higher dispatch costs for cycling units (shifting cycling units to the right on the dispatch curve)
  3. Through exceptionally uncertain, region-specific changes in the frequency and severity of scarcity price spikes
  4. By changing both overall and seasonal demand for natural gas

Each of these effects will vary regionally depending on the prevailing resource mix, renewable resources, weather conditions, and market design. Understanding each of them individually is critical to understand how solar will impact any specific power market.

Lower Prices through the Merit-Order Effect

As solar and wind increase, they displace more expensive thermal generation and decrease power prices in hours in which they operate.

In essence, variable renewable energy shifts the entire dispatch curve to the right, leading to a lower wholesale power price for a given level of demand. At their most extreme, prices can go negative if there is too much power on the system (although this may reflect systemic inflexibility more than anything else).

California’s infamous duck curve provides a good way to conceptualize how the merit-order effect impacts prices in certain hours. As solar is closely tied to sunlight, its generation levels rapidly increase around dawn and rapidly decrease around dusk. In particular, the rapid decrease as night approaches and power demand is still high requires other generation to ramp up quickly.

CAISO Duck Curve

California's duck curve indicates that net energy demand will fall significantly during daylight hours during each of the next five years

Source: CAISO

Most discussion of the duck curve focuses on what it means for evening ramping requirements or what it indicates about integrating high levels of solar. However, these analyses often miss a critical point: the generation/net load duck curve also leads to a duck curve in power prices.

A quick comparison of California electricity prices illustrates this price duck curve: the graphic below compares average hourly electricity prices between May 2012 and May 2016.

Day-Ahead Average Hourly Electricity Prices at SP-15 (CAISO), May 2012 versus May 2016

Difference in electricity prices between 2012 and 2016 closely resembles the net load duck curve

Source: SparkLibrary, based on data from CAISO

These two years provide a solid point of comparison: the vast majority of California’s solar capacity has been added since 2012 and natural gas prices were near similar levels. The bottom of the price duck curve, where solar is reducing power prices significantly, is the merit-order effect in action. Very high levels of solar generation push out higher marginal cost thermal units, lowering overall power prices.

Critically, power prices are lowered the most during peak hours, when electricity is usually the highest. Average prices between 8 AM and 6 PM were only $17/MWh in 2016, almost half of the $32/MWh level in 2012.

Over time, these reduced hourly prices can really add up. In Germany, one study found that increasing renewable energy reduced average wholesale power prices by 6-10 €/MWh in 2010-2012 with the potential to reduce prices by 14-16 €/MWh this year. These declining prices have severely damaged the economics of electric utilities using traditional generation sources.

Higher Prices from Increased Cycling

While the downwards price pressure from merit order effects is significant, solar can also impact prices through increased cycling/ramping costs.

The need to ramp up generation rapidly in the evening (as indicated by the duck curve) can lead a corresponding jump in power prices. With solar having minimal impact during these hours, prices will equal the dispatch cost of the marginal unit (likely natural gas or coal in most cases).

Critically, these dispatch costs may be higher in a system with high levels of solar than they would be otherwise. This is because starting units and ramping their generation is costlier than maintaining constant generation. These increased cycling costs could mitigate some of the downwards pressure from the merit order effect.

Hypothetical Price-Duration Curves with Solar PV Effects

Hypothetical price duration curve indicates that the merit order effect will reduce prices in some hours while cycling costs increases them in others

Source: MIT Future of Solar

The actual price effects of cycling in a region will depend on the region’s demand profile, its generating mix, weather, and the solar penetration level. In general, upwards price effects will be lower in systems with higher flexibility from:

  • Readily-dispatchable natural gas generation
  • Available energy storage
  • Interregional transmission
  • Demand response resources

Overall, in the United States, the large prevalence of rapid response natural gas capacity will likely limit increased costs from cycling. While cycling will impact some of the higher cost hours, it will only be limited to a few hours a day. Comparably, merit-order effects will dominate most peak hours, leading to a net decrease in average electricity prices.

Uncertain Changes in Electricity Price Spikes

High solar penetrations could also impact power prices in a way that is relatively unexamined:  by changing the frequency and severity of price spikes (also referred to as scarcity prices).

As electricity cannot currently be stored in significant quantities, the electric grid needs to meet demand at all times or the system collapses into a blackout. When electricity demand approaches available supply, two things can happen.

First, the highest marginal cost resources (primarily natural gas or oil peaking facilities) are dispatched, greatly raising either day-ahead or real-time prices. These types of price spikes happen relatively frequently: most ISOs experience them during summer heat waves, other extreme weather events, or as a result of transmission or generation outages.

Second, if operating reserves becomes sufficiently limited, the ISO/RTO institutes shortage pricing procedures, where market clearing prices become (more or less) administratively determined. While these procedures are implemented rarely, they play an outsize role in ERCOT, Texas’ grid operator.

In August 2011, high temperatures in Texas drove ERCOT demand to record highs while drought caused key forced outages. Although the state narrowly avoided blackouts, prices spiked severely – average day-ahead peak power prices broke $100/MWh on many days. On particularly severe days, average peak prices were higher than $500/MWh, around ten times higher than normal prevailing summer power prices.

Daily average electricity prices in ERCOT during August 2011 frequently broke above $100/MWh

Source: EIA

Why are these price spikes important? Because they are highly profitable for electricity generators and very costly for consumers. In a system like ERCOT, one day of peak power prices at $500/MWh will raise average electricity prices for the whole year by $0.50-$1.00/MWh.

While ERCOT provides a poignant example of the effects of price spikes, major price spikes occur throughout the country. The causes of price spikes are highly variable and are (likely) impossible to model: prevailing weather conditions, generating fleet composition, electric trade, and contingent forced outages, to name a few.

Accordingly, it is exceptionally difficult to determine how solar growth will impact the frequency and severity of electricity price spikes. Realistically, scarcity pricing could either become more or less frequent or severe:

  • By generating at their highest levels during sunny heat waves, solar’s generation profile is well matched to the heat waves that most often cause price spikes. Thus, for most hours, solar could prevent the occurrence of scarcity pricing or limit its severity if it does occur.
  • Conversely, solar’s rapid drop off in the evening hours could cause price spikes in the evening to ensure sufficient ramping generation comes online. Although this ramping generation would be for only a few hours, severe conditions could cause shortage price conditions more severe than would occur in the absence of solar.

On balance, solar is likely to reduce the severity and occurrence of summertime price spikes in most regions. In particular, unlike thermal generation, solar does not really suffer from forced outages – as the system penetration level of solar increases, the system actually becomes less vulnerable to individual forced outages.

Reduced Overall Natural Gas Demand

There is a final major way that solar could impact wholesale power prices: indirectly by reducing power sector natural gas demand overall and by impacting seasonal demand.

During most hours in U.S. electricity markets, the market clearing price set by the electricity dispatch curve is determined by either natural gas or its main competitor coal. Accordingly, power prices usually have a direct relationship with natural gas prices.

By reducing the need for natural gas or coal generation, increased solar will tend to lead to decreased natural gas consumption. On average, this leads to lower natural gas prices and lower wholesale power prices. For example, a recent LBNL report found that the renewable energy required by state RPS policies reduced natural gas prices by $0.05-0.14/MMBtu in 2013. Higher solar penetrations will similarly keep natural gas and electricity prices down by limiting natural gas consumption.

Greater Natural Gas Price Volatility

Increasing solar generation will also impact natural gas prices by changing seasonal natural gas demand patterns and potentially altering the dynamics of natural gas price volatility.

Compared to other energy sources, natural gas has the most diverse end uses. In the U.S., only about a third is used for electricity, with the rest of demand coming from residential (16.9%), commercial (11.7%), and industrial sectors (27.3%). Most residential and commercial sector consumption of natural gas is for heating in wintertime.

This heavy demand for heating directly leads to natural gas’ price volatility: the natural gas market needs to ensure sufficient natural gas supplies to get through the next winter. A cold winter (like 2013-2014) leads to large consumption of natural gas, depleted storage, and higher prices to refill that storage. A warm winter (like this last winter) limits consumption of natural gas, leads to overflowing storage, and requires very low prices to burn off ‘excess’ natural gas.

During the last five years, this volatility has led to Henry Hub prices generally ranging between $2-6.50/MMBtu. As natural gas prices set power prices either directly or indirectly (through competition with coal), natural gas price volatility leads directly to electricity price volatility.

High generation from solar, as well as from wind, could change the dynamics of natural gas price volatility. Wind capacity factors reach their highest in spring while solar capacity factors reach their highest in summer.

2015 Monthly Natural Gas Demand versus Wind and Solar Capacity Factors

Monthly natural gas demand, concentrated around winter months contrasts sharply with the seasonal max capacity factors of wind and solar

Source: SparkLibrary, based on data from EIA

As wind and solar grow, they may increasingly displace natural gas during these seasons. Power sector natural gas demand could become even more concentrated towards both winter and fall, further increasing the impact of variable winter weather on natural gas demand and prices.

As such, solar could actually lead to greater volatility in natural gas prices. The final effects will depend on the degree to which wind and solar reduce natural gas consumption and how closely winter weather severity correlates with subsequent wind and solar resource availability.

Critically, unlike the other factors covered in this article, solar’s impact on natural gas prices is a national, not regional, phenomenon. This means that even regions with relatively low levels of solar will see reduced and more volatile power prices indirectly through natural gas prices.

Solar May Not Hurt Itself as Much as Many Think

In sum, solar is likely to have an extremely disruptive effect on U.S. power markets. It will:

  1. Lower power prices during most peak hours (historically the highest priced hours);
  2. Slightly increase power costs due to ramping generators;
  3. Reduce the prevalence and severity of scarcity price spikes in most hours;
  4. And reduce overall natural gas prices while also making them more volatile.

Recent discussion of these effects have primarily focused on the merit order effect (#1) and what it means for solar value deflation. Jesse Jenkins and Alex Trembath argue that these downwards price impacts will limit solar’s penetration levels to near its capacity factor (solid critique of this argument here). Meanwhile, Shayle Kahn and Varun Sivarum argue that the solar industry can mitigate solar value deflation through continuing to drive down costs through innovation.

The impression from this coverage is that solar will definitely ‘eat its own lunch’ and will be the resource hurt most by its success.

The reality for solar is considerably more complex.

First, the actual market impacts of high penetration solar on wholesale markets will depend heavily on regional characteristics, system flexibility, prevailing weather, and even market design. Over time, price effects from solar can encourage greater electricity trade, shift solar generation to favor generation later in the evening, reduce overall net peak demand, and even make short term energy storage more valuable. All of these will tend to limit the effects of solar value deflation.

Second, there is a critical difference between the wholesale market impacts of an energy supply and how it receives compensation. Most renewables today are on long term contracts, making them largely insensitive to short term electricity prices.

Long term contracts are based on perceptions of wholesale prices, but it may take a while for downwards price impacts to actually filter through to long term contracts. Similarly, distributed solar is almost entirely insulated from wholesale power prices – aided by net metering, it is largely competing with all-in rates.

Thus solar may be more resilient to its wholesale market effects than the current discussion indicates.

Lower Electricity Prices to Hurt Baseload Generation

However, the impacts on other energy resources could be much greater. With the potential exception of wind and hydro, solar’s impact on power markets will hurt the economics all other energy resources: coal, nuclear, natural gas, biomass, and even energy efficiency.

For most of these resources, the challenge comes from a glaring tension at the heart of U.S. competitive power markets: daily energy markets are dispatched on the basis of short term marginal costs that do not match the all-in cost structures of energy resources. To put it another way, energy resources are dispatched based on what it takes to run today not on the costs to keep the plant running tomorrow.

In particular, coal and nuclear have large fixed costs that they do not always recover in energy markets. Unlike renewables, most thermal plants do not use long term contracts, increasing their sensitivity to wholesale power prices. Thus the ultimate effect of high penetrations of solar could be to accelerate coal retirements and potentially exacerbate the financial troubles facing nuclear. The reduction in peak power prices from solar will significantly hurt these baseload resources.

Critically, natural gas capacity may be the least impacted by solar’s growth. With limited capital and fixed costs, natural gas’ cost profile closely matches its energy market revenues. Natural gas is able to ramp up and down quicker than nuclear or coal, making it better able to capture any price fluctuations from solar’s intermittency or from short term price spikes. Solar will generally reduce natural gas prices while also making them more volatile – however, natural gas will still often set marginal clearing prices, limiting financial impacts on natural gas units.

 

Read More

  1. An in-depth examination of the duck curve: http://www.nrel.gov/docs/fy16osti/65023.pdf
  2. Good discussion of potential cost innovation in solar to overcome solar value deflation: http://www.vox.com/2016/4/18/11415510/solar-power-costs-innovation
  3. Solid (but limited) discussion of the merit order curve and cycling costs: https://mitei.mit.edu/system/files/Chapter%208_compressed.pdf