The design of electricity markets varies around the world. In regulated markets a central governing body, such as a public utility commission, sets the rates for electricity and approves or disapproves capital investments in new facilities. In deregulated systems, market forces set rates and make those decisions more freely. Within those two broad categories exist many variations, as not all regulated markets are identical, and deregulated markets have several different structures.
Texas serves as a useful example for discussing electricity markets because it has its own grid and a mix of competitive and regulated market structures. Its power sector includes a competitive market at the wholesale level and for many retail areas; regulated markets (for transmission); municipally-owned utility monopolies for power, gas and water; and the largest rural electric co-operatives in the nation, which have special challenges with low meter density per mile of distribution lines. For the Texas grid, like most other grids in the world, power plants are dispatched by price and availability, which makes price a critical factor. Each plant’s price is determined by the costs to build and operate it. The market function and those costs are described below.
For competitive wholesale markets like the one in Texas (or Germany, northeastern United States, or many other locations around the world), grid operators dispatch power plants based on which are available and their marginal cost of operation. They create a dispatch curve, also called a bid dispatch stack (or simply bid stack), that plots capacity (in megawatts (MW)) against marginal cost (in dollars per megawatt-hours (MWh)). Operators for each power plant issue a bid that informs the market operator how much capacity is available for the next time period and at what price they want to operate the plant. Grid operators assemble the bid stack by sorting (or stacking) the bids from lowest to highest with cheapest bids at the bottom of the stack. The resulting merit order determines which power plants will be dispatched. Then, once the desired capacity is met, grid operators dispatch the required power plants and pay each one the same price, the market-clearing price of energy (MCPE).
Importantly, the sorted stack of bids generally remains the same for the different power plants, though their fuel costs can vary year to year for thermal plants that require fuel as an input. Lower demand results in fewer plants turned on. Several of the power plants have stable, low marginal costs. The renewable sources such as wind, solar, and hydro have a stable cost of approximately $0/MWh. Nuclear has a stable, low cost because its fuel is relatively inexpensive. Coal is in a similar situation, though its costs are slightly higher than nuclear. By contrast, natural gas prices fluctuate significantly. When natural gas prices are low, some of its more efficient plants, such as natural gas combined cycle, are lower in the bid stack than coal plants. Less efficient natural gas power plants would be higher in the bid stack than coal. However, when natural gas prices are high, almost all natural gas plants are higher in the stack than coal. In that case, grid operators dispatch coal before the natural gas because the coal plants would be cheaper.
This sequence can be illustrated with an example. If the total demand is 50,000 MW, grid operators use the dispatch curve to determine the price per MWh based on the bid stack. Wind, hydroelectric, and solar power plants have approximately zero marginal costs, as their costs are almost entirely in the capital costs to build the plants. They bid approximately $0/MWh. Nuclear costs about $10/MWh, or $0.01 per kilowatt-hour (kWh). A typical existing coal-fired power plant costs about $40/MWh, or $0.04/kWh. Natural gas plant costs vary depending on the type of power cycle that is used. Typical bids are as follows: natural gas combined cycle bids $0.04/kWh, natural gas boilers bid $0.08/kWh, and natural gas combustion turbines bid from $0.06 to $0.14/kWh. In addition to the cost, power plant operators also have to include their available capacity in their bid. In this example, the typical Texas market capacity (presented in gigawatts) resembles the following: wind (5 GW), nuclear (4 GW), natural gas combined cycle (35 GW), coal (10 GW), natural gas boilers (10 GW) and natural gas combustion turbines (1 GW). Grid operators evaluate this “stack” of capacities to determine which plants are needed to meet demand. The last power plant is the one that is “on the margin” and sets the price for the market. Power plants that bid $0/kWh, such as wind, do so because their fuel is free. Even though they bid zero, they receive the same market-clearing price as everyone else, meaning they will make a profit anytime they are dispatched. If the demand is only 30 GW, then fewer plants are dispatched, which establishes a lower market-clearing price.
Because marginal costs determine the merit order, it is important to understand their structure. The marginal cost for a power plant is a function of the variable operations and maintenance (O&M) cost, the heat rate of the power plant, and the fuel cost. Power plants with low heat rates tend to have a lower marginal cost because they buy less fuel. Therefore, efficient power plants have lower marginal costs. As discussed, hydro, nuclear, and wind maintain low marginal costs because their fuel costs are either free or come at very low costs. Power plants with lower variable O&M costs have a lower marginal cost as well. Two types of costs contribute to the overall marginal cost of a plant. Variable costs such as fuel and maintenance are spent running the plant. Fixed costs are spent even if the plant is off. The total expenses include the fixed O&M expenses, taxes, insurance, interest, and the cost of building the power plant.
The heat rate of the power plant, a measure of its efficiency, determines some of the overall cost. The lower the heat rate, the lower the cost. Plant capacity (in MW) and fuel prices (in dollars per British thermal unit (Btu)) also drive the cost. To illustrate this difference, let’s consider two power plants with the same capacity and the same heat rate: one is a coal-fired power plant, and the other is a natural gas combined cycle plant. Even though they have the same heat rate and overall capacity, the coal plant’s marginal cost of operation would be cheaper, because in the United States in 2013, the cost for coal was about $2 per million Btu (MMBtu), whereas the cost for natural gas was about $4 per MMBtu. Additionally, the O&M costs vary for every facility. While nuclear has lower fuel costs, it tends to have higher O&M costs because of stricter safety and security standards.
The final influence on variable costs is capacity factor, or the fraction of the year that the power plant generates power. Nuclear plants in the United States typically have the highest capacity factors, achieving greater than 90% because they stay on year-round, followed by coal, natural gas combined cycle, and other sources that are operated less often. Capacity factor depends partly on the plant’s reliability, design, and fuel. Capacity factor is calculated by a ratio in which the denominator is the electricity output for maximum rated capacity for the plant running 24 hours a day for 365 days, and the numerator is how much electricity was actually generated in that year. Power plants using nuclear or coal usually have high capacity factors because they are designed to provide baseload power, which means they will operate as much as possible. They are reliable; their fuel is available; and they are relatively affordable. Some power plants are designed to have low capacity factors. For example, peaking plants operating on natural gas combustion turbines are designed only to work at peak hours in the summer months, so they have low capacity factors by design. Backup generators running on expensive diesel are only used when the power goes out, so they also have intentionally low capacity factors. Natural gas combined-cycle plants serve not only intermediate load but also baseload when gas prices are low, so they have a middle-range capacity factor. The renewables such as wind and solar have lower capacity factors because their fuel availability varies based on the weather or time of day.
Analysts incorporate all of these different factors into an integrated assessment of the variable costs expressed in terms of dollars per kWh or dollars per year. Sometimes the figurative bundle of variable costs is known as OPEX, which is shorthand for “operational expenditures.”
Capital costs are treated differently, as they are based on the initial up-front costs to build a power plant. They are also known as CAPEX, which is shorthand for “capital expenditures.” Analysts estimate the costs to build a power plant in terms of dollars per watt (W) or kilowatt (kW) of capacity. This estimate is often called the overnight price, which is what someone would pay for the plant if it were built overnight. However, this designation is misleading because capital costs are distributed over the lifetime of the plant (decades) and plants take years to build. Since few entities can write a check for a few billion dollars, they tend to procure a loan whose payback term (usually between 20 and 40 years), interest rate (which varies by technology and fuel), and principal amount determine the monthly loan payment.
The principal amount is determined by the overnight price. A general rule of thumb is that the CAPEX for power plants is $1 per W, $1,000 per kW, or $1 million per MW. However, around that rule of thumb, capital cost varies for each fuel source and for each plant. Natural gas turbines cost roughly $650 per kW. Natural gas combined cycle costs slightly more than $1,000 per kW. Wind turbines cost between $1,500 and $2,000 per kW. Solar and coal approach $3,000 per kW of capacity, and nuclear costs well over $6,000 per kW of capacity. However, the installed capacity costs only tell part of the story. While nuclear plants run 90% of the time, coal-fired plants run about 75% of the time, and solar operates about 25% of the time. Even though solar and coal have similar capacity costs to build, the coal plant would operate about three times as often, distributing costs over many more kWh of electricity. In other words, 500 MW of coal generate much more electricity than 500 MW of wind. Thus, while capital costs are similar, the cost recovery from sales of electricity is dramatically different.
The operations of each plant affect how capital costs per kWh are normalized. Building an expensive power plant that is seldom used is an inefficient use of capital, whereas building one that is used frequently is a more efficient and potentially profitable use of capital. Because the demand for power varies so much over the course of the year, much of the capital investment in power plants is underutilized. In 2011, the peak demand in Texas was over 68 GW for just a few hours in the summer. In April 2011, the lowest demand of the year was 22 GW. Most of the year, load falls between 35 and 40 GW. Engineers designed the entire electrical generation system around peak load, but the rest of the time producers do not use the power system to its full capacity. Therefore, on average, power plants system-wide only use between 40% and 45% of their capacity. However, the debt incurred in power plant construction must be paid whether they are operated or not, so if a power plant stays off for more than half the year, the cost of capital is inefficiently embedded in the plant’s pricing.
Power Plant Lifetime
Because producers must annualize capital costs over the lifetime of the power plant, an important factor to determining profitability and rate of return is the lifetime of the power plant. While most large-scale natural gas and coal power plants are built for a 25- to 40-year lifetime, nuclear power plants are usually built for a 40- to 60-year lifetime, and dams are often built for a 100-year lifetime. If a company continues to invest capital into the retrofitting and updating of equipment, plants can last 80 to 100 years and will continue to generate power as long as the owner continues to invest in their upkeep. A typical period for wind and solar farms is 20 to 25 years. Those initial expected lifetimes usually establish the loan period or amortization period for the capital cost. The total cost, expressed in $/kWh, includes the variable and the capital costs. Other costs might arise in the future, such as a cost for emissions. For example, producers might be assessed emissions fees in dollars per ton of NOX (Nitrogen oxides, a pollutant and precursor for smog formation and acid rain), dollars per ton of SOX (sulfur oxides, a pollutant and precursor for acid rain), or dollars per ton of CO2 (a greenhouse gas). There might even be costs for water use in dollars per gallon.
Integrating these different costs into a single framework is useful to planners, operators, and policymakers. Total costs can be levelized over the power plant’s expected operational lifetime to estimate the all-in cost per kWh of electricity generated. This integrated lifetime cost is known as the levelized cost of energy or levelized cost of electricity (LCOE). Dividing total annual cost (including OPEX and amortized CAPEX) in dollars per year by annual electricity generation in kWh per year yields LCOE in dollars per kWh. LCOE is an estimate, because one cannot predict fuel prices for the next 40 years. However, power plant operators try to estimate fuel costs when making a decision about power plant construction. Those estimates, along with estimates for variable and fixed costs are used to create the LCOE.
The LCOE is a useful metric for comparing the cost of these different power plants over their entire lifetimes. The LCOE varies for power plants based on the fuel, technology, location, and the year the plant was built. Coal plants built in the 1970s without environmental scrubbers have a different LCOE than new coal plants built with additional emissions controls. The cheapest plants tend to be established nuclear plants, hydroelectric dams, combined cycle natural gas turbines, coal-fired steam turbines, and some wind turbines.
While the costs of building and operating a power plant greatly affect producers and investors, and subsequently the wholesale price, they are not directly correlated to the consumer price for electricity. The price paid by the retail customer is related to wholesale electricity costs but can vary across the nation. In a regulated market, a rate-maker, such as a politically-appointed or elected government official, sets the price that consumers pay. Even though demand fluctuates throughout the day and throughout the year, and even though the generation of wind or solar might change throughout the day or the year, consumer prices are fixed despite fuel costs and supply that vary by time of day, season, or location. This approach has the advantage that producers can capture a reliable rate of return on their CAPEX and consumers are not caught by surprise from price volatility. However, this approach misses the underlying volatility in supply, demand, and price for electricity, and as a result, the price signals that consumers see can get misaligned from the sector’s fundamentals. That misalignment can lead to expensive outcomes such as blackouts or the retention of old, paid-off, dirty power plants.
Market-based systems determine the price based on market fundamentals like supply and demand. In such systems, the price for electricity in Texas would be much higher on summer afternoons than through winter nights because the demand is higher (for air conditioning) and the supply (from wind and hydro, for example) is lower. Many power companies and regulators are switching to market-based systems with real-time pricing as a way to rectify capital underutilization and to align public interests more closely with price signals. For the residential retail sector, smart meters can pass feedback and pricing directly to consumers in real time. Typically, consumer prices range between $0.10 and $0.14 per kWh for commercial and residential users in the United States, and as described in the earlier section on smart grids, dynamic price signals allow more opportunity for hedging or changing behavior. ■