Keeping the Lights On: When Electricity Gets Scarce

Exploring how power grid operators handle shortages using "Scarcity Pricing", based on the paper "Reserve and energy scarcity pricing in United States power markets: A comparative review of principles and practices" by Mehrtash, Hobbs, and Ela.

Imagine the electric grid as a giant, complex balancing act. At every single moment, the amount of electricity generated must exactly match the amount consumed. If there's too little power, you get blackouts. Too much, and equipment can be damaged. Grid operators work constantly to maintain this balance.

But what happens when things go wrong? A power plant might unexpectedly shut down. Demand might surge during a heatwave. Clouds might suddenly block the sun, reducing solar power output. To handle these surprises, grid operators keep some power plants ready in reserve, like a spare tire for your car. This backup is called Operating Reserve.

When these reserves start running low, the grid is in a state of scarcity. It's a warning sign that the system is stressed and blackouts are more likely. How should the electricity market react? This is where Scarcity Pricing comes in. It's a way for grid operators to signal this scarcity by administratively increasing the price of electricity and reserves. These higher prices encourage generators to produce more power (if possible) and motivate large consumers to use less electricity.

Different regions in the U.S. have different ways of implementing scarcity pricing. With the rise of renewable energy sources like wind and solar (which have near-zero fuel costs), these administrative scarcity prices are becoming increasingly important for setting market prices and ensuring the grid remains reliable. This post explores the core ideas and differences in these approaches, using interactive visualizations to build intuition.

The Foundation: Reserves and Demand Curves

Operating reserves are crucial for reliability. They come in different flavors, based on how quickly they can respond:

• Regulation Reserves: Respond within seconds to minor fluctuations.
• Spinning Reserves: Synchronized to the grid, ready to ramp up energy output within 10 minutes.
• Non-Spinning Reserves: Can be started and ramped up within 10 minutes.
• Flexiramp (Newer): Specifically designed to handle the rapid changes in output from wind and solar power over short periods (e.g., 5-15 minutes).

When reserves are plentiful, their "value" in the market might be low – just the cost of having a power plant ready but not producing much energy. But as the amount of available reserves dwindles, the risk to the grid increases. The "value" of that last remaining megawatt (MW) of reserve becomes much higher because it might be the only thing preventing a blackout.

Market designers represent this changing value using an Operating Reserve Demand Curve (ORDC). It's not a demand curve in the traditional sense (consumers deciding how much to buy), but rather an administrative curve set by the grid operator. It defines the price (or penalty) the system is willing to pay for reserves based on how much is available.

Here's a simplified, conceptual ORDC. Drag the slider to change the amount of available reserves and see how the "scarcity price" changes. Notice that as reserves get very low, the price shoots up dramatically.

Apples and Oranges: Different Approaches to Scarcity Pricing

The paper groups the methods used by U.S. Independent System Operators (ISOs) / Regional Transmission Organizations (RTOs) into three main categories:

1. The Ex Post Adder (ERCOT - Texas)

Texas (ERCOT) uses a unique approach. They run their main real-time energy market first *without* explicitly including reserve demand curves. After the market clears and energy prices (Locational Marginal Prices, or LMPs) are determined based on generator offers, they calculate a separate Price Adder. This adder reflects the scarcity level and is added *on top* of the energy LMP.

The adder calculation is based on two key concepts:

• Value of Lost Load (VOLL): An administratively determined estimate of the economic cost consumers bear if their power is cut off (e.g., $5000 per Megawatt-hour). This represents the maximum price the system is willing to pay.
• Loss of Load Probability (LOLP): The estimated probability, based on forecasts and historical data, that the available generation and reserves will be insufficient to meet demand in the near future. This probability increases as available reserves decrease.

Roughly speaking, the adder is calculated as: $Adder \approx LOLP \times (VOLL - Energy LMP)$. As reserves ($R$) decrease, $LOLP$ increases, pushing the adder up towards the $VOLL$. If reserves fall below a critical threshold ($R_{req}$), the adder jumps straight to $VOLL - Energy LMP$.

2. Co-optimized Stepwise ORDCs (Most Other ISOs)

Most other ISOs (like PJM, MISO, CAISO, NYISO, ISO-NE, SPP) use a different approach. They co-optimize energy and reserves simultaneously. This means the market clearing process considers both the cost of producing energy *and* the value of holding reserves at the same time. They achieve this by building stepwise ORDCs directly into the market model.

These ORDCs typically have several steps. At high reserve levels, the penalty (price) is low. As reserves decrease past certain thresholds, the penalty jumps up in steps. The highest step usually corresponds to a maximum penalty value (often related to the energy offer cap, e.g., $1000/MWh or $2000/MWh, though sometimes higher like MISO's $3500/MWh).

A key feature in these co-optimized markets is nesting. This means higher-quality, faster-responding reserves (like Regulation or Spinning) can also satisfy the requirements for lower-quality reserves (like Non-Spinning or 30-minute reserves). If there's a shortage of, say, Non-Spinning reserve, the market might use more expensive Spinning reserve to fill the gap. This causes the scarcity price for Non-Spinning to effectively "add up" with the price for Spinning, leading to potentially very high prices for the highest-quality reserves during severe shortages.

Sub-flavors: With vs. Without Flexiramp

The paper further divides the stepwise ORDC approach based on whether the market includes the newer Flexiramp product:

• With Flexiramp (MISO, SPP, CAISO): These markets explicitly procure reserves to handle short-term ramps. Because flexiramp is often needed even when traditional reserves aren't critically low (e.g., to handle solar ramp-downs), including it tends to create scarcity prices more often and extends the "right tail" of the energy scarcity premium curve (meaning prices start rising at higher levels of overall reserve availability). The penalties for flexiramp shortages are often lower than for contingency reserves, based on different rationales (e.g., cost of starting a fast generator, or just avoiding leaning on neighbors).
• Without Flexiramp (NYISO, PJM, ISO-NE): These markets rely on traditional contingency reserves (spin, non-spin, 30-min) and regulation. Their scarcity price curves tend to have shorter right tails, primarily activating during contingency events or more severe shortages.

Putting It All Together: A Comparative Simulation

The different assumptions and methods used by ISOs lead to significantly different scarcity prices under the same physical conditions. The paper presents a numerical example (Section 4) to illustrate this. Let's explore a simplified, interactive version inspired by Figure 11.

Imagine a system with 50 GW of generation capacity facing a 40 GW load. This leaves 10 GW of potential reserves. However, let's assume a minimum contingency reserve requirement of 3 GW must always be met. The chart below shows how the scarcity premium added to the energy price might vary as the *actual* available reserves (total capacity minus load) change, according to the rules of different ISOs.

Use the dropdown menu to select different ISO approaches and drag the slider to see how the scarcity premium changes. Notice the differences in:

• The maximum price reached during severe scarcity (0-3 GW available).
• How quickly the price increases as reserves decrease (the slope).
* How far the "right tail" extends (at what reserve level prices start to rise above zero).

As the simulation shows, ERCOT's VOLL-based adder leads to the highest prices during extreme scarcity. CAISO, with a lower energy offer cap (in the base case scenario), shows lower peak prices. Markets with flexiramp (CAISO, SPP, MISO) tend to show prices rising earlier (longer right tail) compared to those without (PJM, NYISO, ISO-NE). The exact shapes and step locations vary significantly based on each ISO's specific parameter choices.

Why Do They Differ So Much? And What Makes a Good ORDC?

The paper's sensitivity analysis (Section 4.3) confirms what the comparison suggests: the biggest driver of differences between the ORDCs is the penalty level chosen for severe scarcity (e.g., the VOLL or the highest penalty step). The number and size of the steps, and whether flexiramp is included, also play significant roles.

So, what makes a "good" scarcity pricing mechanism? The authors conclude that an effective ORDC should ideally have three features:

1. Reflect Marginal Value: The penalty/price should increase as shortages become more severe, reflecting the rising marginal value of reserves in preventing blackouts.
2. Consider Probabilities: The calculation should incorporate the likelihood (probability) and magnitude of potential contingencies (like generator outages or forecast errors).
3. Avoid Abrupt Discontinuities: Smooth curves (like ERCOT's adder) or many small steps are generally preferable to large, abrupt jumps in price, which can cause excessive market volatility unrelated to fundamental changes in conditions.

Getting scarcity pricing right is crucial. As our grid relies more on variable renewables, these mechanisms will increasingly determine not just reliability, but also the economic signals for investment in new generation, storage, and demand response. The significant differences across U.S. markets highlight an ongoing evolution in market design, striving to balance reliability and economic efficiency in a changing power system.