Alberta-Power: The Change is Structural, not Cyclical!
[This article is not financial or investment advice, but provided for general information purposes only. All information is subject to change and should not be relied upon for any decision making. See Webpage Terms of Use.]
In our October and November electricity market snapshots, we stated that the changes to Alberta’s power market appear more structural than cyclical. This would clearly have big implications for prices.
What makes us so confident in this assessment? In short, it’s the big shift in the composition of generating assets. In the Q&A for the October 2024 market snapshot, we shared a picture summarizing the nameplate capacity of assets that are operating & under construction vs demand scenarios.
This triggered a number of questions on how more realistic supply & demand scenarios could look - clearly we will never have all power plants operating at their nameplate capacity at the same time. Great question - time to dive deeper!
Status Quo and Growth
As of November 2023, the Alberta Electric System Operator (AESO) reports total generating nameplate capacity of 22,383 MW for the province, with an incremental 4,589 MW under construction and an additional 4,341 MW with approval by the Alberta Utility Commission (AUC). The AESO also reports a further 21,251 MW as “announced”. Since these projects have far lower probability of materializing, we will ignore them for the purpose of this article. Most of the expected growth is coming from Solar, Wind and Natural Gas Fired Cogen units:
Alberta Electricity Generating Assets, Nameplate Capacity (MW)
Existing generation mix as of November 2024 (left), adding assets under construction (center), and adding assets with approval by the Alberta Utility Commission, AUC (right).
In this article, we focus on projects that are almost certain to come onstream, i.e. those already under construction. Here, a timeline view of their expected in-service dates and nameplate capacities by type of generation:
Alberta Electricity Generation Projects under Construction
Nameplate capacity of assets under construction by expected In-Service Date (ISD). Note that usually, commissioning and ramp up take several months following ISD. For example, one 403 MW Cogen unit at Suncor’s Base plant was already counted as in service by AESO, but as of early December 2024 is still operating at around half of its nameplate capacity (see Dispatcho for more details).
Most of the capacity under construction is expected to come online over the next 12-18 months. Clearly, nameplate for Cogen is not the same as nameplate for Wind, given different capacity factors and variability, thus the chart in a way compares “apples and oranges”, but more on that later.
Cyclical vs Structural Change
Most commodity markets exhibit cyclical change patterns, with the longer-duration cycles driven by high prices leading to more investments chasing those high prices, and those investments then increasing supply and reducing prices and in turn investments. “The best cure for high prices are high prices, and the best cure for low prices are low prices”. The cycle length is driven by project lead times. In power markets, we had very low average prices from 2015-2018, which led to underinvestment that in turn drove up prices between 2020-2023. And those high prices in turn led to a large wave of new investments that now leads to cratering prices.
If that’s the case, why do we argue that the change is more structural than just cyclical? It’s due to the different type of new generation: The most recent capacity expansion wave is driven by inflexible assets, i.e. gas-fired Cogen (dominated by Suncor’s new 806MW unit at Base Plant) and several Gigawatts of wind and solar. Cogen power plants usually “have to” run as they primarily need to satisfy steam demand. And wind & solar have negligible incremental operating costs. So let’s see what this means under a variety of supply & demand scenarios.
Supply & Demand Scenarios
Alberta Internal Load (AIL), i.e. electricity demand, has recently fluctuated between ~7.9 and 12.4 GW, with an annual average of 10.3 GW forecast for the next 2 years.
Now let’s look at supply scenarios:
“Nameplate (NP)”: Nameplate basically means output at full capacity. Here, Cogen (“always on” for steam generation) plus a fraction of non-dispatchable wind & solar are sufficient to cater for all demand scenarios. But whilst somewhat illustrative, all power plants operating at nameplate capacity at the same time is not realistic, as it doesn’t make sense. The following scenarios are more likely to occur in real life, with each illustrating different aspects of the market.
“At 2023 Capacity Factors (CF)”: The capacity factor is the quotient of actual output over nameplate output over a certain period of time. What if we apply this annual capacity factor to all generating assets? We end up with average demand, as over the year, supply and demand have to balance out (the delta is due to exports & imports etc). The corresponding bar chart thus shows which source generates electricity “on average”, assuming 2023 capacity factors. For the Cogen category, assuming stable output year-round at nameplate times capacity factor is quite realistic, as they tend to be “always on” (though there is some seasonality).
“CF, 90% wind, 0% solar”: For wind & solar, applying average capacity factors makes a lot less sense than for Cogen assets - there is zero output when it’s windstill or dark, and almost nameplate output on a windy or sunny day. It’s like the “head in the freezer and feet in the oven, and on average it’s comfortably warm” analogue - i.e. the average is not very useful. In this scenario, we assume generation at 2023 capacity factors for Cogen and other fossil fuels, 90% output from wind (very windy throughout the province), and 0% from solar (nighttime). Here, Cogen and wind suffice for low and average demand scenarios without any contribution from other gas-fired generation. Therefore, if this scenario materializes during lower demand hours, one could expect prices to fall to zero dollars per MWh, which is exactly what we increasingly see.
“CF, 90% solar, 0% wind”: What if it is windstill but very sunny, for example during the middle of the day in summer? Inflexible generation is just about high enough to cater for low demand scenarios, but above that level we’d need more flexible gas generation, which usually gets bid at more than zero dollars, thus solar alone is unlikely to lead to zero dollar prices.
“CF, 70% W&S”: Now let’s assume a reasonably sunny and windy situation, with Cogen facilities again running at average 2023 capacity factor, and wind & solar each at 70% of nameplate capacity. Here, the inflexible supply exceeds even the average demand scenario. The outcome - zero-dollar power prices during many hours of the day. As this scenario requires the sun to shine, it would coincide with higher daytime demand as well, thus the amount of excess supply would likely be limited somewhat.
“Dunkelflaute”: the German term Dunkelflaute means “dark and windstill situation”, i.e. zero output from wind and solar. In that situation, the gas fired generation fleet would need to run above average capacity factors most of the time - “Dunkelflaute at CF” shows how otherwise supply would be outstripped by demand most of the time. This is what these assets are designed to do. The question is just whether at or near capacity these other assets are sufficient to cater for high demand scenarios. And the answer is basically “yes”, with flexible supply at over 15GW nameplate capacity in a max 12.4 GW market. But we need to keep in mind that Cogen units in such previous situations only barely exceeded 90% output - thus we should cap them at that level in the “Dunkelflaute at NP” (Nameplate) scenario, leading to just under 14.7 GW peak capacity. Or in other words, there is around 2.3 GW spare capacity for the Dunkelflaute scenario. This is reassuring given lower reliability of gas fired assets in e.g. ultra-cold situations. The bigger challenge will be whether these plants will earn enough revenue throughout the year to stay available for these rare but critical situations - and we may see asset retirements as a result.
Supply and Demand Scenarios for Alberta Internal Load (AIL) and Generating assets operating and under construction as of November 2024. Data per AESO Long-Term Adequacy Report, November 2024. See article for a more detailed description of scenarios.
The scenarios show that during a material amount of hours per year, the power price will be almost entirely driven by the availability of Cogen, wind and solar generating assets - no matter what other generating assets are available on the grid. Those other assets are almost irrelevant during many hours. And that’s the big structural change.
As indicated, reality is a bit more complicated - here are other key considerations:
Interties: Alberta has only limited interties to neighboring markets, but nevertheless these influence supply and demand. As of 2023, “Total Transfer Capability” was around 1 GW. If prices are low, neighbors tend to buy power from Alberta, and if prices are high, they tend to sell into the market. Thus interties tend to increase minimum demand, and decrease maximum demand. The impact tends to be bigger on the minimum than the maximum demand scenarios though, as maximum demand scenarios tend to be driven by extreme temperatures, which tend to extend into BC and Saskatchewan and driving up demand there as well - they may have no “spare power” to sell even if prices are high.
Batteries: Alberta is lagging other markets in the deployment of batteries, but this may change and increase supply flexibility, reducing supply during low price hours and increasing it during high price periods. In other markets, their impact has been more on reducing peak prices than driving up low prices however, as they simply won’t be charged unless prices are (ultra-)low.
Power demand: We made it almost to the end of the article without mentioning AI, but here we go - if new large AI loads came onto the grid, they could absorb a lot of the excess power generation. But this would affect average loads more than peak supply, which will remain driven by inflexible renewables. And deploying Gigawatt-sized data centers is not fast either.
One thing is certain - it’s a fascinating market that will only get more “interesting” over the coming years. And the opportunities it offers for flexible demand and storage assets are huge. Let’s use them!