Saturday, March 9, 2024

Saskatchewan's Case for Nuclear Power

Since September 2022, I’ve been collecting SaskPower daily generation data with my Saskatchewan Electricity Mix project. Using that data I will show how nuclear power is ideally suited to supply Saskatchewan’s power demand. 

Power demand in Saskatchewan is fairly constant on average; the minimum load experienced in SaskPower’s 2022-23 reporting year was 2,032 Megawatts (MW). This is the minimum amount of constant power required to run homes, businesses, industry, hospitals, schools, farms, mines, and more. 

The chart above shows one year of daily average power demand numbers, with the 2,032 minimum load labeled as “base load”. 

The bulk of our power demand and consumption is constant. 

Dispatchable power is needed to serve base demand

Capacity Factor is the ratio (%) of the power produced by a generating asset (or fleet of assets) from its installed capacity. 

This chart illustrates the per-source daily capacity factor for each of SaskPower’s power reporting categories: coal, gas, wind, hydro, solar, and “other” (which includes contracted Manitoba Hydro imports, bioenergy, and small IPPs). 

Each dot represents one day’s capacity factor. The box-and-whisker plots highlight the median value (line in the middle box), operating quartiles (each box or whisker contains approximately 25% of values), and outliers (more on how to read box plots).

We can see that Saskatchewan uses coal, gas, and hydro to meet demand continuously, because the capacity factor of these assets never nears zero. They are dispatchable, controllable power sources. With wind and solar we take what power we can get, and some days the capacity factor of those asset fleets is at or near zero. 

Saskatchewan wind is great, but… 

At about 37% annual capacity factor (the middle bar in the box plot above), wind power in Saskatchewan is a great resource relative to wind power in other regions. But the wind is highly variable. 

Let’s define a “calm” day as bottom-quartile wind performance: the one-quarter of days contained in the bottom whisker of the plot above, where wind produced less than ~17% of its installed capacity. 

In the last 12 months, the longest “calm streak” we’ve experienced - concurrent bottom-quartile days with our geographically-distributed wind fleet - is seven days (two of them, in fact). 

Plotting wind performance as a histogram may help further visualize the variability. For example, there were 14 days where the wind fleet’s capacity factor was between 0% and 5% of installed capacity. 

Best-in-Canada solar…?

Saskatchewan is said to have some of the best solar resources in Canada, but this is a bit like saying the Okanogan Valley is the best place in Canada to grow avocados. The best is not that good. 

Using SaskPower’s daily generation data we calculate an annual solar capacity factor of about 17.2% (consistent with a City of Saskatoon study; PDF). For every 100 MW installed, we average (over a year) 17.2 MW of generation. In contrast, solar’s capacity factor for sunny equatorial regions can exceed 30%.

Solar performance is especially poor in Saskatchewan winters:

A common claim (or misconception) is that solar generation aligns with peak loads; the highest load experienced by a grid in a year. This is true for some sweltering southern grids with late afternoon summer A/C loads, but not Saskatchewan’s winter-peaking grid. 

The chart below shows the last 8 months of daily peak loads, and if the peak load occurred when the sun was up (and solar would be producing) or down. Note the 8-month timespan is due to data availability, not deliberately excluding data. 

Saskatchewan’s annual peak - the day or hour with the highest load - consistently occurs in winter, when the sun is down (at 4pm!). The conclusion is that solar’s effective load carrying capacity during winter peak loads - where the peaks are the highest - is zero. 

Hydro is not as dependable as we might hope.

In the last year, hydro’s capacity factor has averaged about 32%, likely due to low moisture levels, drought conditions, and the fact that hydro dams support other functions of irrigation, flow, and control. 

As a result, in the last 6 months we’ve been utilizing just 200-300 MW (daily averages) from our nearly 900 MW of hydro generation capacity, often at much lower levels than the year before. 

In their most recent Integrated Resource Plan, Manitoba Hydro indicated they will not have excess hydro to export to Saskatchewan, in the long term. 

Critical analysis of renewables is not anti-renewable

Believe it or not, I am not opposed to renewable (wind/solar/hydro) energy. If decarbonization is the goal, many solutions will be required to transform and update Saskatchewan’s grid, including renewables and energy storage. 

My concern, looking at the data, is hearing unrealistic ideals and perceptions about the real-world utility of renewable power. It is not realistic to depend solely on renewable power, unless we want blackouts, curtailments, or a large dependency on imported power. 

Could battery storage mitigate renewable intermittency? 

Some batteries on our grid will be beneficial for tight frequency control and near-instant response to other failures. However, the quantity of batteries required to store long-duration power from renewables is staggering. 

Here's the math: if our daily average power consumption is 70 GWh, we’d need 490 GWh of energy storage to get through a 7-day calm streak (as described above) on a wind-dominant, zero-fossil grid in the dead of winter. This amount of storage is equal to about 25,000 20 MWh batteries (like the one SaskPower is deploying in Regina), or 36,000,000 Tesla Powerwall units. 

This is an extreme and slightly absurd example, but a small fraction of that storage is a huge amount.

Nuclear power is the most reliable source of electricity 

Let’s finally discuss nuclear power and start with reliability. Ontario's fleet of reactors (below) provides uninterrupted zero-carbon power 24/7/365 under any weather conditions.Just this week, one of Ontario’s nuclear reactors reached 600 days of continuous runtime.

Omitting hydro due to Saskatchewan's limited options, Ontario is a great model for Saskatchewan's future grid.

Nuclear is the anchor. Wind and solar reduce natural gas use. Gas and hydro flex to meet demand and manage wind and solar intermittencies.

Nuclear power excels at decarbonizing electrical grids.

Pictured is the daily carbon intensity of electricity of fossil-heavy Saskatchewan and neighboring Alberta, renewable-heavy Germany and California, and nuclear-heavy France and Ontario.

If decarbonization is the goal, what is the answer? 

Nuclear power has the smallest lifecycle land footprint.

Here we illustrate the lifecycle land use intensity of our current grid, and grid decarbonization using a mix of wind and solar (with approximately 4:1 ratio of installed capacity, per 2030 projections from SaskPower) OR nuclear power.

Lifecycle land use includes mining (of fuel and construction materials), the actual plant site, and waste management (landfills, nuclear spent fuel)..

If the goal of environmentalism is minimizing impacts on our natural resources, nuclear is the clear leader.


Geographically, Saskatchewan is challenged to produce reliable electricity from wind, solar, and hydroelectricity to meet demand, as shown by this real-world data, despite “best-in-Canada potential”. 

Electricity produced from coal, natural gas, and hydro power share many of nuclear power’s positive attributes, and all of these base load power sources have varying levels of social license and support. Nuclear power however, not without its own challenges and drawbacks, is a clear win for reliability, energy security, environment, carbon intensity, and the ability to serve base load. 

An “always live” version of the charts in this piece can be found here, including links to sources and other notes about the data. 

Saturday, February 25, 2023

Four Saskatchewan SMRs isn't enough.

I hope to persuade you that Small Modular Reactors (SMRs) should form a key part of Saskatchewan's power future, and that the four we're talking about building (from 2034 through 2042) are both too little and too late, given our current power generation mix and expected growth.

Saskatchewan should build more than four SMRs, and we should start building them as soon as reasonably possible. 

Background & Assumptions 

A few of my old posts may be worth reading (see index of energy posts) but this post builds significantly on my Path to 2030 post in methodology and what I've learned since August. 

For a deep dive on the BWRX-300 in Saskatchewan and how we can to be evaluating this SMR in particular, see this post

I've held these key assumptions through the analysis below:

  • There is no (or very limited) capacity for new large-scale hydro. This has been frequently repeated by SaskPower and is intuitively true for a very flat province.  
  • Generating assets have approximate lifespans of 20 years (batteries*), 25 years (wind, solar, sawmill waste*), 30 years (coal, natural gas, flare gas*, waste heat*, bio-refinery*), 60 years (nuclear), 100 years (hydro), before major refurb, replacement, or decommissioning is required. Lifespans marked with a * are an educated guess (and the sources are small enough not to impact the analysis), the rest are based on surveys of industry expectations. 

My usual caveat applies: I'm not an expert in this space, but I am highly interested and motivated to learn more. More assumptions are noted in the spreadsheet linked at the bottom of this post. 

The Path to 2050 2055

I've combed through SaskPower's system map and asset list, planning and construction projects list, news releases and more to try and build a basic-but-comprehensive model of our provincial power grid. 

The basic model has just a few parameters:

  • Power plant type (natural gas, coal, wind, etc.) 
  • Power generation capacity (in MW)
  • Expected asset life (in years, and specific dates)

Aggregating this information into one chart, we can visualize SaskPower's planned power mix to 2055. 

Why 2055, and not 2050? It shows the end of life of a large amount of mid-2020s wind and solar assets with expected 25-year lifespans.

Emphasizing my comment above about this being a "basic" model, here are some comments on the above image: 

  • Coal drops off by 2030 per federal regulations. (I don't think SaskPower has shared if it will be a gradual ramp-down or an abrupt stop, so I've modelled it as the latter)
  • Four SMRs are built, commissioned in 2034, 2037, 2040, and 2042. This timeline is based on the 2021 inter-provincial SMR Feasibility Study (but the province has made no firm commitment; the decision on the first SMR is scheduled for 2029. See this post). 
  • It is uncertain if new federal clean electricity regulations will mandate the phase-out of natural gas for electricity by 2035. This model assumes a fast phase-out is unlikely, so I've shown gas capacity tapering off as plants hit end-of-life (30 years). 
  • Of course, most power assets can be refurbished, rebuilt, or replaced-in-kind instead of being taken out of service, but that is not shown. 

Declining Dispatchable ("Base load") Power Poses Risk, Uncertainty 

The most concerning outcome of this model is the uncertainty about where we'll get secure, dispatchable, "base load" electricity in the future. This is power that can be called upon when needed: boilers lit, dam water spilled, atoms split, batteries discharged, etc. 

Intermittent power (like electricity from wind or solar) depends on favourable weather conditions and may not be available when we need it (or: too much may be available when we don't need it). 

On the following chart I've added two lines: 

  • The red line is my forecasted/estimated peak load: the maximum instantaneous power demand in the province. I've taken our historical peak load (3910 MW, set December 2021) and increased it by 0.5% per year. This is an extremely conservative assumption (as we'll see later) given the push to "electrify everything" to reduce emissions, and given Saskatchewan's growth plan which seeks to increase our population to 1.4M by 2030. More people, more industry, more power. 

  • The black line is the total dispatchable power capacity. I've both over-estimated dispatchability by assuming 100% of hydro can be called upon (typically it runs <50% capacity factor, see for historical utilization), and under-estimated dispatchability by assuming imports are not dispatchable. Because this is a simple model, let's say these two assumptions cancel each other out.

The area of concern on this chart is where dispatchable power drops below the conservatively-estimated peak load. This means on a dark, cold, winter night when we need electricity the most, our hospitals, care homes, industries, and homes will be depending on imports and weather-dependent energy instead of "firm" dispatchable power.  

It's fair to object to my suggestion imported power is not dispatchable or reliable. It might come from firm sources... but it might not. 

Consider the new 650 MW interconnect to the Southwest Power Pool (SPP), due to be completed in 2027. SPP is a huge grid: over 85 GW of generating capacity (compared to our 5.5 GW). However, one third of that capacity is wind, and the wind may not be blowing when we need it. 

We cannot absolve ourselves of responsibility for our own energy security. Electricity is critical to homes, businesses, farms, industries, emergency services, and life in general. Like healthcare, we cannot outsource this critical service to other jurisdictions. 

The Impacts of Increasing Intermittency

The final version of this chart condenses different generation types into three categories: dispatchable power, variable (intermittent) power, and imported power. 

Note: Most imported power from SPP would likely be from variable sources, so the % Variable estimate is likely low. 

The dark gray line highlights a growing percentage of provincial electricity will come from intermittent sources. This means:

  • Power prices will increase due to redundancy required by intermittent power sources, including:
    • Dispatchable backups like natural gas for when it is calm/dark. (I don't think batteries are a credible grid-scale solution to get us through 7 days of calm and dark - see this post.)
    • Tons of extra transmission lines to remote generating sites.
    • Lots of empty transmission line capacity waiting for favourable weather. 
  • Our grid will generally be less stable (easy to manage) with increased supply-side variability in addition to existing demand-wide variability. 
  • Dispatchable assets will wear out faster as they ramp up and down harder to match variable generation, requiring more maintenance and/or early replacements. 
  • Prolonged or indefinite lifespan extensions of natural gas generation, which could be costly due to escalating carbon prices and/or possible 2035 federal regulations requiring we discontinue using gas for electricity or retrofit plants with expensive carbon capture equipment

Books that should be required reading for any electricity policymaker and/or wannabe energy pundit are: 

To close this section, I want to re-emphasize that the further we peer into the future, the hazier it gets. 

This model is not the definitive plan, merely a simple extrapolation projects and lifespans we know about today, waiting for questions to be answered and more projects to be proposed to fill in the blanks. 

So, let's try to fill in some of the blanks! 

A Nuclear Future for Saskatchewan

Let's imagine a future where we prioritize and value these qualities for electricity:

  • Reliability 
  • Affordability 
  • Carbon-free

Reliability should be the top priority for any grid. If electricity is not reliable, it is therefore unaffordable: for unreliable electricity generation the grid operator must provision expensive redundancies and raise rates to cover those costs. Or they make reliability the responsibility of consumers, in effect forcing people to procure backup generators, solar panels, batteries, etc. to keep their service reliable. Nuclear power plants in North America run at ~95% capacity factor: that's reliability.

Affordability is tightly coupled to reliability. A predictable, boring grid is an affordable grid. Despite the high capital (e.g. construction) costs of nuclear, operating costs are low and extremely predictable. Thanks to the long asset life of nuclear plants (60+ years), we can "mortgage" construction costs over the life of the asset. Nuclear fuel prices do not fluctuate like the price of fossil fuels, and the fact the fuel is so energy-dense means that only a little bit of fuel is required. (See also: my post on SMR costs)

Carbon-free is the icing on the cake. Nuclear energy has the smallest land footprint per unit of electricity delivered, and uses an extraordinarily small quantity of raw materials, rare minerals, and fuel to unlock a tremendous amount of atomic energy, all to boil water, make steam, spin a turbine, and generate power without emitting carbon. (What about the waste? see the last few sections of my BWRX-300 post)

Nuclear energy is clean energy. Image source: Glex

Using my simple model, I've calculated that Saskatchewan would need to build not four, but eleven small modular reactors between 2034 and 2049 to bring our dispatchable power back above our  (conservatively) forecasted peak load:

The only other change I made to this model was to delete 1,500 MW of wind/solar power from being brought online in 2035, since it would no longer be required. Think about it: why would we generate zero-carbon electricity from gigantic, energy-dilute farms harvesting sporadic energy from the weather when we could simply manufacture electrons in a dozen tiny factories? 

We should be concerned about what happens in 2030 when coal is phased out, and dispatchable power capacity drops below a conservatively-estimated peak load. Yes, coal is bad for emissions - but at least it's available on demand. Unless new dispatchable projects are announced to fill that gap, we will have to depend on a mix of intermittent renewables and expensive imports (see this post - imports are 2-3x more expensive than generating electricity at home) to meet demand during our most challenging conditions. 

It would be ideal (and optimistic) to pull SMR commissioning forward from 2034 to 2030, so that some nuclear can offset the last of the coal. 

Eleven SMRs!? I'll have some of what this guy's smoking.

Is it so impossible to build eleven SMRs by 2050? That's 27 years away. 

Between 1999 and 2028 SaskPower (or a private IPP supplying SaskPower) will have built or refurbished a natural gas generating station every 2.4 years. That's exactly the pace of this SMR scenario: natural gas and SMRs are both thermal generating technologies with very similar components (except for the heat source!). Perhaps we could build them at existing coal and gas plant sites as they retire, and take advantage of existing infrastructure. 

Eleven SMRs is also about the number of coal and gas plants we have in service today. Perhaps we could take advantage of existing infrastructure (roads, sites, transmission) as those assets are retired. 

France took 15-20 years to revolutionize their energy mix with over 50 zero-carbon nuclear plants:

Source: Wikipedia, Nuclear Power in France

The province of Ontario has the one of the cleanest grids in the world with over 10 GW of nuclear generating capacity providing (on average) 60% of the province's electricity. Ontario commissioned 22  CANDU reactors in 22 years from the early 70s to the early 90s. 

Ontario's Independent Electricity System Operator (IESO) recently released their Pathways to Decarbonization Report which calls for 17,800 GW of new nuclear capacity by 2050 (vs. the meagre 3,300 GW I am proposing for SK), in addition to boat-loads more power and storage from other sources. 

Source: IESO Pathways to Decarbonization, p31. Also note: the IESO is forecasting just 160 MW of their existing 5,000 MW of wind generation capacity will still be functional in 2050

I can hear anti-nuclear readers going, "ah-ha! that graphic also shows lots of wind, solar, hydrogen, and more. This guy's just obsessed." Keep reading! 

What does the Canadian Climate Institute have to say?

In their 2022 report, Bigger, Cleaner, Smarter: Pathways for Aligning Canadian Electricity Systems With Net Zero (PDF link) the Canadian Climate Institute notes that "a range of studies conclude that achieving net zero will require" installed generation capacity to grow by 2.2x to 3.4x by 2050, and that this is a "safe bet" action to pursue (p21):

Source: Bigger, Cleaner, Smarter: Pathways for Aligning Canadian Electricity Systems With Net Zero (PDF), Canadian Climate Institute, p22

Holy smokes! This suggests my growth model is extremely conservative

If we smash together these two forecasts - my simple 11-SMR scenario and the CCI's prediction of required ~2.5x growth in installed capacity, we can see a huge generation shortfall of 5,780 GW in Saskatchewan by 2050 - a shortfall the size of our 2023 grid (!):

Note: I've simplified the CCI's non-linear model from the previous image into a linear line because... it is simple.

What conclusions can we draw from this? 

  • One is: it may be that eleven SMRs isn't nearly enough. Maybe we need something like 30 SMRs! That would close the 2050 gap. 

  • A second might be: we need lots more of everything: wind, solar, batteries, hydrogen, and hydro (if we can find it). 

  • A final conclusion is: whether future demand is modest or dramatic, we need be decisive, focus our resources, and start building in order to meet our future needs. If our power future is one of energy shortages, that will be expensive for families and unworkable for industries. 

The "lots of everything" conclusion is probably politically palatable - it keeps everyone happy. If we pursue that path we need to understand and accept the impact of massive intermittent and/or dilute energy deployments, which require additional raw materials (per kWh), redundancies, complexities, and interconnects compared to dispatchable sources, and end up costing more in the long term. We also need to understand that only nuclear and large-scale hydroelectric have the high capacity, long-term, zero-carbon power we need. 

Over at I've built a Fossil Friday dashboard that asks "what if we converted all of our existing fossil capacity from the last 7 days to nuclear OR wind OR solar?", illustrating the requirements of energy-dense vs. energy-dilute sources: 

This is why nuclear energy is such an elegant solution for the Saskatchewan: concentrated, energy-dense electricity factories that maximize the ultimate energy output of construction materials, energy, and fuel inputs used to build and operate them. 


Four Saskatchewan SMRs by 2050 isn't enough. 

Eleven Saskatchewan SMRs just barely satisfies a back-of-napkin-math scenario predicated on very conservative assumptions. 

If the Canadian Climate Institute's forecast is credible, if we manage to build 11 SMRs by 2050 we'll still face a generating capacity shortfall equal to our entire 2023 grid! 

We need to be decisive, make commitments, and start planning and building SMRs now. 

If you've made it this far - thank you so much for reading. I always welcome constructive feedback, corrections, and suggestions. Leave a comment or fire me a Tweet/DM/email. 

Data and calculations.