Can battery energy storage systems replace gas peaker plants for grid stability? The evidence says yes - and it's already happening at the exact scale Nova Scotia needs.
The two proposed 300 MW gas plants at Marshdale and Salt Springs are not conventional power plants. According to the RFP functional specifications, they would operate as synchronous condensers more than 70% of the time - spinning their turbines without burning fuel to provide grid stability services like inertia, voltage regulation, and frequency support.
They would generate electricity less than 25% of the time, functioning as spinning reserve for peak demand periods. In other words, Nova Scotia is being asked to build two fossil fuel plants primarily to provide services that do not require fossil fuel.
Toronto's Portlands Energy Centre was sold as a peaker that would run "a few hours per month." By 2023 it was running 21 hours/day at 87.5% capacity factor in summer. Emissions quadrupled from 188,406 tCO2e (2017) to 807,500 tCO2e (2023). The contract was extended to 2046 without a public hearing. The proposed Marshdale tolling agreement has the same gap: no cap on operating hours.
Grid-forming (GFM) battery energy storage systems can provide the same grid stability services as synchronous condensers - inertia, frequency response, voltage regulation - plus they store and dispatch energy. A March 2025 joint study by the Energy Systems Integration Group (ESIG), GridLab, and American Transmission Co. (ATC) used real-world electromagnetic transient (EMT) modelling on an actual transmission network and found:
| Capability | Grid-Forming BESS | Synchronous Condenser |
|---|---|---|
| Reactive power | Inverter-based, adjustable to 100% | Mechanical, limited by design |
| Voltage regulation | Fast (milliseconds) | Slow (seconds) |
| Inertia | Synthetic via GFM controls, configurable | Physical from rotating mass |
| Frequency support | Active frequency response + grid-forming | Limited to inertia + reactive power |
| Black start | Yes (with GFM inverters) | Inherent to synchronous machines |
| Energy storage | Yes - stores and dispatches real power | No - reactive power only |
| Maintenance | Lower (mostly electronic) | Higher (rotating equipment) |
| Emissions | Zero | Zero (but paired gas turbine emits) |
| Water consumption | Zero | Zero (but paired SAC turbine uses 175 m³/hr) |
The ESIG study concluded that GFM BESS is a "do no harm" solution: it provides grid-stabilizing benefits in weak grids and has no adverse impact in strong grids. Controls require no retuning between grid conditions. The study specifically found that GFM batteries can defer more costly solutions including synchronous condensers and transmission buildout.
Nova Scotia's grid is relatively small and isolated - classic weak grid characteristics where GFM BESS provides the greatest benefit. NS Power's own modelling projects ~30% wind curtailment by 2030, climbing past 40% by 2040. GFM BESS reduces curtailment by 80% compared to conventional battery systems (ESIG study: 250 MW curtailed with grid-following vs 50 MW with grid-forming).
In March 2025, Europe's largest battery energy storage installation became operational at Blackhillock, Scotland. At 200 MW, it is roughly two-thirds the scale of each 300 MW gas plant proposed for Nova Scotia — and a 300 MW grid-forming BESS at Kilmarnock South followed in January 2026, matching the per-plant scale exactly.
The Blackhillock project was the inaugural installation under the UK's National Grid ESO Stability Pathfinder programme, which was specifically designed to procure grid stability services - short-circuit level and synthetic inertia - from non-fossil sources. It uses Wartsila Quantum BESS with the GEMS Digital Energy Platform and is owned/operated by Zenobe.
Its purpose is directly relevant to Nova Scotia: it provides the stability services needed to integrate wind power from three major North Sea offshore wind farms, replacing grid services previously provided by fossil fuel plants.
"Our batteries at Blackhillock are the first to use advanced power electronics to enable a much higher uptake of renewable power on the grid." - James Basden, Founder, Zenobe
Ten months after Blackhillock went live, Zenobe powered up its second grid-forming BESS in Scotland: Kilmarnock South, 300 MW / 600 MWh. Operational since January 2026, it matches the rated power of each proposed Nova Scotia gas plant — and it is doing the same grid stability job those gas plants are being built to do.
Like Blackhillock, Kilmarnock South was procured under NESO Stability Pathfinder 2 — the National Energy System Operator’s competitive tender for grid stability services from inverter-based resources. The project provides short-circuit level support and synthetic inertia using SMA grid-forming inverters, the same technology as Blackhillock. Wartsila supplied the Quantum energy storage system. EDF Energy serves as route-to-market provider, dispatching the asset via the Kraken platform.
Where Blackhillock proved that a transmission-connected grid-forming BESS could provide stability services at all, Kilmarnock South proves it can be done at the scale of a major peaker plant. Zenobe alone has now secured roughly 65% of the ~6.8 GVA of total inertia procured under NESO Stability Pathfinder 2 — battery storage is no longer an experiment in the UK’s grid stability market. It is the dominant resource type.
The case for the Marshdale and Salt Springs gas plants rests on the claim that batteries cannot provide the inertia and stability services these plants would deliver as synchronous condensers. Kilmarnock South is the operational, commercially-procured, working refutation of that claim — at the same rated power as each proposed NS gas plant, in a grid (Scotland) with similar characteristics: relatively small, isolated, integrating large amounts of wind power, transitioning away from synchronous fossil generation.
It is also relevant that this happened on the same NESO Stability Pathfinder 2 procurement as Blackhillock. The UK didn’t have to invent a special framework for grid-forming BESS — it ran a technology-neutral procurement for grid stability services and battery storage won. The IESO Nova Scotia gas RFP, by contrast, is scoped to natural gas only. Battery storage cannot bid even though it is now demonstrably capable of providing the same services.
There is a shovel-ready battery storage project proposed for Nova Scotia right now. NRStor Trenton Energy Storage is a joint venture between NRStor Inc., the Wskijinu'k Mtmo'taqnuow Agency (WMA, owned by all 13 Nova Scotia Mi'kmaw First Nations), and Aecon Concessions.
The project would be located in the Town of Trenton, Pictou County - adjacent to the existing Trenton Generating Station (a coal plant retiring by 2030) on approximately 7 acres of already industrially-zoned land with an existing grid connection. It would store surplus renewable energy and release it during peak demand, replacing coal-fired generation capability as Trenton coal retires.
The case for the Marshdale and Salt Springs gas plants rests on Nova Scotia Power's 2022/2023 Evergreen Integrated Resource Plan (IRP), which used the PLEXOS capacity expansion model run by consulting firm E3. The model identified the need for "fast acting generation in the form of combustion turbines" to balance intermittent renewables and maintain reliability. The IESO Nova Scotia procurement followed directly from this finding.
But the IRP modelling has a structural assumption that determines its conclusion before any optimization runs: battery energy storage was modelled exclusively as 4-hour lithium-ion energy storage. Grid-forming BESS does not exist as a resource option in the model.
The model enforces a "minimum synchronous inertia constraint" that only gas combustion turbines and synchronous condensers can satisfy. The "Domestic Integration" option in the model is always a paired bundle: 200 MW BESS + 200 MVA synchronous condenser. There is no option for grid-forming BESS providing both energy storage and inertia from the same asset. The optimizer is therefore structurally forced to select gas turbines or synchronous condensers for inertia — not because batteries lost a fair comparison, but because the technology that's now operational at Blackhillock didn't exist as an input.
The strongest evidence that grid-forming BESS can replace gas-plus-synchronous-condenser combinations was published after the IRP modelling was complete. The IESO is now procuring gas based on a model that didn't — and couldn't — consider the alternative.
Even within the IRP's own constraints (where BESS provides energy storage only), the cheapest scenario is one where battery costs decline faster. NS Power's "Battery Decline" sensitivity scenario produced the lowest net present value of partial revenue requirement of any scenario tested — approximately $2.8 billion lower than the base case over the 2025–2050 horizon. Even handicapped to energy-only operation, more batteries reduced system cost.
The same modelling shows wind and solar curtailment rising to over 4,700 GWh per year by 2050. When a wind or solar farm is curtailed, it’s told to stop producing because the grid can’t absorb the power, and ratepayers still pay for it. Storage would solve this. The ESIG study found grid-forming BESS cuts curtailment by 80% — but it wasn’t available to the optimizer.
IESO Nova Scotia's VP of Planning and Procurement, Chris Milligan, told CBC News: "Batteries are an important part of the future resource mix in Nova Scotia. Our plan for that is to do a competitive procurement for energy storage resources and we plan to start that process near the end of 2026."
By that timeline, the gas plant tolling agreements will likely already be signed and ratepayers locked in for 20 to 27 years. A battery procurement that begins after gas contracts are signed cannot displace gas — it can only add to it. The shovel-ready NRStor Trenton project, which could compete with gas if procured concurrently, will instead be evaluated against a fait accompli.
The IESO is also currently hiring a consultant to develop its own IRP, with proposals due April 10, 2026. The province is committing to a 25-year fossil fuel build-out before its own independent system operator has completed its first integrated resource plan.
When NS Power finalized its Evergreen IRP in August 2023, grid-forming BESS at transmission scale didn’t exist anywhere in the world. Blackhillock was still under construction. The ESIG/GridLab study confirming GFM BESS meets grid-forming specifications had not been published. NESO’s Stability Pathfinder 2 procurement, which validated the entire commercial model, was still in early stages. If NS Power had moved straight from the IRP recommendation to procurement in late 2023, gas plant contracts could have been signed by 2024 — before the alternative existed in any operational form.
That procurement didn’t happen. In April 2024, Nova Scotia passed Bill 404 (the Energy Reform Act) and transferred procurement responsibility from NS Power to a brand-new Independent Energy System Operator. The IESO had to be built from scratch — staff, governance, board appointments, market rules, procurement framework, all of it. It became the 10th independent system operator in North America. The transition consumed most of 2024 and 2025. The first gas procurement call didn’t go out until October 2025, more than two years after the Evergreen IRP recommended the gas plants.
In that two-year gap, grid-forming BESS went from “promising lab demonstration” to “operational at transmission scale.”
By the time the IESO’s draft gas RFP landed in March 2026, the cleaner, cheaper alternative the 2023 IRP couldn’t model had become a proven, deployed technology with multiple reference projects at the exact scale Nova Scotia needs.
NS Power publishes annual IRP Action Plan Updates that summarize the state of the resource plan. The April 2025 Update was published one month after Blackhillock went operational and one month after the ESIG/GridLab study was released. It is a 42-page document. It mentions Blackhillock zero times. It mentions the ESIG study zero times. It mentions grid-forming inverters zero times.
The April 2025 Update continues to recommend gas combustion turbines, citing the same 2023 Evergreen IRP modelling unchanged. The document does acknowledge that “Updated information... can be assessed in the future through the Evergreen IRP process” — in other words, NS Power explicitly knew the modelling could be refreshed and chose not to. Six months after the April 2025 Update, the IESO issued the gas RFEoI on that basis.
The procurement delay caused by the IESO transition accidentally created the Leapfrog Opportunity. Nova Scotia can now skip past legacy gas peaker technology entirely and go directly to grid-forming BESS — the same way some countries skipped landlines and went straight to mobile phones. The technology matured during the delay.
There is a very easy way to take advantage of this opportunity. Make the RFP technology-neutral. Define the performance requirements — MW capacity, response time, inertia contribution, availability — and let any technology that meets them compete. If gas peaker plants are still truly the least-cost, most reliable option, they will win a technology-neutral procurement on merit.
Some potential objections to a technology neutral RFP and why they aren't valid in this case:
| Objection | Response |
|---|---|
| “It would delay the timeline.” | It's true, it would add some time... but given what we're looking at, it's worth it. |
| “It’s harder to evaluate diverse bids.” | Technology-neutral energy procurements are standard practice across Canada and internationally. Ontario's IESO has run a technology-neutral Capacity Auction since 2020, and its LT2 RFP lets wind, solar, hydro, and storage compete head-to-head. BC Hydro's 2025 Call for Power does the same. So does the UK's Stability Pathfinder — the procurement that selected Blackhillock. There is no reason we can't do the same in Nova Scotia. |
| “Alternative technologies aren’t proven at this scale.” | Kilmarnock South is a 300 MW grid-forming BESS that is similar in scale of each proposed Nova Scotia gas plant — operational since January 2026. If a technology can’t meet the performance requirements, it won’t bid. That’s the market working as intended. |
| “Batteries can’t run for extended periods.” | Then set minimum duration and availability requirements in the RFP and let bidders demonstrate they can meet them. A worst-case scenario with supporting napkin math is offered below: under stacked worst-case assumptions, five consecutive extreme cold days, zero wind, zero solar, zero demand response, a 300 MW / 4-hour battery recharging overnight looks like it could cover 99.5% of peak demand. If the IESO believes batteries can’t meet the capacity requirement, they should publish their analysis showing why instead of just saying it will not work. |
| “The IRP already determined gas is needed.” | The IRP determined that dispatchable capacity and grid stability services are needed. It recommended gas because grid-forming BESS wasn’t an available resource type in the model. The technology has since been demonstrated at scale. A technology-neutral RFP lets the market test whether the IRP’s 2023 conclusion still holds in 2026. |
If gas peaker plants are still truly the best option in 2026, make the RFP technology-neutral and let's see.
All dollar figures in this section are Canadian dollars. I'm using two industry-standard lifecycle cost metrics that bundle everything — capital, financing, O&M, fuel, and carbon — into a single number showing what each megawatt-hour of delivered energy costs over the asset’s life.
The everyday analogy is the all-in cost per kilometre of driving a car. If you add up the purchase price, financing, insurance, maintenance, and fuel over the car’s lifetime and divide by the total kilometres driven, you get a single number — say, 45¢/km — that tells you what each kilometre actually costs you. That lets you compare a gas car against an electric car on equal footing, even though their cost structures are completely different (gas cars spend more on fuel, EVs spend more upfront). LCOE and LCOS do the same thing for power plants:
Both resolve to a single unit — $/MWh — which lets us compare batteries and gas directly. These are the metrics Lazard publishes annually in their widely-cited LCOE+ report, and they are the industry standard for comparing across technologies.
Every LCOS/LCOE figure on this page is built the same way: annual fixed cost spread across the MWh delivered per year, plus a variable cost per MWh. The inputs differ by technology — batteries use charging electricity, gas plants burn fuel and pay for carbon — but the structure is identical:
LCOS or LCOE = (Annual fixed cost ÷ Annual MWh delivered) + Variable cost per MWh
The Fixed $/MWh column in the table below is the first term of that formula — it converts the annual $/kW-yr fixed cost into dollars per MWh of energy delivered, by dividing by the number of MWh each kW of capacity produces in a year:
Fixed $/MWh = ($/kW-yr) ÷ (8,760 hours × capacity factor ÷ 1,000)
The 8,760 is just the number of hours in a year; the ÷1,000 converts kWh to MWh. Worked example: NRStor’s 8-hour battery at 33.3% capacity factor delivers 8,760 × 0.333 ÷ 1,000 ≈ 2.92 MWh per kW per year. At $175/kW-yr fixed cost, that’s $175 ÷ 2.92 ≈ $60/MWh — which is the value in the Fixed $/MWh column for the NRStor base case. Every row in the table is computed the same way. Add the variable cost and you get the LCOS/LCOE in the last column.
The notes below explain where each input comes from: $/kW-yr (annual fixed cost), capacity factor (batteries and gas plants are treated differently), and the variable cost per MWh (charging electricity for batteries, fuel + carbon + O&M for gas plants).
A note on annual fixed costs: the $/kW-yr figures in the first column of the table are the annualized capital recovery plus fixed O&M that ratepayers pay whether or not the resource dispatches. Each project’s fixed cost is sourced differently:
A note on capacity factor for batteries: unlike gas plants (which can run any number of hours), a battery’s output is limited by its duration. NREL’s standard assumption for grid-scale batteries is one full cycle per day — charge overnight, discharge during the day’s peak hours. Under that assumption, a 4-hour battery has a capacity factor of 4÷24 = 16.7%, and an 8-hour battery has 8÷24 = 33.3%.
A note on capacity factor for gas plants: a gas plant’s capacity factor is an operational choice — it can theoretically run anywhere from 0% to 90%+ of the year depending on how often the operator dispatches it. The gas plant scenarios in the table use two reference values:
A note on battery charging cost scenarios: the cost of electricity to charge the battery depends on when it charges and what else is running on the grid at that moment. Unlike a gas plant — which must burn whatever fuel is in the pipe at the moment it dispatches — a battery operator chooses when to charge, and would simply not charge during the rare hours when gas is on the margin at $200–300/MWh. The table therefore uses two reference values, both representing realistic charging windows:
A note on gas fuel cost scenarios: Nova Scotia imports natural gas via the Maritimes & Northeast Pipeline, which is constrained during winter. Gas prices at the Algonquin Citygate (the relevant market hub) are more volatile than almost any other North American trading point. The table uses three reference values:
All gas scenarios use the actual marginal carbon cost a new NS gas plant would face in 2026 under Nova Scotia's Output-Based Pricing System (OBPS). Under OBPS, a new gas plant only pays carbon on emissions above a fuel-specific benchmark that tightens each year. A new gas peaker emits ~0.5 tCO&sub2;/MWh; the 2026 OBPS benchmark for new gas is 0.164 tCO&sub2;/MWh; the gap (0.336 tCO&sub2;/MWh) is charged at the federal benchmark price ($110/tonne in 2026) — about $37/MWh. The benchmark tightens to zero by 2030 (compliance period 8), at which point the full 0.5 tCO&sub2;/MWh is charged at $170/tonne — ~$85/MWh, with continued escalation as the federal price rises after 2030.
The table below shows both technologies under the same calculation framework, plus sensitivity scenarios showing how changing the input assumptions affects the result. Battery LCOS stays in a narrow band across all reasonable scenarios. Gas LCOE varies by a factor of 5 or more depending on fuel price and utilization — and ratepayers pay every dollar of that volatility under the tolling agreement.
| Project / scenario | Annual fixed cost | Capacity factor | Fixed $/MWh | Variable $/MWh | LCOS / LCOE |
|---|---|---|---|---|---|
| NRStor Tantramar RFP (8-hr battery, base case: $83/MWh charging) | $175/kW/yr | 33.3% | ~$60 | ~$94 ($83/MWh ÷ 0.88 round-trip efficiency) | ~$154/MWh |
| — same project, cheap charging ($25/MWh curtailed wind) | $175/kW/yr | 33.3% | ~$60 | ~$28 ($25/MWh ÷ 0.88 round-trip efficiency) | ~$88/MWh |
| NS Power e-STORAGE (4.7-hr battery, base case: $83/MWh charging) | $242/kW/yr | 19.6% | ~$141 | ~$94 ($83/MWh ÷ 0.88 round-trip efficiency) | ~$235/MWh |
| — same project, cheap charging ($25/MWh curtailed wind) | $242/kW/yr | 19.6% | ~$141 | ~$28 ($25/MWh ÷ 0.88 round-trip efficiency) | ~$169/MWh |
| NS gas plant (base case: 25% CF from EA, CA$18/MMBtu typical winter fuel, OBPS 2026 carbon) | $280/kW/yr | 25% | ~$128 | ~$208 ($164 fuel + $37 carbon + $7 O&M) | ~$336/MWh |
| — 25% CF, bad winter month (CA$28/MMBtu, Jan 2022 actual) | $280/kW/yr | 25% | ~$128 | ~$299 ($255 fuel + $37 carbon + $7 O&M) | ~$427/MWh |
| — 25% CF, polar vortex days (CA$110/MMBtu, 2014/2018 actual peaks) | $280/kW/yr | 25% | ~$128 | ~$1,045 ($1,001 fuel + $37 carbon + $7 O&M) | ~$1,173/MWh |
| — 10% CF (Lazard peaker default), CA$18/MMBtu typical winter fuel | $280/kW/yr | 10% | ~$320 | ~$208 ($164 fuel + $37 carbon + $7 O&M) | ~$528/MWh |
| — 10% CF, bad winter month (CA$28/MMBtu, Jan 2022 actual) | $280/kW/yr | 10% | ~$320 | ~$299 ($255 fuel + $37 carbon + $7 O&M) | ~$619/MWh |
| — 10% CF, polar vortex days (CA$110/MMBtu, 2014/2018 actual peaks) | $280/kW/yr | 10% | ~$320 | ~$1,045 ($1,001 fuel + $37 carbon + $7 O&M) | ~$1,365/MWh |
The asymmetric risk: across every realistic scenario, NRStor’s 8-hour battery LCOS sits between ~$88 and ~$154/MWh — a range of about $66. The gas plant LCOE ranges from ~$336 to over $1,365 — a range of about $1,029, or roughly 16× wider. Batteries are bounded by a narrow range because the operator picks the cheap hours; gas plants have to burn whatever fuel is flowing in the pipe at the moment they dispatch. Under the tolling agreement, ratepayers absorb 100% of that fuel price volatility.
Gas scenarios above use the 2026 OBPS marginal carbon cost (~$37/MWh). By 2030 the OBPS benchmark for new gas drops to zero, raising the marginal carbon cost to ~$85/MWh — an additional ~$48/MWh on every gas scenario above (e.g., base case $336 → $384; worst case $1,365 → $1,413). The federal carbon price continues rising after 2030 (~2% real per year), so a plant built in 2027 would see escalating carbon costs across its 25–30 year operating life.
The LCOE/LCOS numbers above bundle everything into a single figure, but it helps to see what drives the difference. The LCOE for gas is dominated by costs that simply do not exist for batteries:
| Variable cost | Battery | Gas peaker |
|---|---|---|
| Energy input / fuel | Electricity to charge — cost depends on when it charges (a scheduling choice, not a commodity exposure). Approaches zero when charging from curtailed renewables or overnight surplus. | Natural gas at market price. NS is at the end of a constrained pipeline; Algonquin Citygate winter daily peaks have hit ~US$80/MMBtu (~CA$110) in 2014 and 2018, while typical recent winter averages run US$13–14/MMBtu (~CA$18–19) vs Lazard's US$3.45/MMBtu (~CA$4.76) North American benchmark. Volatile, unpredictable, and permanently tied to global commodity markets. |
| Carbon pricing | Zero — no combustion, no emissions. | Full exposure. Canada’s carbon price is legislated to reach $170/tonne by 2030 and continues rising thereafter. At 25% capacity factor, a 300 MW plant faces tens of millions per year in carbon costs. |
| Efficiency losses | 10–12% round-trip loss per cycle (charge 100 MWh, deliver ~88 MWh). Predictable and fixed. | Heat rate losses inherent in combustion (~40–45% thermal efficiency for aeroderivatives). Worsens with age and ambient temperature. |
| Water consumption | Zero. | Up to 175 m³/hr per plant (SAC engines with water injection for NOx control). Drawn from local groundwater. |
| Startup / cycling wear | Negligible — batteries can cycle thousands of times with minimal wear. No thermal stress. | Each start-stop cycle wears hot gas path parts, combustion liners, and thermal coatings. Maintenance intervals are triggered by start count. |
| Emissions consumables | None — no combustion. | SCR systems (if used) consume urea or ammonia reagent continuously during operation. |
| Cost trend over 25 years | Declining — more renewables = more surplus energy for low-cost charging. Battery “fuel” gets cheaper over time. | Rising — carbon price legislated to increase, gas prices volatile, water and emissions constraints tightening. |
Who pays for gas variable costs? Under the tolling agreement structure proposed for Marshdale and Salt Springs, fuel costs and carbon costs are passed through directly to ratepayers — the plant operator has zero exposure to gas price volatility or carbon price increases.
The dominant battery variable cost — charging electricity — is under the operator’s control. In Nova Scotia, two charging windows keep costs low:
NS Power’s e-STORAGE program (150 MW / 705 MWh across three sites) was approved at $354M in capital costs ($237.7M net of $116.6M in federal funding), or approximately $2,360/kW installed — nearly double the current North American market price of ~$1,208/kW.
Several factors explain the gap:
The takeaway: the e-STORAGE cost reflects 2023 market conditions and first-mover premiums, not what batteries cost today. Any future battery procurement in Nova Scotia should come in substantially lower.
Battery storage is not a silver bullet. Lithium-ion batteries are commercially viable for 2 to 10 hours of storage. They cannot store summer energy for winter use. They are not a direct replacement for baseload generation that runs 24/7. Nova Scotia’s coal retirement requires a portfolio approach: wind, solar, batteries, interprovincial transmission, and potentially some dispatchable generation.
But the most common objection to batteries in Nova Scotia is not about seasonal storage or baseload. It is about multi-day winter cold snaps: what happens when it is bitterly cold and the wind stops blowing for days at a time? Can a battery survive that?
A 4-hour battery at 300 MW provides 1,200 MWh per cycle and must recharge between dispatches. It cannot run for five days straight, but it does not need to. Demand drops every night, and the rest of the fleet has enough spare capacity to recharge it before the next day’s peak. The question is whether the math actually works.
A note on what follows: what you are about to read is napkin math, not a sophisticated grid model. I built this simple hour-by-hour simulation because I could not find any published analysis showing that batteries can’t handle NS’s winter capacity needs, just assertions they can’t. This analysis uses public data and transparent assumptions to show that batteries look like they are at least in the ballpark for the job given the stacked worst-case assumptions. It is an invitation to comment and respond, not a definitive answer. It would be great if anyone claiming batteries are inadequate for NS’s winter reliability could publish their own analysis showing the work. And please send me a link!.
Even with zero wind for 5 consecutive extreme cold days, a 300 MW / 4-hour battery recharges every night and looks like it could cover 99.5% of peak demand. Nova Scotia's non-wind supply is projected to provide ~2,310 MW by winter 2029/2030. On extreme cold nights, ~217 MW of spare capacity is available for charging. That is enough to refill the battery overnight and have it discharge during the next day's peaks. The only shortfall: 6 MW in a single hour (10 PM), when the battery empties after covering the entire evening peak. And this analysis stacks four worst-case assumptions on top of each other. If one of those assumptions was relaxed even slightly, it eliminates even that tiny gap.
The flat crimson dashed line across the top of the chart. This is the total firm generating capacity Nova Scotia will have by winter 2029/2030 after coal retirements, excluding the proposed IESO gas plants and all wind and solar. Wind and solar are excluded because this is the worst case: five consecutive days with zero renewable output.
| Resource | Capacity (MW) | Source |
|---|---|---|
| Hydro + Maritime Link | 494.8 | 10-Year Outlook, Figure 4 |
| Tufts Cove steam (Units 1–3) | 318 | 10-Year Outlook, Figure 4 |
| Tufts Cove combined cycle (Units 4–6) | 144 | 10-Year Outlook, Figure 4 |
| Lingan 1, 3, 4 (coal → HFO conversion) | 459 | 10-Year Outlook, Figure 7 |
| Point Tupper (coal → gas conversion) | 150 | 10-Year Outlook, Figure 7 |
| Port Hawkesbury biomass | 43 | 10-Year Outlook, Figure 4 |
| Existing oil CTs (Burnside, Tusket, Victoria Junction) | 231 | 10-Year Outlook, Figure 4 |
| IPPs (non-wind) | ~120 | 10-Year Outlook, Figure 4 |
| Subtotal generation | ~1,960 | |
| NS–NB reliability tie (import capacity) | 350 | Approved November 2025 |
| Total supply (no coal, no wind, no solar) | ~2,310 |
What is excluded and why: The proposed 600 MW of IESO gas plants (because those are what this analysis evaluates alternatives to). All wind and solar (because this is the zero-renewables worst case). NS Power’s approved e-STORAGE 150 MW BESS (because batteries are storage and cannot charge other batteries during a supply shortage). This is the generation fleet that would exist to both serve demand and recharge batteries.
What is retired by then: Trenton 5 (−150 MW, winter 2027/28), Lingan 2 (−148 MW, winter 2027/28), Point Aconi (−168 MW, winter 2029/30), Trenton 6 (−154 MW, winter 2029/30). Total retirements: −620 MW. These are already factored out of the 2,310 MW figure above.
Average winter demand (blue solid line): Actual hourly load data from NS Power’s OASIS system, averaged across all winter days (December through February 2024), projected to 2030 at the Outlook’s 1.1%/yr peak growth rate. Projected trough: ~1,556 MW at 4 AM. Projected evening peak: ~1,839 MW at 6 PM. The peak growth rate is applied to the entire curve, which is conservative — the Outlook says total energy demand is actually declining at −0.2%/yr while peaks grow. So this curve overstates average demand.
Extreme cold day (red dashed line): The average day’s shape is scaled so the peak matches NS Power’s gross peak forecast of 2,474 MW for 2030 (10-Year Outlook, Figure 3). The daily profile shape — when the trough and peaks occur, how the curve rises and falls — comes from the actual 2024 OASIS data. The same trough-to-peak ratio is kept (0.846) as the average day, giving an extreme overnight trough of approximately 2,093 MW.
The trough number is the single critical variable that determines whether this whole approach works. Every MW of overnight demand is a MW that can’t be used to charge the battery. A flatter profile (higher trough) means less surplus and a battery that can’t fully recharge before the morning peak. A deeper trough (lower overnight demand) means more surplus, more headroom, and the battery sails through. Everything downstream, the 1,270 MWh charged, the 1,118 MWh delivered, the 99.5% coverage depends on this one number. The sensitivity section below shows exactly how the outcome shifts as the trough moves.
On an extreme cold day with the 2,093 MW trough, the fleet has surplus capacity for 12 hours (overnight from roughly 11 PM through 7 AM, plus a brief dip mid-afternoon). Demand exceeds supply for the other 12 hours, creating a deficit of 1,124 MWh.
The 300 MW / 4-hour battery has 1,200 MWh of storage and approximately 88% round-trip efficiency — for every MWh drawn from the grid, about 0.88 MWh is delivered back. The natural unit to think about is one full charge cycle: 7 AM to 7 AM. The cycle begins with the battery full at 7 AM after the overnight refill. It discharges through the morning peak, takes a small top-up during the midday surplus, discharges hard through the evening peak, empties at 10 PM, and begins recharging an hour later when supply recovers, returning to 100% by 7 AM the next day. Across the cycle the battery draws 1,270 MWh from the grid (1,200 MWh overnight refill + 70 MWh midday top-up) and delivers 1,118 MWh back during the morning and evening peaks. The 152 MWh difference is round-trip loss. This covers all but 6 MWh of the deficit — one hour at 10 PM, when the battery hits zero after covering the entire evening peak. Because each cycle ends in exactly the same state it began (full battery, 7 AM), the pattern repeats indefinitely.
| Hour | Demand (MW) | Grid surplus | Battery action | State of charge |
|---|---|---|---|---|
| Cycle begins at 7 AM — battery full from overnight refill | ||||
| 07:00 | 2,283 | +27 MW | Full — surplus wasted | 1,200 MWh (100%) |
| Morning peak — battery discharges to cover deficit (delivers stored × 0.88) | ||||
| 08:00 | 2,388 | −78 MW | Discharge 78 MW | 1,111 MWh (93%) |
| 09:00 | 2,425 | −115 MW | Discharge 115 MW | 981 MWh (82%) |
| 10:00 | 2,414 | −104 MW | Discharge 104 MW | 862 MWh (72%) |
| 11:00 | 2,383 | −73 MW | Discharge 73 MW | 780 MWh (65%) |
| 12:00 | 2,360 | −50 MW | Discharge 50 MW | 723 MWh (60%) |
| 13:00 | 2,340 | −30 MW | Discharge 30 MW | 689 MWh (57%) |
| Brief mid-afternoon surplus | ||||
| 14:00 | 2,297 | +13 MW | Charge 13 MW | 702 MWh (59%) |
| 15:00 | 2,270 | +40 MW | Charge 40 MW | 742 MWh (62%) |
| 16:00 | 2,293 | +17 MW | Charge 17 MW | 759 MWh (63%) |
| Evening peak — battery discharges, covers nearly all deficit hours | ||||
| 17:00 | 2,379 | −69 MW | Discharge 69 MW | 680 MWh (57%) |
| 18:00 | 2,474 | −164 MW | Discharge 164 MW | 494 MWh (41%) |
| 19:00 | 2,473 | −163 MW | Discharge 163 MW | 309 MWh (26%) |
| 20:00 | 2,451 | −141 MW | Discharge 141 MW | 148 MWh (12%) |
| 21:00 | 2,414 | −104 MW | Discharge 104 MW | 30 MWh (2%) |
| 22:00 | 2,343 | −33 MW | Discharge 27 MW | 0 MWh — 6 MW unserved |
| Overnight refill — battery recharges from grid surplus | ||||
| 23:00 | 2,250 | +60 MW | Charge 60 MW | 60 MWh (5%) |
| 00:00 | 2,204 | +106 MW | Charge 106 MW | 166 MWh (14%) |
| 01:00 | 2,165 | +145 MW | Charge 145 MW | 311 MWh (26%) |
| 02:00 | 2,129 | +181 MW | Charge 181 MW | 492 MWh (41%) |
| 03:00 | 2,102 | +208 MW | Charge 208 MW | 700 MWh (58%) |
| 04:00 | 2,093 | +217 MW | Charge 217 MW | 917 MWh (76%) |
| 05:00 | 2,106 | +204 MW | Charge 204 MW | 1,121 MWh (93%) |
| 06:00 | 2,163 | +147 MW | Charge 79 MW (full) | 1,200 MWh (100%) |
| 7 AM next day — battery back to 100% — cycle repeats indefinitely | ||||
With 88% round-trip efficiency included, the battery covers every deficit hour except for 6 MW at 10 PM — when it empties after covering the entire evening peak. The overnight refill brings it back to 100% by 6 AM, with surplus still wasted at 7 AM as one cycle ends and the next begins. That 6 MW shortfall is 0.5% of the cycle’s total deficit. Maximum discharge in any single hour: 164 MW at 6 PM, well within the 300 MW rating.
Because each 7 AM–7 AM cycle ends in exactly the state it began (battery full), the simulation is genuinely steady-state. Every cycle is identical:
| Charged from grid | Delivered to grid | Unserved | SoC at 7 AM (end) | |
|---|---|---|---|---|
| Cycle 1 | 1,270 MWh | 1,118 MWh | 6 MWh | 1,200 MWh (100%) |
| Cycle 2 | 1,270 MWh | 1,118 MWh | 6 MWh | 1,200 MWh (100%) |
| Cycle 3 | 1,270 MWh | 1,118 MWh | 6 MWh | 1,200 MWh (100%) |
| Cycle 4 | 1,270 MWh | 1,118 MWh | 6 MWh | 1,200 MWh (100%) |
| Cycle 5 | 1,270 MWh | 1,118 MWh | 6 MWh | 1,200 MWh (100%) |
Because every cycle is identical and ends fully refilled, this pattern is sustainable indefinitely — there is no Cycle 6 problem, no Cycle 10 problem. The 152 MWh lost to RTE each cycle is the cost of storage; the grid still gets 1,118 of the 1,124 MWh it needs.
The critical variable is the overnight trough on extreme cold days. If demand doesn’t drop as much overnight — if the extreme day profile is “flatter” than the average day — there is less surplus to charge the battery. Here is how the result changes:
| Overnight trough | Trough/peak ratio | Charged from grid | Delivered to grid | Result |
|---|---|---|---|---|
| 2,050 MW | 0.829 | 1,168 MWh | 1,028 MWh | Battery works with large margin |
| ~2,091 MW | 0.845 | ~1,270 MWh | ~1,118 MWh | Break-even |
| 2,093 MW (used here) | 0.846 | 1,270 MWh | 1,118 MWh | 6 MWh unserved (99.5% covered) |
| 2,100 MW | 0.849 | 1,260 MWh | 1,109 MWh | Short by 32 MWh |
| 2,150 MW | 0.869 | 893 MWh | 786 MWh | Short by 489 MWh |
The 2,093 MW figure (ratio 0.846) is the data-driven default: the same trough-to-peak ratio observed in the actual 2024 OASIS winter data. At this trough, the battery looks like it could cover 99.5% of the deficit.
To make sure the battery scenario isn’t cherry-picked, four worst-case assumptions are baked in. Each one makes the scenario harder for the battery than reality is likely to be. Here’s what relaxing any one of them would do:
This analysis assumes Nova Scotia’s entire 611 MW wind fleet produces nothing for five consecutive days. In reality, winter is when wind performs best, with a typical capacity factor of 30–35%.
What relaxing it does: Even a token 2.6% capacity factor (16 MW average across the fleet) adds 382 MWh per day — enough to erase the 6 MWh shortfall many times over. At a realistic winter average of 30%, the fleet contributes ~4,400 MWh per day — nearly four times the entire daily deficit. The battery wouldn’t even need to fully discharge.
The 2,474 MW figure is NS Power’s gross peak forecast for 2030 — the single coldest hour of the single coldest day. This analysis applies that extreme peak to all five days in a row. Real cold snaps moderate: day 2 and day 3 are usually less severe than day 1.
What relaxing it does: A modest 5% softening on days 2–5 (peak drops from 2,474 to 2,350 MW) frees roughly 120 MW of headroom every hour — cutting the daily deficit by ~600 MWh and pushing coverage well past 100% for most of the cold snap.
NS Power already has demand response programs that can shed 50–100 MW during peak hours, plus industrial interruptible load contracts and time-of-use pricing. None of it is counted in this analysis.
What relaxing it does: Just 50 MW of demand response during the worst evening hour (18:00, where the deficit hits 164 MW) cuts that hour’s gap by nearly a third. Across the full deficit window, 50–100 MW of shaving could eliminate 400–800 MWh of the daily deficit on its own.
The whole analysis hinges on the overnight trough — the lower the trough, the more surplus there is to charge the battery. This analysis uses 2,093 MW, derived by applying the actual 2024 OASIS winter trough-to-peak ratio (0.846) to the 2030 peak forecast. That’s the historical observed ratio, not a hand-picked low.
What relaxing it does: A modest demand-management push could meaningfully deepen the overnight trough below 2,093 MW. The sensitivity table shows that even small movements matter: a trough of 2,050 MW (ratio 0.829, only 2% deeper than observed) eliminates the shortfall entirely and gives the battery comfortable headroom.
An obvious question: if relaxing any single assumption would erase the 6 MWh shortfall, why not just relax one and report the battery covering 100% of the deficit?
Because that would be cooking the books, and the whole point of this exercise is the opposite.
The 6 MWh shortfall — one hour, 6 MW, on day after day after day of peak-of-peak demand with zero wind, zero solar, and zero demand response — is what the math actually produces under the harshest reasonable inputs. Reporting it honestly is more useful than tuning the numbers until they hit a round figure. Three reasons specifically:
The point of this analysis isn’t to declare batteries the winner. It’s to show that the case for 25-year gas contracts can’t rest on “batteries can’t cover the worst day.” Under stacked worst-case assumptions, it looks like they could cover 99.5% of it, and any movement toward reality erases the rest.
A 300 MW gas peaker covers 100% of the daily deficit (1,124 MWh in the base case, more with a flatter profile) on extreme cold days. It runs when needed, idles the rest, and has no energy storage constraint. That is its advantage on the very worst days.
But for most of the year, it either sits idle or spins as a synchronous condenser — burning no fuel but providing no benefit that a grid-forming battery cannot also provide. The gas plant’s capacity factor on an extreme day in the scenario above is only 15.6%. In that same scenario, on an average winter day, the fleet covers demand without any peaker at all; neither the gas plant nor the battery needs to dispatch.
The question is whether the nightmare scenario of no wind and a handful of extreme days and no demand response, where the gas plant is needed and we don't have enough battery capacity, justifies locking ratepayers into 25 years of fuel costs, carbon costs, and water consumption.
The grid stability services that justify 70%+ of the proposed gas plants' operating time can be provided by grid-forming batteries, without emissions, without water extraction, without fuel cost exposure, and at declining costs. The remaining peaking generation need is also increasingly served by batteries at lower cost than gas. The question is not whether the technology exists. It does, it is proven, and it is being deployed at the exact scale Nova Scotia needs. The question is whether we will take the opportunity to choose future technology and not drag the past along for another 25 years.