An interactive tool for calculating and comparing the Levelized Cost of Energy (LCOE) for Solar, Wind, and Natural Gas across all 10 Canadian provinces. Built as part of a personal portfolio demonstrating applied energy economics and financial modeling for renewable energy infrastructure.
Live Dashboard → (https://lcoe-optimization-tool.streamlit.app/)
The Levelized Cost of Energy (LCOE) is the average cost, in dollars per megawatt-hour, of building and operating a power plant over its entire lifetime, divided by the total electricity it produces. It collapses very different cost and output profiles — a cheap-to-build but low-output solar farm, versus an expensive gas plant that runs almost constantly — into one comparable number, expressed the same way for every technology.
LCOE is the first number an energy company calculates before deciding whether a project is worth building at all. Get it wrong, and a "cheap" project can turn out to be the expensive one once real-world capacity factors and financing costs are accounted for. This tool lets you build that number yourself — pick a province and technology, adjust the real inputs, and see exactly how each one moves the final cost.
- Select any of Canada's 10 provinces
- Select a technology: Solar, Wind, or Natural Gas
- Adjust real inputs — Capital Expenditure, Fixed O&M, Capacity Factor, Discount Rate, Project Life, and (for gas) Fuel Cost — and see the resulting LCOE calculated live
- View a sensitivity tornado chart showing which input matters most for each technology/province
- Toggle between $/MWh and $/kWh display
- Read a dynamic "why this number?" insight explaining each result in plain English
Following the standard NREL/IESO methodology:
LCOE = (CapEx × CRF + Fixed O&M) / (AEP_net / 1000) + Variable O&M + Fuel Cost
Where CRF (Capital Recovery Factor) is derived from a real discount rate and project life:
CRF = r(1+r)ⁿ / ((1+r)ⁿ − 1)
The formula was validated against IESO's (Ontario's grid operator) March 2024 published LCOE figures for Wind, Solar, and Natural Gas. Initial testing with a single shared financing rate came within 1–5% — investigation revealed IESO uses technology-specific fixed charge rates (Wind 6.40%, Solar 6.11%, Gas 6.53%, not one shared rate). Refitting with these brought results to within 0.05 $/MWh of IESO's published numbers.
To make discount rate and project life independently adjustable (rather than baked into one fixed rate), the formula was extended to derive financing cost via CRF. Reverse-solving for the discount rate each technology implied — tested at both 20-year and 25-year project life — found that 25 years produced discount rates (3.56–4.19%) landing almost exactly on NREL's independently-published real WACC range (3.66–4.01%) for reference wind projects, external validation that 25 years is the correct assumption.
| Data | Source | Coverage |
|---|---|---|
| CapEx, O&M, Capacity Factor, LCOE | IESO 2024 Annual Planning Outlook — Resource Costs and Trends | Ontario (officially sourced) |
| CapEx | AESO 2024 Long-Term Outlook | Alberta (officially sourced) |
| Capacity factor (8 remaining provinces) | Reasoned estimates from documented Canadian wind/solar resource geography and provincial generation mix | BC, SK, MB, QC, NS, NB, PE, NL |
Research finding worth stating plainly: IESO and AESO appear to be the only two Canadian grid operators publishing standardized, technology-level $/kW cost tables. BC Hydro and SaskPower were checked directly — both publish project-level totals and qualitative resource assessments, but not comparable granular tables. This is a genuine characteristic of the Canadian data landscape.
Wind is the cheapest technology in every single province (10 of 10, no exceptions), ranging from $35.09/MWh (Alberta) to $65.13/MWh (BC).
Alberta's wind is the cheapest option in the entire dataset — not because of an unusually strong wind resource (its 37% capacity factor is mid-pack), but because AESO's real, sourced capital cost ($1,105/kW) is 39% lower than the Ontario-derived default used elsewhere ($1,824/kW) — concrete evidence of Alberta's mature, merchant-driven renewable market delivering genuinely lower build costs.
Natural gas shows a 3x spread — $71.14/MWh (Alberta) to $214.14/MWh (Quebec) — driven entirely by how each grid actually dispatches gas, not the hardware cost. Hydro-dominant provinces (Quebec, BC, Manitoba) barely use gas (10–20% capacity factor), making it their most expensive option; gas-reliant provinces (Alberta, Saskatchewan) run it closer to baseload (50–55% CF), keeping it competitive. LCOE is a property of a technology as deployed in a specific grid, not the hardware alone.
Solar tracks provincial sunshine almost exactly — Saskatchewan (Canada's sunniest, 25% CF) is cheapest at $66.19/MWh; Newfoundland (Canada's cloudiest, 12% CF) is most expensive at $137.90/MWh.
Full sensitivity analysis (tornado charts, per-parameter swing data) available in
notebooks/02_sensitivity_analysis.ipynb.
lcoe-optimization-tool/ ├── data/ │ ├── lcoe_defaults.csv # 10 provinces x 3 technologies, validated defaults │ ├── sensitivity_summary.csv │ └── sensitivity_full.csv ├── notebooks/ │ ├── 01_lcoe_fundamentals.ipynb # Formula derivation + IESO validation │ └── 02_sensitivity_analysis.ipynb # CRF extension, tornado charts, 10-province expansion ├── src/ │ └── lcoe.py # Reusable calculation, sensitivity, and plotting functions ├── app.py # Streamlit dashboard ├── requirements.txt └── README.md
git clone https://github.com/tmugomba/lcoe-optimization-tool.git
cd lcoe-optimization-tool
python -m venv venv
venv\Scripts\activate # Windows; use `source venv/bin/activate` on Mac/Linux
pip install -r requirements.txt
streamlit run app.py- Nova Scotia, BC, Saskatchewan, Manitoba, Quebec, New Brunswick, PEI, and Newfoundland's capacity factors are reasoned estimates, not directly sourced — clearly tagged in-app and in the data
- CapEx, Fixed O&M, and discount rate are treated as consistent across provinces (equipment/financing markets are largely national) except Alberta, which uses AESO's real, lower figures
- Sensitivity analysis uses a flat ±5 percentage point capacity factor range across all technologies, which disproportionately affects low-capacity-factor cases (e.g. Quebec's 10% gas CF) — documented in Notebook 2 as a known limitation, not smoothed over
Python, pandas, NumPy, Plotly, Streamlit, Jupyter
IESO and AESO data used under their respective public reporting terms. This tool is for educational and portfolio purposes; do not use for actual investment decisions without consulting current, verified figures from the relevant grid operator or a qualified financial advisor.
Tendekai Mugomba — Electrical engineering student passionate about renewable energy, building data science skills at the intersection of energy systems and machine learning. Currently gaining hands-on experience in solar PV analytics, M&V, and energy modeling, and building this portfolio to apply that growing skill set to broader ML-driven energy problems.