Capacity Expansion Modelling
Summary Description
Capacity expansion models (CEMs) are long-term electricity system planning tools used to identify optimal investment and retirement pathways for generation, storage, and supporting infrastructure required to meet projected electricity demand while satisfying policy, technical, and reliability constraints. These models typically optimize system outcomes over multi-decadal horizons by minimizing total system cost subject to assumptions on demand growth, technology performance and costs, fuel prices, resource availability, and planning constraints (Boyd (2016), EPRI (2024)).
CEMs are widely applied to support integrated resource planning, policy evaluation, and long-term system transformation analysis, including decarbonization pathways and electrification scenarios. Recent literature emphasizes the needs of adaptation of climate change and the corresponding system vulnerability in the CEMs, such that climate impacts are translated into planning-relevant model inputs rather than treated implicitly or assumed stationary. Climate change affects capacity expansion outcomes, including changes in electricity demand requirements, generation resource availability, and infrastructure performance limits (EPRI (2024), Cooke et al. (2021)).
On the demand side, temperature increases and extreme heat events alter energy consumption levels, peak demand magnitude, and load profiles, which in turn affect capacity needs and investment timing. Planning studies therefore treat demand as an exogenous requirement informed by climate conditioned load projections rather than as an endogenous outcome of capacity expansion modelling (Homer et al. (2023)). On the supply side, climate change influences generator performance and availability, particularly for thermoelectric and hydropower resources. Thermal generation performance is sensitive to ambient air temperature and cooling water conditions, which can result in capacity deratings or reduced availability during extreme heat or water stress. Hydropower generation is affected by changes in precipitation, snowmelt timing, evaporation, and interannual hydrologic variability, which challenge the assumption of stationary inflows traditionally used in long-term planning (Cooke et al. (2021), Ray et al. (2018), Hydro-Québec (2022)). These effects are increasingly represented in CEMs through climate informed resource availability assumptions or scenario-based stress testing, rather than single deterministic forecasts (EPRI (2024)).
CEMs play a central role in screening long-term investment strategies and identifying candidate resource portfolios that are economically efficient and policy compliant under assumed future conditions. In response to deep uncertainty surrounding future climate conditions, CEMs are used to explore multiple plausible futures, enabling planners to compare system costs, resource mixes, and vulnerabilities across climate, policy, and technology scenarios (Marchau et al. (2019), Homer et al. (2023), NARUC-NASEO (2021)).
Role in the Electricity System
CEMs play a strategic planning role within the electricity system by informing long-term investment and retirement decisions that shape the future composition of generation, storage, and supporting infrastructure. Their primary function is not to operate the power system, but to support future planning processes, such as integrated resource planning, long-term transmission planning, and policy evaluation, by identifying resources availability for meeting projected demand and policy objectives at least system cost (Boyd (2016), NARUC-NASEO (2021)).
From a reliability perspective, CEMs contribute indirectly by ensuring that sufficient capacity is planned to meet forecasted energy and peak demand requirements under planning reserve criteria. However, they generally do not resolve short-duration events, extreme weather stress, or correlated outage in sufficient detail to assess loss-of-load risk. Rather, their outputs are commonly used as inputs to resource adequacy planning, where probabilistic methods are applied to evaluate reliability under uncertainty (Homer et al. (2023), IESO (2024)).
In the context of climate change, the role of CEMs has expanded to include system-level screening of climate risks and adaptation options. Planning guidance emphasizes that CEMs are well suited to evaluating how climate-driven changes, such as increased temperature-sensitive demand, altered hydropower availability, or reduced thermal generator performance during extreme heat, may affect long-term capacity needs, technology choices, and investment timing, provided these impacts are represented through climate-informed planning assumptions. For example, Cooke et al. (2021) highlights that climate change can affect both generation availability and electricity demand through changes in water availability, cooling conditions, and temperature-driven load patterns, and recommends incorporating climate-adjusted weather and water projections into planning analyses rather than relying solely on historical observations. Similarly, EPRI (2024) emphasizes the need for climate-informed planning inputs that translate climate hazards into changes in generator availability or efficiency within planning models. In addition Hydro-Québec (2022) describes a structured process to identify climate hazards, assess the vulnerability of electricity system assets and operations, and evaluate adaptation measures to maintain reliable electricity supply under changing climate conditions.
Climate Scenarios and Capacity Expansion Models
More recent planning frameworks further highlights the role of CEMs in decision-making under deep uncertainty, where future climate, policy, and technology pathways cannot be described by a single forecast. For example, Cohen et al. (2014) applied CMIP3-based surface-water projections in the ReEDS capacity expansion model to represent constraints on water availability for new generation capacity, explicitly linking climate-driven water supply to technology and siting choices in the expansion optimization. Macknick et al. (2015) further implemented water availability constraints in ReEDS such that limited freshwater can force the model to build smaller plant capacities, select alternative cooling systems, or shift to less water-intensive technologies. More recently, Szinai et al. (2024) used 15 climate scenarios from downscaled GCM projections under RCP8.5 for estimating changes in hydropower generation potential and demand changes using load sensitivity factors from ReEDS model. These climate-conditioned estimates were then used as inputs to the Solar, Wind, Transmission, Conventional, and Hydroelectric generation model (SWITCH) to evaluate generation and transmission expansion pathways, In such contexts, CEMs are used to compare system outcomes across multiple plausible scenarios, supporting the identification of resource portfolios that remain viable across a wide range of future conditions rather than optimizing for a single expected case (Marchau et al. (2019), Homer et al. (2023), NARUC-NASEO (2021)).
Key Inputs
- Electricity demand and load characteristics: including annual energy demand and peak demand. These projections may reflect assumptions that are included in the demand forecasting, such as economic growth, electrification, demand-side measures, and weather sensitivity.
- Technology cost and performance characteristics: These parameters explain which technologies are selected in least-cost solutions.
- Fuel price assumptions: Prices of natural gas, coal, biomass, and fuel supply or transportation constraints affect operating costs and dispatch decisions.
- Policy and planning constraints: Emission limits, renewable portfolio standards, clean energy targets, retirement schedules, or technology deployment limits.
- Resource availability and system constraints: Renewable resource potentials (e.g., wind, solar, hydropower), transmission limits or expansion options, reserve or adequacy requirements, and operational simplification.
- Climate and weather-related inputs: Climate variables, and hydrologic inflow scenarios derived from historical data or future climate projections reflect climate variability and change on system planning.
- Planning reserve margin requirement: Minimum planning reserve margin to ensure that total available generation capacity exceeds projected peak demand by a specified percentage.
Model Outputs
- Capacity expansion and retirement decisions: What types of resources are built or retired, in what quantities, and in which years over the planning horizon.
- Generation and energy mix: Electricity production by technology in a specific time step, which provides insight into how demand is met across different resources.
- System costs: Total system cost, investment costs, operating costs, and fuel expenditures.
- Emissions and environmental indicators: Carbon dioxide and other pollutant emissions.
- Reliability and adequacy indicators: Reserve margins, capacity credits, or curtailment levels, which provide a screening-level indication of whether future capacity portfolios can meet demand.
- Scenario comparison insights: Trade-offs among cost, emissions, technology deployment, and resilience outcomes across different assumptions, including climate-informed scenarios.
Discussion, Gaps, and Recommendation
Several studies have quantified the sensitivity of thermal power plant performance to elevated ambient temperatures, reporting reductions in power output of thermal efficiency relative to standard operating conditions (often ISO conditions at 15 °C). For gas turbines, De Sa & Al Zubaidy (2011) reported that for every 1 °C increase above ISO conditions, gross useful power decreased by 1.47 MW, corresponding to a 0.554 % reduction in rated power per degree Celsius. Pinilla Fernandez et al. (2021) similarly examined gas turbine performance and found that thermal efficiency decreases by approximately 0.06 % per °C increase in ambient temperature above ISO conditions. The same study also reported that total power output reductions under high-temperature conditions could reach up to 22 %. In addition to direct power and efficiency losses caused by higher ambient temperatures, climate change can also constrain thermal power generation through impacts on cooling-water availability. Van Vliet et al. (2012) showed that increases in river water temperature and reductions in summer river flows can significantly limit cooling water for thermoelectric power plants, resulting in decreases of power plant capacity by 4.4-16 % in the United States and 6.3-19 % in Europe. Van Vliet et al. (2016) extended this analysis globally and found that a majority of thermoelectric power plants worldwide are projected to experience reductions in usable capacity under future climate conditions.
In current practice, climate-related inputs in CEMs primarily affect resource availability for weather-dependent generation sources such as wind, solar, and hydropower. Hydropower availability, for instance, is typically represented using historical hydrologic data rather than forward-looking climate projections. Similarly, renewable generation profiles are generally derived from historical weather observations. Another characteristic of CEM highlighted by practitioners is the largely deterministic nature of many expansion models. Rather than explicitly representing probabilistic weather variability, these models often rely on representative weather profiles, such as median conditions, to estimate system performance and resource availability. While this approach simplifies the optimization process and reduces computational complexity, it limits the ability of expansion models to capture future climate variability and extreme weather conditions.
For conventional thermal generation, temperature-dependent performance effects are also rarely represented explicitly in CEMs. Instead, simplified assumptions such as seasonal firm capacity ratings are typically used to approximate temperature-related performance differences between summer and winter operating conditions. These modelling simplifications reflect practical trade-offs between model complexity and computational tractability in large-scale CEMs.
Coupling insights from climate projections with CEMs could help represent long-term shifts in renewable resource availability and hydrologic conditions. Methods that incorporate climate non-stationarity may improve the robustness of expansion planning to evaluate long-term investment decisions under multiple economic and policy scenarios. However, incorporating detailed weather variability or ensembles of climate projections directly into optimization models would significantly increase computational requirements and complicate model implementation. A practical approach could be to use climate information to shape a set of climate-informed planning scenarios. Under this framework, insights from climate projections can be translated into alternative assumptions regarding resource availability, hydrologic conditions, or temperature-driven demand patterns. These assumptions can then be incorporated into CEMs through representative scenarios, allowing planners to evaluate the robustness of investment pathways under different plausible climate futures.