Overview
Summary
The U.S. Regional Economy, Greenhouse Gas, and Energy Model (US-REGEN) is an energy-economy model developed and maintained by the Electric Power Research Institute (EPRI). It combines a detailed dispatch and capacity expansion model of the United States electric sector with a technologically detailed consumer choice model of end-use service and energy demand as well as a fuel supply model representing alternative primary resources and conversion technologies for non-electric fuels. The three models are solved iteratively to convergence, allowing analysis of policy impacts on the electric sector taking into account electricity demand responses, and conversely allowing analysis of how economy-wide energy policies and technological improvements impact electric demand and load shapes. The electric sector model can also be run separately with higher regional and temporal resolution, or for a single year with hourly resolution and unit commitment constraints. This makes US-REGEN capable of modeling a wide range of environmental and energy policies in both the electric and non-electric sectors.
US-REGEN is a regional model of the United States. It can consider multiple sub-regions of the continental U.S., to account for differences in resource endowments, energy demand, costs, policies, and policy impacts, as well as transmission of electricity. By default, the model uses 16 sub-regions, each an aggregation of states, but with the datasets currently included, can be configured to consider any arbitrary aggregation of the lower 48 states plus the eleven New York Independent System Operator (NYISO) zones within New York state. The electric model can be extended to include additional sub-state detail for other states based on the boundaries of independent system operators (ISOs) and regional transmission operators (RTOs). A separate version of the model to include Canadian provinces is also available. Figure 1‑1 shows a map of the default sub-regions in the model.[1] US-REGEN is an inter-temporal optimization model. By default it solves in five year time-steps from 2015 through 2050, but can be configured to use other time-steps and base years.
The electric sector component of US-REGEN is a detailed generation planning model. In each time step, the model makes decisions about capacity (e.g. new investments, retrofits, and/or retirements) and dispatch to meet energy demand for both generation and inter-region transmission. It uses a bottom-up representation of power generation capacity and dispatch across a range of intra-annual time segments. It models transmission capacity between regions and requires that generation and load plus net exports and line losses balance in each time segment and for each region. In electric sector only mode, the model is capable of representing all lower 48 states as well as various intra-state zones as separate regions.
The end-use component of US-REGEN represents trade-offs between end-use technologies and fuels for a wide range of disaggregated sectors and activities with economy-wide coverage. The model includes structural detail across several dimensions relevant for fuel and technology choice, such as building size, type, and vintage, climate zone and location, and vehicle ownership and driving intensity. Within each structural category, service demand may be met with a range of options, characterized as combinations of fuels and technologies. The model evaluates the total cost of each option in each new vintage based on assumed technology cost and performance, fuel prices, structural attributes of service demand, and non-economic factors. The resulting allocation across the options is based on a logit model translating relative costs to equilibrium market shares, with a lagged process to simulate a gradual transition toward the model's calculated equilibrium shares. The model then calculates annual and hourly fuel use by region as a function of the resulting mix of end-use technologies. The electric and non-electric fuels demand outputs of this model are inputs to the electric and fuels model, respectively.
The fuels component of US-REGEN represents the primary supply of fossil fuels and bioenergy, as well as conversions of these resources to non-electric end-use fuels. It also represents the storage and delivery of fuels and includes carbon management options such as direct air capture and natural climate solutions. It uses input fuel demands from both the electric and end-use models to resolve economy-wide market-clearing constraints, potentially including policy constraints on economy-wide emissions, providing endogenous delivered fuel prices to the other model components.
The electric and fuels models are formulated as linear programming problems. The end use model is a set of modules representing different sectors, which are formulated using a mix of logit simulation and linear programming methods. Each model component is implemented within the GAMS software environment.
US-REGEN uses several primary sources to calibrate to observed data:
- Economic data sourced from the IMPLAN database
- Energy data from the Energy Information Administration (EIA) of the U.S. Department of Energy
- Gridded weather data from the National Aeronautics and Space Administration (NASA)
- Household travel survey data from the U.S. Department of Transportation Federal Highway Administration
- U.S. generation fleet and emissions data from ABB Energy Velocity (formerly Ventyx)
- The National Emissions Inventory (NEI)
- Technology cost and performance data from the Electric Power Research Institute (EPRI)
- A variety of other sources providing detail on economic growth, policy, and biomass, among others
Further details on data sources are provided in the Resource and Technology Assumptions section. Note that this documentation is not intended to convey results and analysis arising from use of US-REGEN, although some limited results are shown to demonstrate how the model works. To access US-REGEN results, please see Reports and Publications.
Limitations and Assumptions of Energy-Economic Models
EPRI's US-REGEN model and other energy-economic models are necessarily numerical abstractions of the complex economic and energy systems they represent. US-REGEN is designed to show how changes in technology or policy constraints impact economic and energy-system activities relative to a baseline case; the optimization method and assumptions have been chosen and calibrated carefully to best align with this use. Doing so inevitably makes strong assumptions about the function of markets and energy electricity systems. The use of inter-temporal optimization implicitly assumes that actors will behave rationally in a competitive market with perfect information about current and future market conditions, including prices. For example, the model "knows" future gas prices and where and when the wind blows. The model is deterministic; that is, it does not include uncertainty, so that a given set of inputs and assumptions will produce the same outputs. This has the advantage of demonstrating the least-cost deployment mix of generation subject to a scenario-dependent range of technological and policy constraints and specified input assumptions. Apart from the impacts of uncertainty and imperfect information, economic theory suggests that it is reasonable to expect a least-cost outcome from a competitive market, which in some respects the wholesale generation of electric power has become.
These types of limitations and assumptions are common to the field of energy-economic modeling. Other prominent energy-economic models used in similar scenario analyses include the US-REP model developed by the Massachusetts Institute of Technology[2], the ADAGE model used by the Environmental Protection Agency (EPA)[3] and developed by RTI International, and the NEMS model developed by the Energy Information Administration (EIA)[4]. Prominent models similar to the US-REGEN electric sector model include the ReEDS model developed the National Renewable Energy Laboratory (NREL)[5], and the IPM model developed by the U.S. EPA and ICF[6].
Major Variations of US-REGEN
The US-REGEN model can be run in various configurations to examine a variety of research questions. The electric sector module has three modes: a dynamic long-run formulation that can be optionally coupled with US-REGEN's end-use module or run standalone, a static hourly formulation that runs for a single year, and a unit commitment formulation that disaggregates the capacity blocks into individual units with ramp rates, start-up/shut-down costs, and other operational details. The latter two models can be "seeded" from dynamic model outputs. Below is a table of the characteristics of each version of US-REGEN, detailing the fundamental differences.
These characteristics are explored in more detail throughout the documentation. The Electric Sector Model section describes the functionality of the dynamic electric sector model, describes how the static and unit commitment models differ from that functionality, and outlines their use cases in more detail. The End-Use Model and Fuels Model sections explore the structure and functionality of those components and how they integrate with the electric sector model.
| Dynamic (Electric Only) | Static Equilibrium (Electric Only) | Unit Commitment (Electric Only) | Dynamic (Integrated Electric/End-Use/Fuels) | |
|---|---|---|---|---|
| Optimization Horizon | Multi-decadal (35 years) | Annual | Annual | Multi-decadal (35 years) |
| Temporal Granularity | ~100 segments per year | 8,760 hours | 8,760 hours | ~100 segments per year |
| Capacity Mix | Endogenous | User Preference | Exogenous (from dynamic model run) | Endogenous |
| Existing Unit Aggregation | 100+ capacity blocks per region (dispatched together) | 100+ capacity blocks per region (dispatched together) | Individual units | 100+ capacity blocks per region (dispatched together) |
| Storage | N/A | Endogenous | Exogenous | N/A |
| Geographical Detail | User-specified regions, up to state based | All user-specified regions | All user-specified regions | User-specified regions, less than 20 |
| Dispatch Constraints | Load balancing | Load balancing | Balancing, min. load limits, ramp rates, start costs, etc. | Load balancing |
| Optimization Type | Linear program | Linear program | Mixed-integer program | Linear program |
| Sectors | Electric Sector | Electric sector | Electric sector | Entire economy |
| Energy Demand | Exogenous, may use supply curves for some inputs | Exogenous | Exogenous | Endogenous; detailed end-use model |
| Fuel Prices | Exogenous, may use supply curves for some inputs | Exogenous | Exogenous | Endogenous; detailed fuels model |
Reports and Publications
Please visit EPRI's Energy Systems and Climate Analysis public website (https://esca.epri.com/models.html) for a list of freely-available EPRI reports and peer-reviewed publications that include US-REGEN modeling results.
For a list of all EPRI publications that have been tagged with US-REGEN, please visit https://www.epri.com/search?productkeywords=US-REGEN.
In defining the regions of the integrated model, there is necessarily a tradeoff between optimally representing the economy, where data is typically available along state lines, and optimally representing the U.S. electricity sector, the internal boundaries of which frequently cross state boundaries. ↩︎
https://globalchange.mit.edu/sites/default/files/MITJPSPGC_Rpt194.pdf ↩︎
https://cfpub.epa.gov/si/si_public_record_report.cfm?Lab=OAP&dirEntryId=198000 ↩︎