ENERGY 2020 Model Overview

Summary:

The ENERGY 2020 model is an integrated multi-region energy model that provides complete and detailed, all-fuel demand and supply sector simulations.  These simulations can additionally include macroeconomic interactions to determine the benefits or costs to the local economy of new facilities or changing energy prices.  The model can be used in regulated as well as deregulated and transitioning environments.  It portrays the interaction of market competitors in a realistic, as opposed to an idealized, fashion, including transmission-system market-dynamics.  It focuses on the imperfections of the market, including market gaming, and therefore, is extremely useful for M&A and asset evaluation.  Pollution emissions and costs, including allowance and trading, are endogenously determined, thereby, allowing assessment of environmental business-risks.

Overview:

ENERGY 2020 is an outgrowth of the FOSSIL2/IDEAS model developed for the US Department of Energy (DOE) and used for all national energy policy since the Carter administration.[1]  This early version of the ENERGY 2020 model was developed in 1978 at Dartmouth College for the DOE’s Office of Policy Planning and Analysis.

The basic implementation of ENERGY 2020 for North America now contains the user-defined level of aggregation down to the 12 provincial and 50 state (and sub-state) level.  ENERGY 2020 is historically parameterized to simulate all 3500 interacting energy suppliers in North America as needed.   This historical validation captures limits to future actions that  market players can pursue as market rules change.  Current efforts are adding the South American and European databases to the model, and allowing holding companies to see the combined portfolios crossing the continents.

ENERGY 2020 is parameterized with local data for each region/state/province as well as the all the associated energy suppliers it simulates.  Thus, it captures the unique characteristics (physical, institutional and cultural) that affect how people make choices and use energy.  National models typically reproduce history from 1975 to partially validate structure and results.  Collections of state and provincial models are currently validated from 1988 to the latest quarterly numbers because of limited historical data associated with electric utilities.[2]

ENERGY 2020 model can be linked to a detailed macroeconomic model to determine the economic impacts of energy/environmental policy and the energy/environmental impacts of national policy.  For US regional and state level analyses, the REMI macroeconomic model is regularly linked in ENERGY 2020.[3]  The macroeconomic model (that includes inter-state/provincial, US and world trade flows) simulates the real-time impact of energy and environmental concerns on the economy and vice versa.

The structure of the model is well tested and has been used to simulate not only US and the Canada energy and environmental dynamics but also those of several countries in Western, Central and Eastern Europe.  Current efforts include strategic and tactical analyses for South America deregulation.  The US EPA uses ENERGY 2020 to perform the regional (energy, environmental and macroeconomic) impacts of proposed Kyoto initiatives at the 50-state level.  Further, the model has been used successfully for deregulation analyses in over 50 energy suppliers and in all the US states and Canadian provinces.  Many US and Canadian energy suppliers currently use the model for the analysis of combined electricity and gas deregulation dynamics.[4]   The model contains confidence and validity packages that allow it to determine how to take maximal advantage of RTO rules.  The ISO NE used the model to find “gaps” in its rules and to develop more efficient market conditions.  The model was used for the CAPX/ISO to model to show, before-the-fact, many of the “games” played in the California market.

The default model simulates demand by three residential categories (single family, multi-family, and agriculture/rural), commercial, industrial by 2-digit SIC, and three transportation services (residential, commercial, industrial).  There are approximately six end-uses per category and 6 technology/mode families per end-use.[5]  Currently the technology families correspond to six fuels (oil[6], gas, coal, electric, solar and biomass).  The transportation modes include automobile, truck, bus, train, plane, marine and electric vehicles. Added end-uses, technologies and modes can be added as data allow. (Added sectoral detail comparable to the 13 building-types in the national model’s commercial sector can be added as well.)  For all end-uses and fuels, the model is parameterized based on historical locale-specific data.  The load duration curves are dynamically built up from the individual end-uses to capture changing condition under consumer choice and combined gas/electric programs.

Each energy demand sector includes cogeneration and distributed generation simulation including mobile-generation and fuel-cells. Retail wheeling and fuel-switching responses are rigorously determined. The technology families (which can be split, as an option, to portray specific technology dynamics) are aggregates that, within the model, change building shell, economic process and device efficiency and capital costs as price or other information that the decision makers see, changes.  The ENERGY 2020 model utilizes that data the group develops for parameterizing and disaggregating the model.  ENERGY 2020 provides feedback on the implications of future assumptions.  Its demand and prices forecasts are impeccably accurate even under extreme market conditions

The supply portion model includes endogenous detailed electric supply simulation of capacity expansion/construction, rates/prices, financial/accounting, load shape variation due to weather, and changes in regulation.[7] 

The electric sector can additionally simulate the full spectrum of deregulated markets, whether these include a power exchange, ISO, Poolco, Gridco, Transco, or any RTO configuration.   The model dispatches plants according to the specified rules whether they are optimal or heuristic and recognizes transmission constraints as well as the associated costs.[8]  A sophisticated dispatch routine selects critical hours along seasonal load duration curves as a way to provide a quick but accurate determination of system generation.  Peak and base hydro usage is explicitly modeled to capture hydro-plant impacts on the electric system.

Where the model departs from conventional (idealized) approaches is in the overall behavior of the market players. Each utility (or energy provider, as appropriate) is represented in the model by four business units: distribution, transmission, marketing, and generation. The first two remain regulated but the last two can be deregulated to any degree.  All market participants use the rules to their best self-interest.   Many organizations do not have the financial or physical where with all to undertake or survive certain activities in the market. They can be (and maybe should be) easily victimized. Other players with locational, financial or generation advantages play them to the detriment of other competitors  -- just as do the competitors in any other industry.  New market entrants, asset sales/purchases, mergers, acquisitions, takeovers, and bankruptcy are explicitly modeled because that will be the realistic behavior of the market. Players may bid what economics predicts on average, but the deregulation transition is volatile and non-linear.  There is no unique economic solution.  This allows players to try multiple strategies that, while inconsistent with the long-term stability, are successful and therefore economically efficient in the local sense.

The process of deregulation requires a careful consideration of market power dynamics. The ENERGY 2020 model can examine how potential rules can be used to by market participants to take advantage of the market. It can then be used to help design rules that limit the potential for exercising market power.  ENERGY 2020 does produce the     

Herfindahl-Hirschman Index of Concentration (HHI). It can also readily generate the indexes found in  DOJ/FTC Merger Guidelines, FERC Order Nos. 592, 888 and other FERC reports.

The gas distribution utility dynamics are also simulated, but the generic state/provincial models does not contain oil or gas production; only a simplified simulation to determine delivered-product prices. E2020 is written in a language called PROMULA (PROcessor of MUltiple LAnguages) that by its nature allows other client analytical or accounting systems to run under it.

ENERGY 2020 can include oil, gas and coal supply sectors (they exist in the FOSSIL2/IDEAS model) and it is an option (as is the alcohol supply simulation) not incorporated in the basic model implementation.  Energy used in primary production and emissions associated with primary production and its distribution is included in the model.

The ENERGY 2020 model includes pollution accounting for both energy (by fuel, end-use and sector) and non-energy (by economic activity) for SO2, NO2, N2O, CO, CO2, CH4, TSP, VOC, CF4, C2F6, SF6, and HFC.  Other (gaseous, liquid and solid)  pollutants can be added as desired. Pollution is not determined directly by coefficients but rather by the accumulation of capital investments that result in pollution emission with usage. National and international Allowance trading is also included.  Plant dispatch can consider emission restrictions.

The model uniquely captures the feedback among energy consumers, energy suppliers and the economy.  For example, a change in price affects demand that then affects future supply and price. Increased economic activity increases demand; increased demand increases the investment in new supplies.  The new investment affects the economy and energy prices. The energy prices also affect the economy.  While this feedback makes for more self-consistent forecasts and characterization of policy impacts, it also adds increased complexity to the detriment of consensus building among stakeholders.  As such, the model can be run without the feedback active.

The ENERGY 2020 user interface allows the user to arbitrarily specify output tables and graphics.  All information in the model can be interrogated and modified interactively.  A MS Windows menuing system allows automated policy analysis and scenario specification.  These same capabilities allow the user to save multiple scenarios and analyses and then compare them with each other graphically or in tabular form.

Finally, the system includes confidence and validity testing software that places uncertainty bounds on simulation results, quantifies confidence intervals, and ranks the contributions to uncertainty in future conditions. This feature can be used to limit data efforts to information important to the analysis and to determine those strategies and tactics that will most likely result in the desired conditions.

ENERGY 2020 can simulate a technology-by technology, asset-by-asset modeling approach.  Via menus, the user can define new technologies and determine their value and impacts in the marketplace.  The ENERGY 2020 model is designed for scenario testing.  The introduction of a new technology is associated with many market considerations that include market applicability, sub-market niche distinctions, marketing/advertising strategy and categorization of the technology as a new “family” or part of an existing family of technologies.  Additionally, incentives such as rebates, tax breaks, and subsidies can be considered.  The impact of potential changes in technical characteristics such as cost, lifetime, operating costs and efficiency can then be addressed.