Electric Sector Model
Overview
The US-REGEN electric sector model is a detailed dispatch and capacity expansion model of the US electric system. It includes a partially disaggregated representation of both existing generation unit capacity and the hourly profiles of load, wind speed, and solar flux. These details allow the model to explicitly evaluate dispatch decisions (when and for how long installed capacity operates) as distinct from capacity decisions (new investment, retrofit, or retirement).
Several unique features of the electric sector make the treatment of such details essential to accurately model these decisions and the impact of new policies:
The "shape" or hourly profile of end-use demand and variable resource availability is crucial for appropriately characterizing the operational patterns and profitability or value of different types of capacity.
These patterns and hence the value of generating assets are also dependent on the mix of installed capacity in a region (and in neighboring regions).
Capital investments in generating capacity tend to be long-lived, creating a strong link between dispatch and investment decisions across time periods.
The electric sector model is formulated as an optimization over several time periods balancing the costs incurred by electric sector producers with the value to electricity consumers. The decision variables include both levels of capacity by region and technology type and the dispatch of these "blocks" of capacity across a range of "segments" that represent the intra-annual profile of load and variable resource availability. Each segment is a block of time that is "representative" of anywhere from one hour to over two hundred hours out of the 8760 hours in a given year. The hours represented by one segment are usually not contiguous. In addition, power may flow between adjacent regions during each segment subject to available bilateral transfer capacity.
The costs incurred by producers include variable costs that scale with dispatch (mainly fuel and variable operating and maintenance (VOM) costs), fixed operating and maintenance (FOM) costs that scale with installed capacity, and investment costs associated with new capacity additions (of both generating and inter-region transfer capacity). The optimization considers the time paths of each of these variables and their associated costs simultaneously, subject to a discount rate reflecting the opportunity cost of capital, even though the costs themselves are incurred on very different schedules.[1]
Electricity demand in any given iteration of the electric sector model is treated as a fixed exogenous quantity. Typically, the load profile is specified as an output of the end-use model, though the electric model can also be run in stand-alone mode with any exogenous load profile (e.g. historical or observed data). The electric sector model outputs electricity prices that serve as an input to endogenous demand decisions in the end-use model. Demand elasticity is simulated by allowing the price output of the electric sector model and the quantity output of the end-use model to converge over multiple iterations, which provides a richer representation of the responsiveness of demand to prices than a simple elasticity. When running the electric sector model in stand-alone mode, elastic demand can be approximated with a linear demand function calibrated to given reference point (e.g. price / quantity pair across regions and time steps in a reference run with fixed demand).
The model optimizes by minimizing the costs of meeting the given level of demand (see Figure 2‑1) in each segment. These costs are summed across regions and time periods and discounted to present value. Importantly, the requirement that demand is met in every segment, in addition to a reserve margin constraint on local firm capacity, simulates the clearance of both an energy market and a capacity market. That is, by requiring that sufficient electricity be produced in each segment to meet the prescribed load, this constraint also stipulates indirectly that sufficient investment in capacity occur such that electricity for the prescribed load in the "peak segment" will be available for dispatch, plus a reserve margin. As will be discussed in more detail below, this stipulation applies even with large deployment of variable renewable capacity which is known to have low coincidence with peak demand, such as wind.
The electric sector model is a partial equilibrium model, meaning that the optimization extends only to the electric sector and does not explicitly account for choices and feedbacks in related markets. These interactions are incorporated through the iterative process outlined above and described in detail in the End-Use Model.
When the price elasticity of demand is assumed to be zero, the electric sector model minimizes (the present value across time and regions of) total producers' costs, represented by the area labeled A. This demand can be an output of the end-use model, or an exogenous projection can be used.
Dispatch in the dynamic electric model is by increasing order of marginal generation cost; i.e. units with the lowest variable costs per MWh are dispatched first. This omits unit commitment constraints, due to both computation constraints on including integer constraints in a linear optimization, and to unit aggregation rendering constraints such as ramp rates less meaningful. A unit commitment variant of the US-REGEN electric model can be used to explore the impact of these constraints in more detail for a single year, using the capacity mix from a dynamic electric model scenario with identical assumptions. This variant is described in the Unit-Commitment Model section, and in more detail in EPRI publication 3002004748 (EPRI, 2015).
Solution Characteristics
US-REGEN's electric sector model solution characterizes a profile of the electricity sector over time. By default, outputs include the model years 2015 to 2050, and the key solution variables of the model are
- capacity levels, which reflect new and retrofit investments and retirements;
- generation (i.e. dispatch) by technology and segment;
- inter-regional power flows by segment; and
- the price of electricity.
From those, other outputs can be derived, including fuel consumption and emissions of CO2 and other pollutants. The solution variables satisfy the optimality conditions; that is, they represent the values that minimize net present value of costs subject to constraints.
Optimality Conditions
The optimization formulation ensures that the key outcomes in any optimal solution will satisfy certain conditions. Equivalently, these optimality conditions describe a long-run competitive market equilibrium.
Dispatch Order: Within each segment, units will be dispatched in increasing order of marginal generation cost. Otherwise, producers' cost could be reduced (total surplus increased) by replacing the highest cost dispatched unit (i.e. marginal unit) with a less costly unit that had been bypassed in the dispatch order.
Complementary Slackness of Trade: Within each segment, if the marginal unit in one region has a higher dispatch cost than the marginal unit in an adjacent region, transmission from the adjacent region must be at its upper bound (i.e. transmission capacity must be fully utilized).[2] Otherwise, costs could be reduced by replacing the marginal unit with electricity imported from the adjacent region. Similarly, whenever the marginal generation cost is equal in adjacent regions (or more precisely, whenever the difference is less than the loss adjustment), transmission between regions during that segment must be strictly less than the upper bound.
Profitability of Investment: For any investment in capacity, the marginal unit added will have the present value of its costs (initial capital and operating costs over the lifetime) less than or equal to the present value of revenues (quantity generated in each segment multiplied by the segment price). This condition applies to both new additions and retrofit investments, and an analogous condition applies to new additions of transmission capacity (where the definition of revenues is related to the marginal value of transmission rather than the electricity price itself).[3] Otherwise, total surplus could be increased by dropping the investment in the marginal unit and its operation and foregoing the consumer benefit associated with the energy it produced, which at the margin is equal to its revenues. Further, if the present value of revenues is strictly greater than the present value of costs for the marginal investment in some technology, then there must be a constraint on that technology and investment must be equal to the upper bound. Otherwise, surplus could be increased by substituting investment of the positive-profit technology for some other investment with marginal revenue equal to marginal cost (i.e. zero net-profit).
Price of Electricity
The electric model reports the price of electricity for each region and time step at both the wholesale and the retail level. The wholesale price is related to the generation component of the price, which reflects energy and capacity costs of providing wholesale electricity subject to policy constraints imposed on the generation mix. This price is an output of the electric sector model's cost optimization. The remainder of the retail price reflects average transmission and distribution (T&D) costs and is estimated ex post of the model solve. Note that the wholesale or generation price also includes the market for inter-regional transmission capacity additions, which are endogenous to the electric model's optimization. However, a significant share of transmission costs is associated with intra-region capacity additions and maintenance, which are not explicitly captured within the optimization. These costs are included in the exogenously calculated T&D component of the retail price. The retail price, constructed as the sum of the generation and T&D components, is sent to the end-use model to evaluate consumer end-use energy decisions; the resulting electric load is fed back to the electric model in the next iteration.
Wholesale Price
The wholesale price is a marginal price that corresponds to the dual variable associated with the market clearance condition (i.e. supply = demand) that is enforced in every segment, region, and time period. At optimality, a dual variable (or shadow price) is equal to the amount by which the objective function could be increased (resp. decreased) if the associated constraint were relaxed (resp. tightened) by one unit. That is, the model's reported price in each segment corresponds to the marginal cost of supplying an additional MWh at that time in that region. Note that the objective function also includes costs for new inter-regional transmission as well as imputed rents accruing to existing inter-regional transmission, hence these components are included in the wholesale price. For many infra-marginal (i.e. non-peak) segments, the shadow price corresponds to the "dispatch cost" (fuel and variable O&M cost) of the marginal generating unit in the region (or in neighboring region plus the imputed transmission cost). However, in certain segments, supplying the last MWh actually requires the addition (or retention) of a unit of capacity (either in the form of new investment or deferred retirement).[4] For these segments, the shadow price will include all or a portion of the cost of that marginal capacity addition/retention. Additionally, the supplemental constraint requiring sufficient local firm capacity plus a reserve margin will bind in at least one such segment. The shadow price of this constraint is added to the shadow price on the market clearance condition to form the full wholesale price in peak segments. The result is that a small number of segments across regions and time periods will have very high prices, several orders of magnitude higher than the dispatch cost, and thus the annual average price across all segments reflects the full long-run marginal cost (including both fixed and variable components) of wholesale electricity supply (including inter-regional transmission).
Retail Price
The generation component of the retail price can be calculated in two ways. First, it can be calculated directly as the annual average wholesale price, weighted across segments by delivered load. This is analogous to the average price a de-regulated load-serving entity (LSE) would pay to provide power to its retail customers. Alternatively, the generation component can be calculated as an average cost, that is, dividing total cost (variable costs plus depreciation and rate of return on the rate base) by total end-use sales. This calculation is conceptually analogous to the process by which the generation component of the retail price is set in cost-of-service regulated regions. However, in the idealized setting of the model's intertemporal optimization, there is little practical or numerical difference between these two approaches. The distinction is primarily relevant for measuring the impacts of certain policy instruments affecting the residual value of existing generation assets. Unless the model is being applied to analyze such an instrument, we use the more straightforward approach of directly connecting the realized wholesale price to the retail price. In terms of calibration, it may not be the case that the average wholesale price calculated by the model for the base year coincides with the observed generation component of the retail rate (as reported by EIA). Such a gap may exist because of legacy investments or out-of-the-money contracts. We include an adjustment factor for observed base year discrepancies but assume that this factor is reduced to zero over time.
The T&D component of the retail price is estimated based on observed T&D costs in the base year, adjusted in future years by projected changes in the load shape calculated in the enduse model. Base year costs are derived from EIA's reported price by component from the electricity market module of NEMS, as well as SEDS reported retail price by enduse-sector and state. For model projection years, the average T&D cost per delivered MWh in each sector and region is scaled from the base year level by the change in the ratio of peak to average load within each model region. That is, total T&D costs are assumed to scale with peak load, so that average T&D costs (i.e. the T&D component of the retail rate) scale with the peak to average ratio. While a more detailed assessment would consider a range of other factors relevant for T&D expenditures, this approximation is intended to roughly capture the additional costs of T&D infrastructure upgrades that may accompany increases in system peaks driven by increased electrification. We continue to refine this calculation to reflect potential impacts of changing patterns of load and generation resources on T&D system requirements and costs.
Auxiliary Markets
US-REGEN includes auxiliary markets for capacity reserves, and for spinning reserves. US-REGEN does not currently represent other auxiliary markets such as black start, regulation, or other reliability services.
The optimization algorithm is designed to ensure that sufficient reserves and capacity are built to cover any event occurring within the model's time horizon. However, many investment decisions are made to hedge against the possibility of a stochastic shock to the system. Such shocks are covered, in practice, by the imposition of a reserve margin. By default, US-REGEN adds a reserve margin to all regions equal to 7% above peak residual load. The peak residual load is calculated as the greatest hourly demand net of the intermittent renewable generation within that hour. The residual peak hour is often distinct from the absolute peak hour, which in some cases coincides with high renewable generation. Dispatchable technologies located within the region contribute their full nameplate capacity to the reserve margin, while variable resources, such as wind and solar PV, contribute their modeled output in the hour with the peak residual load. Rooftop PV does not contribute to the reserve requirement. Imports or import capacity do not contribute to the reserve margin by default. In practice, the model increases capacity of flexible generation such as natural gas relative to scenarios without a reserve margin.
US-REGEN also has a representation of spinning reserve markets. Because this is computationally taxing, it is not deployed by default; only where spinning reserve revenues may be significant for new generation capacity investment decisions.
Interaction with End-Use Model
Load growth and hourly load shape inputs vary depending on whether the model is run in integrated or electric-only mode. In an integrated model run, regional hourly load shapes are calculated by the end-use model for each time step, as described in the Interaction with Electric Model section. In this case, the load shapes are updated for each model year to reflect changes in the patterns of energy demand over time.
In the electric-only model, the regional shapes are based on historical hourly load profiles, which are derived from a dataset from ABB Energy Velocity which assembles data from EIA reported by balancing authority. These shapes are translated to hourly shapes for each region and scaled to match total base year electricity consumption (retail and direct use) reported by the EIA. In this case, the load shape stays constant over the duration of the model. The reference load growth path for future time periods is based on the EIA's Annual Energy Outlook (AEO) but the model can be run under a variety of alternative load growth scenarios.
Design of Aggregated Segments
In order to solve the model as an intertemporal optimization over several time steps, it is necessary to reduce the number of intra-annual segments over which dispatch is resolved. US-REGEN uses a novel approach for selecting and appropriately weighting a subset of representative hours. In choosing a subset of weighted hours from the 8,760 hours of the year, the goal is to maintain the important characteristics of the disaggregated data so that model outcomes in the reduced form version are as close as possible to the hypothetical outcome using the full data.[5] These characteristics include:
- The area under the load duration curve (i.e. total annual load, for each region)
- The shape of the load duration curve (for each region)
- The capacity factors of new wind and solar capacity (for each region and class)
- The shape of wind and solar output relative to load (for each region), in particular the extremes of the joint distribution (e.g. hours when load is high but wind/solar output are low)
If the problem were simply to capture the load profile, only a handful of segments, perhaps a peak, shoulder, and base load, would be necessary. This level of aggregation is often employed by models to approximate a load shape. However, wind and solar are more variable, ranging from near 0% to near 100%, and considering all three together extends the variability to multiple dimensions. Furthermore, the set of representative hours chosen should apply across all regions, so that synchronicity of inter-regional transmission is preserved. This necessarily adds additional hours to the set; while the distributions in adjacent regions may have many similarities, conditions in Florida will bear little resemblance to those at the same hour in the Pacific Northwest.
Given these complexities, trade-offs among the various criteria are inevitable, and the value of a systematic approach is strongly indicated. We have developed a novel heuristic approach that relies on a simpler integer formulation to identify a set of hours that at a minimum covers the extremes of the joint distribution in each region, combined with a clustering algorithm to capture the interior of the joint distribution of load-wind-solar. Including the extremes ensures that, on the one hand, capacity values are not over-estimated (e.g. the moment of highest load and lowest wind/solar output is represented), and on the other that abundance of wind/solar output relative to load (potentially forcing transmission, storage, or curtailment) is included. After the hours are chosen, the selected hours are weighted so as to minimize error in total load and average annual capacity factor across all regions and classes. We provide an abbreviated description of this algorithm below and refer the reader to Blanford et al. (2018) for more detail.
Choosing Extreme Hours
In each region, we consider three synchronous hourly time series corresponding to load (as a percentage of peak), and wind and solar output (as a percentage of the annual maximum, weighted average across available classes). These hourly values can be plotted in three-dimensional space. These are the red markers in Figure 2‑15, which gives the example of Texas. The extremes of interest are the eight vertices of the space spanned by the hourly data, as well the vertices of the one- and two-dimensional projections of this space, which may or may not coincide. These vertices are the hours identified as those closest (in the conventional Cartesian sense) to the actual vertices of the unit cube (or line or plane in the projection spaces). For example, the hour closest to the unit cube vertex (1,0,0) (with co-ordinates referring to load, wind, and solar respectively) will not literally have values (1,0,0) unless the hour with peak load also happened to have zero output of both wind and solar. Instead, we identify the hour whose values are closest to this point, which might be, for example (0.9, 0.09, 0) at 7:00 p.m. (local time). In this case, the vertex hour would capture the moment when load is still high after a hot summer day, wind has not picked up, and the sun has set. It is crucial to the analysis for the model with aggregated hours to know that such moments exist.
The essential principle behind the extreme hour selection algorithm is to identify the minimum number of hours such that at least one hour is selected with sufficient proximity to each vertex in each region. If we define "sufficient proximity" as "exactly equal," we would use the set of vertex hours themselves. However, some of the vertex hours turn out to be vertices in more than one region, and other hours close to the vertex could be used to represent extreme conditions in multiple regions. Thus, if we allow a selected hour to qualify as a vertex if it is within some small distance from the actual vertex, the number of representative extreme hours can be reduced. The bubbles in Figure 2‑15 are centered on the identified vertices for Texas and extend five percentage points in each dimension. The tolerances used in the current version of the model are shown in Table X. When we define "sufficient proximity" to each vertex according to these tolerances, the minimum number of "extreme-spanning" hours turns out to be roughly 100.[6]
Choosing Cluster Hours
For most of the year, conditions are not at any of these extremes, so using only these extreme hours tends to over-represent the tails of the load, wind, and solar distributions. Thus, in addition to the selection of extreme hours (which drive capacity requirements), US-REGEN also employs a clustering algorithm to select additional segments to ensure that the interior region of the joint distribution of load-wind-solar is adequately sampled. These additional interior hours describe shoulder and base operating conditions and better capture region-specific load duration curves, capacity factor distributions, and correlations between load and intermittent resources. Although adding these cluster hours results in longer runtimes, the model demonstrates a better fit for regional outputs like load duration curves. The model uses k-means clustering to partition the observations for regional load and capacity factors for intermittent technologies, including all resource classes, into a specified number of mutually exclusive sets ("clusters") by minimizing the within-cluster sums of point-to-centroid distances. The selected hours are the segments closest to the cluster centers ("centroids"). This partitioning method provides representative clusters that contain hours with characteristics that are as close to each other (and as far away as observations in other clusters) as possible. The blue markers on Figure 2‑15 represent the hours chosen by the algorithm (both extreme and cluster hours) plotted for Texas.
In the 2019 version of US-REGEN, the extreme hours are chosen to achieve the specified tolerances, then sufficient clustering hours are added to total 120 representative hours for the default 16 region version of the model. For comparison, the 2016 version of EIA's National Energy Modeling System (NEMS) uses 9 hours in its electric sector expansion model. Blanford et al. (2018) provides an extended discussion and a comparison to other segment selection methods. When US-REGEN uses different aggregations of states to regions, the number of hours chosen may vary as the number of extreme hours is chosen endogenously to satisfy a given tolerance limit. Additionally, the model chooses a different set of hours for each time period when run in conjunction with the end-use model, as the load shape varies over time as the end-use mix changes.
Weighting Chosen Hours
Once the representative hours have been chosen, they must be weighted such that the sum of weights equals 8,760. That is, for each moment described by a representative hour, in what fraction of the year do those conditions prevail? Since the conditions in a given representative hour likely vary significantly across regions and since only one set of weights can be applied, it is an over-constrained problem to select mean-preserving weights (i.e. weights such that total load and average annual capacity factor for the aggregated distribution are equal to those in the hourly distribution). Thus, the objective of the weighting procedure is to minimize the sum of squared normalized errors between the aggregated averages and the hourly averages across regions for load and each wind and solar class. To avoid numerical problems associated with very small weights, we enforce a lower bound of one (i.e. each representative hour gets a weight of at least one hour). This formulation is easily solved by non-linear optimization in GAMS with errors of 5% or less.
In summary, the first step of the aggregation heuristic is designed to ensure that the shapes of load, wind, and solar relative to each other are adequately represented, and the second ensures that magnitude of load and the wind/solar resource is not significantly altered. A sample of the results is shown for load, wind, and solar photovoltaic output in Texas in Figure 2‑16, Figure 2‑17, and Figure 2‑18. In each figure, the duration curve (i.e. plot of time series values sorted in descending order) is shown in black for the hourly data (smooth curve) and in red for the aggregated representative hour segments (piecewise linear curve). In each case, while there is some deviation, the basic characteristics of the shape and the area under the curve (for which error was minimized) are preserved by the aggregation. Moreover, it can be shown that the distributions for more complex attributes such as load relative to wind are also captured well by the representative hours approach. These graphs are representative of the methodology.
Static Model
The dynamic version of US-REGEN can run in concert with the end-use model over multi-decadal time steps and, as outlined in Design of Aggregated Segments, uses a reduced set of representative hours to reflect load, wind, and solar profiles. To enable full 8760 hourly resolution, US-REGEN can also be run in static mode. This annual version of the model abandons intertemporal optimization in favor of a detailed hourly representation of an annual slice of electric sector operation. Instead of using aggregated representative hours, the static model computes dispatch for each of the full 8,760 hours. In other respects, the operation of the model remains similar. Generation units remain aggregated into capacity blocks that are dispatched together, and the capacity mix is determined endogenously based on a rental formulation for new capacity. Fuel prices and energy demand are fixed, and the model linearly optimizes to balance the load.
The static model is a useful bridge between the sub-hourly timescales that detail the daily operations of the power system and the long-run projections of the sort that US-REGEN's coupled electricity and end-use models generate. The complete hourly generation profile created by the static model helps evaluate the efficacy of the long-run dynamic model by noting the differences between complete hourly generation and demand profiles with the aggregated representative hours methodology of the dynamic model. Because of the rental formulation, there are no retirement decisions. The static model can be started from any given base year reading off the data from a dynamic model run. This allows the static model to test the hour-to-hour response of a projected system state output by the dynamic model, giving a more realistic assessment of the hourly load shape and generation choices for a given future state. In turn, the dynamic model informs the static model's profile of generation assets and outputs a fixed demand, assessing how the intra-yearly challenges affect the shifting equilibrium of generation and end-use demand.
One major advantage of the static model is the ability to consider storage challenges. The hourly model naturally maintains chronology, which allows the model to accurately represent the generation and storage of electricity across adjacent hourly load segments.
The static version of US-REGEN can also be configured to include the supply, transmission, and storage of natural gas. Supply is modeled at the basin level and differentiated by reservoir type. Demand for gas in the power sector is modeled at an hourly level linked to dispatch in the electric model. Demand for gas in the end-use sectors is based on a scenario from the End-Use Model. State-level pipeline and storage facility data is aggregated to the regional level and constrains the movement of gas between supply and demand with daily resolution. The inclusion of this additional structure of the gas market leads to regional and seasonal variation in the natural gas price.
Unit-Commitment Model
The US-REGEN electric model incorporates a relatively simple model of dispatch that excludes several operational costs and constraints such as ramping and minimum load levels due to the high computational cost of including them in an inter-temporal perfect foresight model. In recognition of these limitations, a standalone unit commitment (UC) version of the US-REGEN electric model has been developed to better understand the short-run costs and engineering challenges of operating the different capacity mixes output from the US-REGEN dynamic model. This model runs separately from the full US-REGEN model and does not iterate with the end-use model.
This UC version of the model solves for most units in a region for all hours in a single year, given a fixed capacity mix. It determines commitment and dispatch states for individual units, with the objective of minimizing operating costs, while accounting for technical system constraints and chronological operations. The goal of this approach is to integrate the capacity planning perspective (i.e., examining long-run investment decisions) with a UC and economic dispatch one (i.e., understanding the short-run costs and engineering challenges of operating different capacity mixes). This framework provides a test bed for assessing flexibility needs in the context of endogenous investments and regional heterogeneity.
The UC model determines the startup, shutdown, and operating schedule (including unit-specific output levels) for every unit during each hour of an annual time horizon. Combining economic dispatch with UC constraints results in a mixed-integer optimization problem with the objective of minimizing total system operating costs. The four primary cost elements in this objective function are variable O&M costs, fuel costs (with output-dependent heat rates), startup costs, and shutdown costs. The model accounts for multiple constraints, including a load balance condition for each region, maximum and minimum output levels for each unit, transmission constraints, optional operating reserve requirements, startup and shutdown logic for generators, minimum up and down times, and maximum ramp rates. Fuel use characteristics and emissions are a function of unit-specific output levels.
The UC model retains individual unit detail for a majority of the fleet in the region of interest. Decision variables related to operation are indexed over the set of all units in the US-REGEN region greater than 40 MW. Since intra-regional transmission is not modeled, variable generation resources across a model region are aggregated by their capacity types and dispatched as blocks. Wind and solar technologies can be curtailed during periods of over-generation.
Given the significance of transmission and trade in influencing electricity market outcomes, a novel feature of the US-REGEN UC model is its endogenous treatment of imports and exports. Trade may be an important flexibility resource to facilitate the exchange of electricity across regions during periods of surpluses or deficits, especially as intermittent resources comprise a greater fraction of generation and regional electricity markets become more tightly integrated. However, most UC models make simplifying assumptions about imports and exports, often assuming that future trade flows will mimic historical patterns. US-REGEN's integrated perspective models many regions at once to capture the increasingly interconnected landscape for system balancing. Cross-border flows are restricted by net transfer capacities, which are influenced by transmission investments in the dynamic model. To make the UC model of the entire US computationally tractable, US-REGEN has individual unit detail in the region of interest but aggregates units into capacity blocks for all other US regions. This formulation endogenously determines price-responsive trade positions.
Full documentation for the unit commitment version of US-REGEN is maintained in EPRI publication 3002004748 (EPRI, 2015).
Policies
US-REGEN has the capability to model diverse energy, electricity, and emissions policies. By default, the electric sector model includes a suite of current policies including emissions constraints, regional greenhouse gas targets, taxes and subsidies, and state renewable portfolio standards (RPS) and clean energy standards (CES). The model has also been configured to represent a variety of other proposed policies, such as the Clean Power Plan, national carbon constraints, CO2 taxes, clean energy standards, and emission intensity standards. These can be enabled or disabled, reconfigured, or replaced entirely.
Non-CO2 Emissions Constraints
A variety of policy scenarios covering emissions of non-CO2 pollutants are considered. In addition, the model can simulate regional or national caps on CO2 emissions as well as price-based carbon policies. These policies can be linked with non-electric abatement through iteration with the end-use model. The complete list of current policies changes frequently with announcements, and the below documented policies may not be the full list implemented in the reference current policy scenario.
Tax and Subsidy Policies
The US-REGEN base case includes the following tax incentive programs, and the model can easily incorporate additional tax incentives, restricted by year or region:
The Federal Production Tax Credit (PTC) for wind generation. This is phased out over time per the Consolidated Appropriations Act 2016 (Title III; Sec. 301)
The Federal Business Energy Investment Tax Credit (ITC) currently available for solar generation. This is reduced over time per the Consolidated Appropriations Act (Title III: Sec 303)
The 45Q tax credit for CO2 capture and storage applies a $50/tCO2 tax credit on plants under construction before 2024 with installed capture and storage equipment.
Regional Greenhouse Gas Policies
Regional greenhouse gas policies
The Regional Greenhouse Gas Initiative (RGGI) places a cap on CO2 emissions from the electric sector in nine Northeast states.
The existing California Cap and Trade scheme, informally known as 'AB32', caps CO2 emissions on the whole California economy. In electric-model only scenarios, the California cap can only be represented as a price on CO2.
California's SB100 sets a target of 60 percent generation from renewables by 2030 and 100 percent "clean" generation by 2045.
New York has committed to a goal of 100% GHG reduction from the electric sector by 2050.
State Renewable Portfolio Standard Policies
The model includes a representation of state-level renewable portfolio standard (RPS) requirements as of mid-2019. The renewable targets by state by year are aggregated to regions by taking the load-weighted average. Although exact translation of state targets to those regions containing multiple states is not possible, the key aspects of these policies have been included to the extent possible. In each region there is a minimum for non-hydro renewable generation as a percentage of retail sales based on the targets adopted by individual states within the region and the relative size of that state's load. In states such as New York where the target includes hydro, expected generation from hydro is removed and only the portion of the target likely to be satisfied by non-hydro renewables is included. The resulting renewable minimums for 2030 in each region are shown in Figure 2‑19.
In addition to the targets themselves, there are many implementation details relevant to the compliance with state RPS policies. These include restrictions on trade of renewable energy certificates (RECs); whether RECs must be bundled (i.e. purchased along with physically delivered power from a renewable generator) or can be unbundled (i.e. purchased on a national market); whether an alternative compliance payment (ACP) is allowed and at what price; and carve-outs for particular technologies such as solar PV and offshore wind. Many state RPS policies contain carveouts for solar PV and offshore wind, and these mandates are included as part of the overall standard—as one example, New York currently requires 6GW of rooftop solar PV and 9GW of offshore wind. Each of these features has been incorporated into the model as specified in the various statutes.
Technology Standards
In addition to existing state-level renewable standards, the model can simulate other regional or national standards, such as a clean energy standard including non-renewables such as nuclear and CCS technology. The reference scenario includes state-specific prohibitions on new nuclear construction. Specific design details such as technology crediting or flexibility through trade can also be represented.
Note that investment is assumed to be uniformly distributed across the years within each 5-year time period. ↩︎
When there is a loss factor on inter-region transmission, this condition holds when the marginal dispatch cost exceeds the adjacent region's marginal dispatch cost plus a loss adjustment. In some cases we also include a $4/MWh charge on inter-region transmission flows, as described below in Section 2.3.6. ↩︎
A similar condition exists for non-retirement decisions. All capacity has a fixed lifetime at which it must retire, but because there is also a fixed O&M charge per unit of capacity installed (regardless of dispatch), it may be optimal to retire early capacity that does not justify this cost. ↩︎
Usually this will be the segment corresponding to peak load. However, when there is a large share of output from variable resources with zero dispatch cost, the relevant peak can shift to the segment with the highest load net of variable output. Additionally, the shadow price on some infra-marginal segments may include a smaller capacity component, since the model effectively clears a capacity market in each segment, rather than clearing a single annual market for the peak, Finally, marginal capacity costs may be allocated across time periods due to subtle inflections in the dynamics of load growth and capacity evolution. ↩︎
Although this cannot be tested in the dynamic setting, we can evaluate the success in reproducing results in the static setting, which can be run with all 8,760 hours. ↩︎
This result was obtained through the application of the mixed-integer solver CPLEX-MIP in GAMS. With a smaller "bubble" radius, the minimum number increases. The number of representative hours vs. the tolerance of the "bubbles" reflects a trade-off between speed of computation in the model (which is highly sensitive to the number of segments) and accuracy of the approximation. ↩︎