AVERT Questions and Answers

AVERT Inputs

1. Are users restricted to the energy profiles created within AVERT’s Main Module?
2. Can I review RE options other than wind and solar generation?
3. Is there a way for baseload renewables to be included?
4. How do you handle biomass, waste combustion generators, or CHP generators in AVERT?
5. Are there any plans to incorporate electricity production from biogas facilities into AVERT?
6. How does AVERT account for the dispatch of new RE into the existing system?
7. How does AVERT account for the dispatch of new EE into the existing system?
8. Is there a bound on the smallest change that is appropriate to model?
9. Is there a bound on the largest change that is appropriate to model?
10. Why might actual planned offshore wind projects trigger a warning in AVERT?
11. Are energy changes applied over the whole region? Is there a way to apply them at the state, county, or municipal level only?

AVERT Results

1. Does AVERT account for losses?
2. How can users assess the accuracy of the results returned by AVERT’s Main Module?
3. What is the accuracy of the map chart in AVERT’s Excel-based Main Module?
4. Why do some EGUs show positive increases in generation with decreases in system load?
5. What kinds of emission rates does AVERT produce?
6. How should AVERT's results be applied to resource impacts longer than a year?
7. Does AVERT account for lifecycle emissions?
8. Can AVERT results be used for spatial analyses?
9. Can AVERT results be used for mobile source regulatory analyses?

AVERT Statistical Module

1. Why is the model driven by system fossil generation instead of by demand or total load?
2. Does the sum of all unit generation in any given Monte Carlo run add up to the size of the load bin (i.e., the expected fossil generation)?
3. Is it possible to displace baseload EGUs in AVERT?
4. Can AVERT capture or replicate ramping behavior?
5. Can AVERT capture or replicate spinning reserves behavior?
6. Can AVERT capture transmission constraints or changes in transmission?
7. Are recent fuel prices reflected in AVERT?
8. Are emissions control technologies reflected in AVERT?
9. Are predicted changes in fuel or emissions prices reflected in AVERT?
10. What other tools are available to me to estimate changes in emissions aside from AVERT?

Future Year Scenarios

1. Why is there a different future year template for each historical baseline year?
2. Why are some generators excluded from AVERT’s Future Year Scenario Template?
3. Why do some generators appear in the Future Year Scenario Template but not in the RDF?
4. In the future scenario demo, do the total avoided emissions include the impact of retired units, or just the energy impacts adjusted for retirements?
5. Does EPA provide projections for use in AVERT’s Future Year Scenarios?

Electric Vehicles

1. If a new electric car is expected to be charged for the next 12 to 15 years, how should we consider the results generated by AVERT?
2. Which power plants does AVERT assume charge EVs?
3. Should I use AVERT’s definition of “new” vehicle or “existing” vehicle?
4. What other sources for charging profiles are there? How would I use them in AVERT?
5. Can the vehicle emissions be exported to COBRA?

Energy Storage

1. What kinds of energy storage technologies does AVERT model?
2. Are there any online resources for modeling other kinds of energy storage technologies besides lithium-ion battery storage?
3. Which power plants does AVERT assume charge energy storage?
4. What are the default energy storage parameters in AVERT?

AVERT Inputs

1. Are users restricted to the energy profiles created within AVERT’s Main Module?

No. The Main Module maintains simple wind and solar profiles for various regions of the United States for the convenience of users, but does not restrict users to these profiles. Users are encouraged to create energy profiles that reflect their regions and assumptions. Such profiles can be copied into the manual entry page of the Main Module.

2. Can I review RE options other than wind and solar generation?

Yes. New non-intermittent, must-take renewable generation, such as hydroelectric generation or geothermal generation, can be approximated using either the manual hourly data entry or the preset EE sections. For example, if you want to model the expected changes resulting from a new hydroelectric generator, you could click on “Enter hourly data manually” in AVERT’s Step 2 and enter the expected hourly generation curve. If you assume the non-intermittent resource functions as a purely baseload resource, you could use the “annual GWh” setting as a proxy.

3. Is there a way for baseload renewables to be included?

You can model changes in non-emitting, must-take baseload renewables like geothermal or hydroelectricity in AVERT using two different methods.

1. You can use the “annual GWh” setting in Step 2. This means you are essentially modeling an EE program as a proxy for baseload renewables. In order to account for transmission and distribution losses, you must first correct your total annual GWh value by reducing its value according to the value of the T&D losses for that year. You can find annual T&D losses in Table 2 of the “Library” tab in the Main Module.
2. If you have an 8,760 hour profile of your EGU, you can enter the profile directly into the “Manual Energy Profile Entry” page of the Main Module. Note that in this case adjustments to T&D losses are automatically included.

4. How do you handle biomass, waste combustion generators, or CHP generators in AVERT?

If biomass, waste combustion, or CHP generators are emitting and have capacities greater than 25 MW, they are included in EPA’s Power Sector Emissions Data.

AVERT is not currently equipped to estimate the emissions of emitting generators that do not report to CAMD. However, if you know the expected generation and emissions from a new biomass, waste, or CHP generator, you could review the estimated displaced emissions and generation from the inclusion of that generator using AVERT (assuming an hourly energy profile is known for the new EGU) and then add in that generator’s emissions post hoc. To do so, follow the steps below:

1. Determine the estimated energy profile for the CHP generator and associated stack emissions.
2. Input the energy profile for the CHP generator into AVERT under “manual energy profile entry” in Step 2. If the resource is utility-scale (e.g., a central station power plant providing power to wholesale markets), its impacts should be entered in the first column. If it is distributed (e.g., located at the customer site or behind the meter), its impacts should be entered in the second column.
3. Run the Main Module to determine the emissions offset due to the new CHP generator.
4. Subtract the CHP stack emissions from emissions offsets to determine the total change in emissions:

        net emissions reduction from CHP generator = AVERT displaced emissions + CHP stack emissions

There is no current option to review emissions displaced from new biomass, waste, or CHP generators if they do not already report to CAMD.

5. Are there any plans to incorporate electricity production from biogas facilities into AVERT?

If the facility is an emitting generator and has a capacity greater than 25 MW, it is currently included within the AMP EGU dataset. Otherwise, there are no current plans to incorporate electricity production from these types of facilities. Often, these facilities may generate electricity according to their onsite needs and fuel supply and may not be affected by regional changes in load or dispatch.

6. How does AVERT account for the dispatch of new RE into the existing system?

RE generation sources typically have very low variable operating costs; in other words, they are very inexpensive to operate once they are constructed. Typically, low-operating-cost resources are dispatched first, and increasingly expensive resources are dispatched thereafter. RE sources are assumed to dispatch first (a common assumption across many economic dispatch models), and thus can be modeled as an equivalent reduction in demand. AVERT simply compares the generation and emissions of all fossil resources before the new RE resource (i.e., at the equivalent of full demand in each hour) and after the new RE resource (i.e., at the equivalent of a reduced demand in each hour). The difference in generation and emissions between the before and after scenarios represents the emissions displaced by RE.

7. How does AVERT account for the dispatch of new EE into the existing system?

EE usually results in a reduction in demand (in some cases and for some types of programs, it may result in a shifting of demand to off-peak hours). AVERT simply compares the generation and emissions of all fossil resources before the new EE resource (i.e., at the equivalent of full demand in each hour) and after the new EE resource (i.e., at the equivalent of a reduced demand in each hour). The difference in generation and emissions between the before and after scenarios are the emissions that are displaced by EE.

8. Is there a bound on the smallest change that is appropriate to model?

No, there is no specified lower bound, but users may find the following guidance useful when analyzing small changes. Users can review the output chart titled “Hourly Results by Week” for an indication of how closely their expected energy changes are captured in hour-to-hour unit changes for one week. For very small inputs, this graphical interface will indicate a rougher hour-to-hour energy profile—i.e., the resulting change in generation will look less like the amount of energy change expected. Note that all numerical results are shown rounded to the nearest 10 unit.[1] Dashes indicate that AVERT reported a value greater than zero, but lower than the level of reportable significance. In some cases, no reasonably sized energy policy will result in reportable changes. For example, the review of monthly results for a single small county in a low load month may often result in low significance results. However, the user can use discretion to determine if an energy policy has resulted in an acceptable level of significance based on the signal-to-noise diagnostic and the degree to which critical results are below the level of acceptable significance (i.e., are obscured by dashes in numerical results).

For some smaller inputs, users may find increased “noise” in a given model run. Although the resulting annual changes in emissions or annual average emission rate may resemble what one might expect for results, based on the inputs, results for individual EGUs may exhibit more volatile behavior. For example, in a modeling run that is intended to reduce fossil load (e.g., the user has input some quantity of EE or RE), some individual EGUs may exhibit decreases in generation (as one might expect), while others may exhibit unexpected increases in generation. For a check of these results in aggregate, the user should view the “Signal-to-noise diagnostic,” found on the “Display Results” page of AVERT’s Main Module. As described in the AVERT User Manual, this scatter plot shows the changes in generation calculated by AVERT (on the y-axis) against the energy change input by the user. More reasonable results (from a program size perspective) will appear closer to 1:1 lines. Smaller load changes have more noise (i.e., scatter) in this plot, while larger load changes have a straighter line relationship. The R² value in the title of the chart indicates how much of the change in generation can be explained by the user-input energy change. For examples, an R² value of 0.9 indicates that AVERT has captured 90 percent of the change in generation required by the user, while a value of 0.7 indicates that AVERT has only correctly captured 70 percent of the energy change input by the user (i.e., noise accounts for 30 percent of the observed variability).

Figure 1 shows two different energy profiles with very different R² values from the same region, and designed similarly. The graph on the top is a 1.5 percent load reduction during the peak 20 percent of hours. The reduction is sufficiently sized such that the generation reduction is able to match the requirement very closely—over 99 percent of the reduction in generation is a direct result of the energy change input. The graph on the bottom is a 0.25 percent load reduction during all hours of the year. The reduction is insufficiently sized in this case and results in a wide range of uncertain results. By comparison, 92 percent of the generation reduction can be attributed to the energy policy—the rest is noise.

In general, modeling runs that produce a high level of noise (i.e., a low R²) may be useful for describing high-level results (such as annual and regional changes in emissions) but may be less useful for describing changes in generation or emissions on an hourly basis or at any one individual EGU. Modeling runs with comparatively less noise (and larger R² values) are better suited for these purposes.

Figure 1. Examples of two different load reductions with different R2 values in the signal-to-noise diagnostic. 

Top: 1.5 percent load reduction in peak 20 percent of hours. Bottom: 0.25 percent load reduction in all hours.

9. Is there a bound on the largest change that is appropriate to model?

There is not a formal bound on the largest project, program, or policy that should be modeled in AVERT. In general, users should note that AVERT is designed to review marginal operational changes in load, rather than large-scale changes that may change fundamental dynamics. As a guideline, EPA suggests that modeled scenarios generally not deviate 15 percent from baseline fossil generation in any given hour.

With this 15 percent guideline, analysts should use their judgment in deciding whether the results are appropriate for their uses. To assess appropriateness of results, analysts can consider the number of hours out of 8,760 (the number of hours in a year) that exceed the 15 percent threshold and how much greater than 15 percent the resultant fossil generation values are. Analysts should also consider their specific interest in using AVERT. For example, an analyst interested in only annual results may likely be less sensitive to the 15 percent guideline than an analyst interested in the hourly results that span the hours where the threshold is exceeded.

In addition, users may encounter situations in which their selected change to load produces a new hourly load that is outside the range calculable by AVERT. This may occur in situations where load is reduced by 50 percent or more relative to the hour with the lowest load (i.e., well outside the recommended threshold of 15 percent). In situations where load is increased, users may encounter this issue at load increases as low as 10 percent for some regions. In these situations, users should refine their load changes such that an error is not produced on the “Manual User Input” page.

For reference, Table 1 provides each AVERT region’s total annual load, in GWh. Knowing the total annual load in each region may be helpful when using AVERT’s “reduce generation by X% in all hours” option to model the impacts of a broad-based EE program. Figure 2 identifies the distribution of hourly loads in each of the 14 AVERT regions, using data from 2022. Table 1 also shows the maximum and minimum possible load levels able to be modeled in AVERT. These values provide helpful context if one knows the absolute size of a project, policy, or program in units such as MW and one wants a sense of the percentage of regional load that it represents. Note that fossil loads in Table 1 and Figure 2 have not been adjusted to reflect the transmission and distribution losses inherent to each region. Demand-side measures (such as EE or distributed solar) avoid not only electricity demand, but also the electricity associated with transmission and distribution losses. As a result, demand-side programs increase the avoided fossil load by an additional 5–9 percent, depending on the region and the year being analyzed.

Table 1. Total regional fossil loads in AVERT regions, 2023.

AVERT region	Total annual fossil load (GWh)	Maximum possible hourly load (MW)	Minimum possible hourly load (MW)
California	81,566	27,005	1,439
Carolinas	96,980	22,874	4,908
Central	144,949	39,314	3,770
Florida	189,241	38,313	11,146
Mid-Atlantic	467,577	105,955	29,681
Midwest	466,291	99,561	26,739
New England	49,092	12,550	1,714
New York	63,511	19,008	3,219
Northwest	119,831	21,063	3,553
Rocky Mountains	51,394	10,335	2,646
Southeast	172,132	36,852	11,301
Southwest	83,235	16,709	3,113
Tennessee	76,192	15,792	3,608
Texas	271,173	61,440	9,865

Figure 2. Characteristics of regional hourly fossil loads, 2023.

10. Why might actual planned offshore wind projects trigger a warning in AVERT?

The Excel Main Module gives the user a warning when they have entered an energy profile that collectively exceeds 15 percent of load in at least one hour of the year. This warning could appear when modeling certain large policies, programs, or projects, including ambitious offshore wind projects that may be implemented in the near future. However, this warning will not prevent the user from modeling large quantities of offshore wind—the user can simply ignore the warning message if desired.

11. Are energy changes applied over the whole region? Is there a way to apply them at the state, county, or municipal level only?

Because AVERT does not model transmission constraints within a region, energy changes are assumed to have change emissions throughout the selected AVERT region. A limitation of AVERT is that it is insensitive to the physical location within a region of new projects, programs, or policies, despite the fact that real-world dispatch decisions may be quite sensitive to specific locations of resources resulting from energy policies as well as EGUs. AVERT assumes that energy changes are spread across the modeled region. It cannot currently identify the differential effects of local versus regional energy changes. Such differentiation requires the use of a production cost model.

For more information, see “Limitations and Caveats” in Section 2 of the AVERT User Manual. Detail on changes at the state and county level are available on the output sheet “Annual Results by County.”

AVERT Results

1. Does AVERT account for losses?

Yes, AVERT accounts for three types of losses. First, gross generation as collected in the Power Sector Emissions Data is corrected to account for parasitic consumption of energy onsite at fossil-fired EGUs. AVERT applies a parasitic loss factor to each EGU based on unit and fuel characteristics and subsequently calculates emissions based on each unit’s “net” output of energy exported to the grid. Second, reductions in fossil load due to EE and distributed PV (and increases in fossil load from increased demand) are corrected to account for avoided grid (transmission and distribution) losses, using region-specific, year-specific grid loss factors. Wind and utility-scale PV profiles are not corrected for these losses, as it is assumed they are located at a similar distance from load centers as fossil-fired EGUs. Finally, additional loss factors are assumed for offshore wind due to the fact that these resources are commonly located far from load centers, and because associated transmission lines may be underground or underwater. Using loss estimates from NREL, these factors act to reduce the hourly capacity factors of offshore wind resources.[2] See Appendix C of the AVERT User Manual for more detail on how offshore wind capacity factors have been developed in AVERT.

2. How can users assess the accuracy of the results returned by AVERT’s Main Module?

The current version of the Main Module is not equipped to return information on the accuracy or uncertainty of the model results.

While the Monte Carlo analysis run by AVERT creates information useful for some forms of uncertainty analysis, using this information to assess the accuracy of the results returned by the Main Module would require simultaneously performing a Monte Carlo analysis on the baseline scenario and on the modified scenario, and returning uncertainty metrics associated with the difference between these two scenarios. The current version of AVERT does not contain this information. EPA is exploring a future version of AVERT that could perform explicit uncertainty analyses and allow users to assess the accuracy of results returned by the model.

3. What is the accuracy of the map chart in AVERT’s Excel-based Main Module?

The map is a visual cue only, and should not be used as a precise rendering of the location or influence of change in generation or emissions from any EGU or cohort of EGUs. Maps could be used for visual presentations to show the general location of emissions.

4. Why do some EGUs show positive increases in generation with decreases in system load?

Some EGUs show positive increases in generation with decreases in load because the EGU statistics indicate either a slight increase in the probability of operation at very low loads or an increase in generation at very low loads. Spot checks indicate that most of the EGUs that show generation increases with decreases in system load are due to baseload EGUs that show a lower probability of generation at mid-range loads than at either very high or very low loads. In other words, these EGUs counterintuitively increase the probability of operation as system load levels become very low. Further inquiry into these EGUs suggests that they have prolonged maintenance outages during spring or autumn—i.e., during periods of generally low load, but possibly not the lowest in the year. Therefore, the EGU will register as non-operational through a wide swath of medium-low loads, but may operate during the very lowest loads of the year. Therefore, the statistics capture this behavior and increase expected generation by a small margin when system load is reduced from very low load periods. This pattern is almost always observed in trough periods.

5. What kinds of emission rates does AVERT produce?

AVERT’s Annual Regional Results table displays information on modeled emissions, generation, and emission rates. Two types of emission rates are shown on the lower box of this page: “Average Fossil” and “Marginal Fossil.”

The first column labeled “Average Fossil” is an emission rate calculated by dividing the mass (tons or pound) of each pollutant in the baseline (“Original”) by the level of generation (MWh) in the baseline (“Original”). These values are derived from the power plants in the AVERT dataset prior to any user-defined change. This is an annual average emission rate of those EGUs in EPA’s Power Sector Emissions database.[3] It is specifically a “fossil” rate because it does not include generation or emissions from other power plants, including nuclear, hydro, wind, solar, or other plant types. See U.S. EPA’s eGRID tool for information on average emission rates that are inclusive of these other EGU types.[4]

The second column, labeled “Marginal Fossil,” is calculated by dividing AVERT’s estimated change in emissions by AVERT’s estimated change in generation. This predicted change in emissions and generation is the impact on the power sector as calculated by AVERT due to the user-defined scenario. This value is called a marginal rate because it describes a change in emissions per unit change in generation. As AVERT only models fossil-fired EGUs in the Power Sector Emissions database—not other types of power plants—we refer to this as a “Marginal Fossil” rate.

Although the Annual Regional Results focus on annual emission rates, it is also possible to calculate more temporally detailed “Average Fossil” and “Marginal Fossil” emission rates using data in the advanced outputs. AVERT reports both original and changes in generation and emissions for every hour. Users can divide an emission quantity by the corresponding hourly generation and calculate either “Average Fossil” and “Marginal Fossil” emission rates for a single hour. Users can also use these data to create weighted average values for weeks, months, or other time periods.

6. How should AVERT's results be applied to resource impacts longer than a year?

AVERT is best used to conduct short-term analyses, and EPA recommends that users restrict their analyses to a five-year time horizon from the year of the RDF. Analyses that extend past this window may intersect with years when the operation and composition of the grid is substantially different from the baseline year and may not include the structural change that may have been caused by the intervention modeled.

More specifically, the emission rates calculated by AVERT can be classified as "short-run" marginal emission rates (SRMER). These differ from “long-run” marginal emission rates (LRMER) in that short-run rates are focused on operations of the current grid, holding the structure of the grid fixed, while long-run rates incorporate both operational and structural changes to the grid. Each approach is appropriate for distinct purposes. Short-run approaches are appropriate for characterizing the near-term impacts of an intervention prior to the point where structural changes occur. 

The medium- to long-term impact of resources like electric vehicles (EVs) and heat pumps on the grid is also of interest to many users. While AVERT is not intended to answer these questions, EPA has built a reference output to help put the AVERT results in context, the output page titled "Reference: Modeled Emission Rates Over Time." On this page, users can compare the CO₂ emission rate as modeled in AVERT with a set of emission rates modeled by NREL in its Cambium data set. CO₂ SRMER and LRMER are provided for different approaches and different years, allowing users to visualize how grid impacts of interventions are likely to change in the future.

For more information on this topic, see the AVERT User Manual.

7. Does AVERT account for lifecycle emissions?

No. At this time, AVERT users are only able to analyze emissions related to direct combustion (EGU and ICE vehicle) and select emissions relating to fueling and the volatilization of fuel in ICE vehicles. Users interested in exploring lifecycle emissions may wish to utilize the Greenhouse gases, Regulated Emissions, and Energy use in Technologies (GREET) Model developed by Argonne National Laboratory.[5]

8. Can AVERT results be used for spatial analyses?

AVERT contains two types of data with geospatial attributes. The first type of data is point source (latitude and longitude) emission data from EGUs. The annual-aggregated EGU emission data with latitude/longitude attributes can be accessed by viewing the tab titled “Summary” in the Excel file. AVERT also presents this annual EGU data aggregated at the county-, state-, and regional-level in some outputs. (Note that AVERT users will have to click the button labeled “Click here to restore default Excel data” on the welcome page of the model to access this “Summary” page.) Users can utilize the latitude and longitude locations to visualize the annual data displayed on this page, or apply this latitude and longitude data to hourly, EGU-specific data for all six pollutants, which can be accessed by viewing the tabs titled “SO₂ (lb),” “NO_X (lb),” “CO₂ (short tons),” “PM_2.5 (lb),” “VOCs (lb),” and “NH₃ (lb).”

The second type of geospatial data is avoided ICE vehicle emissions aggregated at the county-level (AVERT also presents these data at the state and regional level). These data are a generalization of the emissions not occurring on roadways in counties where EVs are being deployed according to the user’s scenario. These data are most easily accessible for geospatial analysis (via FIPS codes) through the page titled “Output: Annual Results by County, Including Vehicles.”

Users may be interested in utilizing these spatial data to understand how policies and programs modeled in AVERT affect overburdened communities or other communities of interest. To support these equity analyses, users could employ a geographic information system (such as ArcMap, QGIS, Google Earth, or another program) to analyze and visualize AVERT data alongside EJScreen[6] or another spatial dataset of interest.

9. Can AVERT results be used for mobile source regulatory analyses?

No. AVERT may not be used for mobile source regulatory analysis, including SIP and transportation conformity analyses. Consult the most recent EPA guidance document for applying EPA’s MOtor Vehicle Emission Simulator (MOVES) model at: https://www.epa.gov/moves/latest-version-motor-vehicle-emission-simulator-moves.

AVERT Statistical Module

1. Why is the model driven by system fossil generation instead of by demand or total load?

AVERT is a statistically based model that tracks and reproduces EGU behaviors. EGUs are forecast to operate in the near future much as they operate today. In electrical system dispatch, determining how an EGU operates is largely driven by two factors—total demand on the system and the cost of operation for any given EGU. In economic dispatch, more expensive EGUs are dispatched at higher levels of demand.

In AVERT, the degree to which an EGU will be dispatched in the future is assumed to be the same as the degree to which it has been dispatched at a historical level of demand. This assumption incorporates the relative operational cost of different EGUs. In other words, EGUs that dispatched at high demand were likely more expensive to operate and thus are likely to be on the margin at the same level of demand.

However, if AVERT were to observe the behavior of EGUs against total system demand, it would miss an important factor: the large number of non-fossil EGUs that are dispatched at very low operating costs—such as solar, wind, hydroelectric, and nuclear operations. These non-fossil EGU have the same effect as lowering system demand, or specifically, system demand that needs to be met by fossil generation. Therefore, AVERT dispatches against demand for fossil resources, rather than total system demand. New EE or RE policies are assumed to reduce the demand for fossil resources. Figure 3 illustrates the difference between total system demand and demand for fossil resources.

Figure 3. Diagram schematic of system demand over two days, divided into fossil and non-fossil components illustrating system and fossil demand.

2. Does the sum of all unit generation in any given Monte Carlo run add up to the size of the load bin (i.e., the expected fossil generation)?

Not necessarily. The generation from each EGU is calculated independently in each Monte Carlo run, meaning that there is no constraint that forces the output of all EGUs to equal the exact size of the load bin. The total sum of all unit generation from any given Monte Carlo run may be slightly larger or smaller than the load bin. However, over large numbers of Monte Carlo runs, the average output of each EGU will sum quite closely to the expected fossil generation, or the size of the load bin.

3. Is it possible to displace baseload EGUs in AVERT?

Yes. AVERT treats baseload EGUs, or units that run during most hours, including during baseload hours of the year, the same as all other EGUs. There is no distinguishing characteristic that either promotes or prevents an EGU from running during any given hour, except for how it has operated in the past. If an EGU has experienced little downtime in the past and operates continuously even a low levels of load, the model will replicate this behavior going forward. This type of EGU is unlikely to be displaced by EE or RE programs in AVERT. However, an EGU that ramps from high output in the daytime to low output on off-peak hours may show a displacement if system demand declines due to RE or EE.

4. Can AVERT capture or replicate ramping behavior?

No. AVERT performs a separate calculation for each hour of the year and does not evaluate the rate at which an EGU increases or decreases generation. Capturing that behavior requires a chronological dispatch model.

5. Can AVERT capture or replicate spinning reserves behavior?

Yes. AVERT captures historical generation patterns. EGUs that maintain a spinning reserve (i.e., maintain a minimum level of generation in most operating hours) will reflect this pattern in the statistics gathered by AVERT. The Lake Hubbard EGU shown in the AVERT User Manual is a classic example of an EGU that appears to maintain a spinning reserve. It maintains an output of about 150 MW per hour for most hours of the year. However, when system demand climbs above 45,000 MW, it quickly climbs towards an output of 500 MW.

6. Can AVERT capture transmission constraints or changes in transmission?

Generally, no. AVERT operates on the simplifying assumptions that there are no transmission constraints between load centers and EGUs within a region and that regions are independent of each other. Therefore, AVERT is insensitive to the location where new EE or RE resources are placed within a region, and thus does not capture transmission constraints. However, the behavior of some EGUs may be influenced by historical transmission constraints, and this behavior is captured by AVERT. For example, in “load pockets,” or areas of constrained inbound transmission, reliability EGUs may run at lower regional load levels than would otherwise be dictated by economic dispatch. Because AVERT is not an economic model, it simply replicates the behavior of these EGUs, which may capture some elements of current transmission constraints.

Due to the same simplifying assumptions that prevent AVERT from operating as a transmission-constrained dispatch model, AVERT cannot capture future changes in transmission, which typically change which future EGUs can compete to provide the lowest-cost energy in a particular area.

7. Are recent fuel prices reflected in AVERT?

Yes. To the extent that fuel prices have influenced dispatch during the base data year you choose, AVERT will reflect those dispatch decisions. AVERT cannot, however, change dispatch based on future economic or regulatory conditions, such as expected fuel prices, emissions prices, or specific emissions limits. AVERT should not be used for this type of analysis, as such changes require an economic dispatch model.

8. Are emissions control technologies reflected in AVERT?

Yes. To the extent that emissions controls were in operation at the time that data were collected in the base data year, emissions will reflect operational (and operating) control technologies. To the extent that a user requires a review of dispatch with different emission rates, they can override observed emission rates using the “Future Year Scenario Template” as described in Appendix F of the AVERT User Manual. Modeling emissions prices, specific emissions limits, or fuel switching requires an economic dispatch model.

9. Are predicted changes in fuel or emissions prices reflected in AVERT?

No. AVERT should not be used for this type of analysis; capturing this behavior requires an economic dispatch model.

10. What other tools are available to me to estimate changes in emissions aside from AVERT?

You can model generation and emissions changes caused by new energy policies in a production-cost dispatch model. These programs simulate real dispatch decisions based on explicit costs and operational constraints, optimizing generator use to minimize costs. Some of these models are sensitive to EGU ramp-rates, transmission constraints, and outage schedules.

For the most part, these models are highly detailed, proprietary, and require specialized labor and licensure to operate and use, as well as some degree of proprietary knowledge for fuel costs and operational constraints.

AVERT provides an alternative, publicly available tool to estimate changes in emissions in near-term years. Users who wish to conduct analyses more than 5 years from the baseline must use AVERT’s statistical module and future year scenario template. This type of analysis requires access to future year hourly, unit-specific generation and emissions data (e.g., from an electric-sector dispatch model designed to forecast future generation) that can be entered in place of AVERT’s historical data.

Future Year Scenarios

1. Why is there a different future year template for each historical baseline year?

AVERT is sensitive to the composition of the electric fossil fuel fleet. Every year, the composition of the fleet changes slightly as new EGUs are added or retired. To accommodate this changing fleet, AVERT creates a new future year scenario template for each historical baseline year. Using a mismatched pair in AVERT’s Statistical Module (e.g., a historical baseline year of 2017 but a future year template of 2019) risks accidentally using proxy “new” EGUs that did not exist in 2017, and thus will not be incorporated into a 2017 analysis.

2. Why are some generators excluded from AVERT’s Future Year Scenario Template?

AVERT considers EGUs that report to EPA’s CAMD only. This may exclude generators with less than 25 MW of capacity or generators that did not operate in a particular year.

3. Why do some generators appear in the Future Year Scenario Template but not in the RDF?

AVERT’s Statistical Module allows users to exclude small, low-generation units from consideration in the emissions analysis. Small peakers have statistics that may be non-representative of expected generation patterns (i.e., they cannot be readily extrapolated or interpreted outside of specific events). By default, AVERT excludes units that have generated less than 1,000 MWh per year. For a 25 MW unit (the smallest reporting unit), this would be the equivalent of 40 hours of generation over the year, or less than 0.5 percent of all possible operational hours.

4. In the future scenario demo, do the total avoided emissions include the impact of retired units, or just the energy impacts adjusted for retirements?

Results from AVERT runs using the Future Year Scenario Template do not include changes in emissions at user-specified retired units. These units are assumed to be retired in both the “before” and “after” cases.

The purpose of the retirements category is to exclude from consideration any units that are likely to be non-operational in the future year, regardless of the energy change modeled.

5. Does EPA provide projections for use in AVERT’s Future Year Scenarios?

At this time, EPA is not providing EPA projections for use in AVERT. Future scenarios are meant to be developed by users. You can use AVERT’s future year scenario template to make known changes in the regional dataset. Users who wish to conduct analyses more than 5 years from the baseline must use AVERT’s statistical module and future year scenario template. This type of analysis requires access to future year hourly, unit-specific generation and emissions data (e.g., from an electric-sector dispatch model designed to forecast future generation) that can be entered in place of AVERT’s historical data.

Along these lines, EPA has partnered with the Eastern Regional Technical Advisory Committee (ERTAC), a group of state environmental agency senior staff and multi-jurisdictional organizations (e.g., LADCO, MARAMA, WESTSTAR, SESARM, NESCAUM), to provide AVERT-compatible RDFs for ERTAC-specified custom future years.

EPA will issue periodic updates to the historical data files available for download, but will not release stand-alone future scenarios. At this time, EPA anticipates releasing new RDFs in the second quarter of each year.

Electric Vehicles

1. If a new electric car is expected to be charged for the next 12 to 15 years, how should we consider the results generated by AVERT?

As described in the above FAQ (“Does EPA provide projections for use in AVERT’s Future Year Scenarios?”), AVERT is best used to conduct short-term analyses. In practice, we recommend that users restrict their analyses to a five-year time horizon from the year of the RDF. Analyses that extend past this window may intersect with years when the operation and composition of the grid is substantially different from the baseline year and may not include the structural change that may have been caused by the intervention modeled.

The medium- to long-term impact of EVs on the grid is also of interest to many users as the power sector’s emission rates for many pollutants are expected to decline over time. While AVERT is not intended to answer these questions, EPA has built a reference output to help put the AVERT results in context: the output page titled “Reference: Modeled Emission Rates Over Time.” On this page, users can compare the CO₂ emission rate as modeled in AVERT with a set of emission rates modeled by NREL in its Cambium data set (for more information on this feature, see the AVERT User Manual). CO₂ SRMER and LRMER are provided for different approaches and different years, allowing users to estimate how grid impacts of charging vehicles are likely to change in the future.

2. Which power plants does AVERT assume charge EVs?

AVERT sums the load change from all entered resources (EERE, EVs, and energy storage) in each hour and applies the net change to the fossil power plants in the selected grid region (as defined by the RDF). When modeling EV adoption in the near future, users are recommended to model the amount of EE and RE resources that are estimated to come online during the same period to understand the joint effect of both changes together. For example, if a user is modeling the number of EVs added to the grid over a three-year period, they should also model the amount of EE and RE expected to be added over the same three years.

3. Should I use AVERT’s definition of “new” vehicle or “existing” vehicle?

With AVERT, users can select an emissions profile for a new or existing ICE vehicle. In many situations, users are likely to want to select “new” for this vehicle type. This allows a user to compare the impact of some number of new EVs relative to the same number of new ICE vehicles. In some cases, users may wish to compare the impacts of replacing an existing ICE vehicle with a new EV. In these situations, users should select “existing” vehicle. In both options, the grid impacts associated with charging EVs will remain the same; changing the option between “new” and “existing” only modifies the emission rate associated with ICE vehicle emissions.

4. What other sources for charging profiles are there? How would I use them in AVERT?

One potential source for EV charging profiles is the Electric Vehicle Infrastructure Projection Tool (EVI-Pro) Lite.[7] EVI Pro-Lite is the source for AVERT’s “light-duty vehicle” charging profile, but users can utilize EVI-Pro Lite to develop their own custom charging profiles. Charging profiles derived from this tool may be a better representation than the defaults available in AVERT, as EVI-Pro Lite allows users to modify many different options currently not available in AVERT. Users should reference EVI-Pro Lite materials at https://afdc.energy.gov/evi-pro-lite/load-profile to learn how to use the tool and generate load profiles.

Users should note that they can export the results from EVI-Pro Lite into a CSV file. Some post-processing may be needed before importing the results into AVERT. As of January 2023, the load profiles will produce results in 15-minute intervals, which must be averaged to produce hourly values usable in AVERT.

5. Can the vehicle emissions be exported to COBRA?

Yes. When a user generates a COBRA text file, emission changes will be generated for the power sector as well as the vehicle emissions from the transportation sector. This text file will contain a row of vehicle emission changes for every county in the region currently selected for modeling. These county-specific changes are developed by allocating the total regional emissions changes for each pollutant to each county based on the share of VMT in that county relative to the regional total. When COBRA reads the AVERT-generated text file, the emissions will be automatically classified to the appropriate sector (i.e., electricity generation or “transportation,” as the sector is called in COBRA). Because vehicle emissions are emitted close to the ground (unlike pollutants emitted from EGUs, which are emitted from a stack high in the air), these pollutants tend to not travel far distances and, as a result, tend to be more impactful to local communities.

Energy Storage

1. What kinds of energy storage technologies does AVERT model?

Currently, AVERT models energy storage resources using parameters (DoD, RTE, and number of cycles) reflective of lithium-ion battery storage. Currently, lithium-ion batteries make up more than 90 percent of utility energy storage capacity installed since 2012. Users have the option to modify the parameters embedded in the tool to be more representative of different types of energy storage.

2. Are there any online resources for modeling other kinds of energy storage technologies besides lithium-ion battery storage?

Yes. The Pacific Northwest National Laboratory (PNNL) released a "2022 Grid Energy Storage Technology Cost and Performance Assessment" report, which detailed the findings of an assessment of a number of different energy storage technologies. Aside from lithium-ion batteries, this report provides typical values for parameters such as DoD, RTE, and duration for lead-acid batteries, vanadium redox flow batteries, zinc-based batteries, compressed air energy storage, and pumped storage hydropower. This is just one example option that users can reference when defining their own energy storage characteristics. It is available at: www.pnnl.gov/sites/default/files/media/file/ESGC%20Cost%20Performance%20Report%202022%20PNNL-33283.pdf.

3. Which power plants does AVERT assume charge energy storage?

When modeling energy storage as being paired with solar, AVERT assumes energy storage can only be charged using solar PV resources added by the user in Step 2 of the Main Module. When pairing energy storage with solar, utility-scale storage can only be charged by utility solar PV, and distributed storage can only be charged by rooftop solar PV. When modeling energy storage as being unpaired with solar, AVERT treats a charging battery like any other load served by the grid. For example, when users only enter energy storage, it will be charged by fossil EGUs at the margin.

4. What are the default energy storage parameters in AVERT?

The default energy storage parameters are:

• Number of discharge cycles: 150 cycles per year.
• Days of discharge: The 150 days with greatest fossil load in 2023.
• Duration: 4-hour. Charges and discharges for each 24-hour cycle.
• Round-trip efficiency: 85 percent.
• Depth of discharge: 80 percent.
• Charging profile: Midday.
• PV-plus-storage: Storage is required to be paired with solar PV.

In the AVERT Web Edition, the user can only modify the number of discharge cycles. In the AVERT Excel Edition, the user can modify each of these default values.