Resources Page

Here you'll find a detailed description of each component used by this calculator, broken out by which calculator project module involves them.

Table of Contents

Additional Resources




General Approach

The Emissions Quantification Tool (EQT) compares emissions of NOx, SO2, and CO2 from fossil-fueled electricity generation before and after implementation of a smart grid project. Modules apply project-specific algorithms to create a new load profile after the desired smart grid project has been carried out. This new load profile is then compared with the original to estimate changes in emissions. This estimation is performed through the utilization of regional historic operation data. Different regions will have different generation mixes, and thus emission rates vary widely. The Environment Protection Agency's (EPA) AVoided Emissions and geneRation Tool (AVERT) is behind the 10 region structure, regional load levels, and pollution rates. More detailed information is described in the section below. The data utilized by the AVERT comes from the EPA's (Air Markets Program Data (AMPD) tool). Please see project specific descriptions below for more detail. The figure below provides a high level flow chart for the methodology utilized by the EQT.


Methodology Overview

The figure below provides more detail as to the general methodology employed by the EQT. Begin with a load profile, either a default AVERT region load, or a user uploaded custom load profile. The profile is then scaled to project size via peak scaling (scaling factor is a ratio of the absolute maximum value contained in the load profile to user input project size). Next, module specific calculations are performed in order to create a new load profile. This new profile, along with the original, are then peak scaled in order to match the AVERT analysis region size. AVERT calculates emissions from total load levels, and these emission results are then scaled to project size and reported to the user.

Calculation Overview

The AVERT takes hourly electrical generation and emissions data from the EPA's Air Markets Program Data (AMPD) tool (presently values reported for 2014) and probabilistically estimates the operation and output of each EGU in a region based on a region's hourly demand for fossil-fired generation to determine emission rates for NOx, SO2, and CO2. For more details, see the AVERT user manual. This generation level to emissions mapping is culminated in regional data files (RDFs), which are available for download through the AVERT's website and exported with EQT results. Note that electrical generation units smaller than 25 Megawatts do not report to AMPD and are thus excluded.


Back to the Top

Foundational Literature

The Emissions Quantification Tool was inspired by previous work performed at Pacific Northwest National Laboratory, specifically the report The Smart Grid: An Estimation of the Energy and CO2 Benefits (http://www.pnl.gov/news/release.aspx?id=776). This report evaluates how the smart grid can contribute to the reduction of carbon emissions from the electricity sector.

Back to the Top

Emissions Quantification Mapping utilizing AVERT

As mentioned in the General Approach section, the Environment Protection Agency's (EPA) AVoided Emissions and geneRation Tool (AVERT) provides the core emissions analysis leveraged by the EQT. Detailed information about this tool can be found here: http://epa.gov/avert/.

The AVERT splits the country into 10 regions in order to group similar generation mixes together and to represent how electricity demand is met on a regional level. A map of these regions is shown below.

AVERT Region Map

The AVERT takes hourly electrical generation and emissions data from the EPA's Air Markets Program Data (AMPD) tool and performs a Monte Carlo statistical analysis to determine emission rates for NOx, SO2, and CO2 at several different load levels for each region. For more details, see the AVERT user manual. This generation level to emissions mapping is culminated in regional data files (RDFs), which are available for download through the AVERT's website. Note that electrical generating units smaller than 25 Megawatts do not report to AMPD and are thus excluded.

The EQT uses the aforementioned RDFs to estimate the resultant emissions from a load profile based on the selected AVERT analysis region. This is done for pre and post smart grid project load profiles. Since the RDF files from the AVERT have emissions grouped into discrete load level bins, the EQT interpolates (or extrapolates, as necessary) to determine emission levels for a specific load level. For every hour of the year, emissions levels are compared, and the differences summed to reach a final reduction (or increase) value for each pollutant.

Back to the Top

Additional Transmission and Distribution loss changes

After a smart grid project is implemented and a new load profile generated, this change in load also affects the line losses: the changes in losses are calculated as a constant factor. Distribution losses are evaluated at 3.5%, and transmission losses at 2.5%. Total line loss estimates come from Energy Information Administration: http://www.eia.gov/tools/faqs/faq.cfm?id=105&t=3, and the split between transmission and distribution losses is based on information found in this pdf: http://www.nyserda.ny.gov/-/media/Files/Publications/Research/Electic-Power-Delivery/epri-assessment-losses.pdf.

A user will choose which losses to include; the EQT team recommends the following based on where the project is located:

  • Project at point of consumption (include transmission and distribution losses)
  • Project at substation (include transmission losses)
  • Project uses same transmission lines as bulk fossil fuel generation (include neither)

Depending on the sign of the change in load (positive or negative), the line losses can either increase or decrease. Before emission differences are calculated, the hourly differences between the pre and post smart grid loads are taken, and multiplied by the line loss factor. Each hourly product (whether positive or negative) is then subtracted from each hourly load value in the post smart grid load profile. This final resultant load profile is then used for analysis of emission reductions.

Back to the Top

Energy Storage for Annual Peak Shaving

This module implements a peak-shaving, valley-filling algorithm to cap system generation at a specified threshold value, defined as the "Post-Project Allowed Peak System Load (MW)," and referred to as target maximum for the remainder of this discussion. This is done through the use of an energy storage device with a defined round trip efficiency. Notably, the device capacity is not defined up front, but rather computed from the service provided (and output with the results). This algorithm starts with a full storage device, discharging when the load exceeds the target maximum, and charging during periods of low generation.

Inputs:

  • AVERT analysis region
  • Pre-Project Load Profile (hourly, in MW)
  • Pre-Project Maximum System Load (MW) (used for scaling)
  • Post-Project Allowed Peak System Load (MW)
  • Storage Device Round-Trip Efficiency
  • Line Losses

Outputs:

  • Changes to gas emissions by weight and percentage
  • Modified (peak-shaved and valley-filled) generation profile
  • Battery energy capacity (MWh)
  • Battery rated power (MW)

First of all, the Pre-Project Load Profile input is scaled to the specified Pre-Project Maximum System Load input via a process referred to as peak scaling. This involves creating a scaling factor that is equal to the Pre-Project Maximum System Load divided by the maximum of the Pre-Project Load Profile, and then multiplying every value of the Pre-Project Load Profile by this scaling factor.

The algorithm begins by examining a 24 hour period of the generation profile, and determining when the target maximum generation level is exceeded. If the target maximum is never exceeded, the next 24 hour period will be examined, and no peak-shaving or valley-filling occurs for that period. If the target maximum is exceeded, each hourly value above the target maximum will be reset to take on the value of the target maximum (peak-shaving). The difference between the original value and the target maximum is the energy that needs discharged for that hour. While the load is given in units of power (MW), it is easily converted to energy by multiplying by one hour. Since it is not known what the load does within the given hour, no other area approximations (trapezoidal or otherwise) are appropriate. This is a midpoint approximation. This energy is summed for every hour that exceeds the target maximum, and this sum represents the total energy discharged by the battery in the given 24 hour period.

The aforementioned 24 hour energy sum is then divided by the round trip efficiency value, resulting in total energy that must be generated in order to recharge the battery. Energy storage introduces several inefficiencies. Batteries are an illustrative example: Alternating current electricity from the grid must be converted to direct current to charge the battery, and back again to discharge. Additionally, batteries have internal resistance. The round-trip efficiency includes all three losses. Other forms of energy storage (such as pumped hydro storage), will also have losses in both the charging and discharging phase. Therefore, a round trip efficiency value is sufficient.

The round trip efficiency results in an amount of energy greater than the amount of energy that was peak-shaved. This final amount of energy now needs filled into the valleys in the generation profile, and is tracked to determine the necessary battery energy capacity (MWh). The interval considered eligible for valley-filling runs from the first instance of peak-shaving in the 24 hour period (first value equal to the target maximum), until whichever comes first: either the first instance of a value exceeding the target maximum outside of the 24 hour period, or 168 hours after the aforementioned first peak-shaving instance within the 24 hour period. For the purposes of this module, extra hours are added on to the end of the load profile to avoid a failure to charge/fence-post issue. If a default load is chosen, 168 hours of load data from 2015 will be tacked on. The user can either upload a file that is 8928 hours long, or a regular 8760 file. If an 8760 file is uploaded, the last 168 hours will be repeated and added on to the end. In all of the aforementioned cases, peak-shaving will be performed up until the last hour, but valley-filling can potentially happen all the way out to the last hour of the file. The data that has been added on will be used in calculating emission differences later on as well.

Bisection is used to determine the generation level at which battery recharge (valley-filling) should occur. After the eligible valley-filling interval is determined, the absolute minimum generation value within that interval is determined. The difference of the target maximum minus the aforementioned minimum value is divided by two, and used as a starting threshold for bisection. Every value in the eligible valley-filling interval is compared against this threshold: if the generation value is less than the threshold, that hour is considered eligible, and the difference between the threshold and the generation value is recorded. After every hour is checked and every difference recorded, these differences are summed up. This sum indicates how much charging would occur if the current threshold were used for battery charging. The error for this threshold is defined as the difference between the required charging energy and this sum (representing energy at current threshold). If the error is greater than 0.001 MW, bisection will repeat, moving the charging threshold up and down according to the sign of error.

If the threshold ever exceeds the target maximum, this indicates that it is not possible to fill the requisite energy into the valley-filling interval without exceeding the user-specified "Post-Project Allowed Peak System Load (MW)," and thus the charging operation has failed. This failure is reported to the user along with the suggestion to either increase the target maximum or round trip storage efficiency.

After the bisection loop converges, a final charging threshold has been determined. Every generation level in the valley-filling interval is examined, and if that level is less than the final threshold, it is reset to equal the final threshold (valley-filling).

This process of peak-shaving and valley-filling is repeated for every 24 hour period of the hourly load profile, resulting in a new "shaved and filled" load. The change in load also affects the line losses: The changes in losses are calculated as a constant factor (see the Additional Transmission and Distribution loss changes section for values and sources) times the difference between the baseline unmodified load and the peak-shaved, valley-filled load. The application of the line losses produces the final modified load - both it and the unmodified load are scaled up to AVERT region size via the aforementioned peak scaling factor and run through the AVERT engine. The difference in emissions is then scaled back to project size, and the scaled difference represents the estimated impact due to the energy storage project. Before scaling for use in AVERT, pre-project and post-project loads are subtracted from each other to determine the necessary battery power rating (MW).


Algorithm Flowchart
This flowchart illustrates the methodology used in the Energy Storage for Annual Peak Shaving module, discussed above.

Back to the Top

Photovoltaic Solar Generation

This module estimates the emission reductions for a load given the nameplate capacity and solar profile location of a given smart-grid enabled solar installation.

Inputs:

  • AVERT analysis region
  • Pre-Project Load Profile (hourly, in MW)
  • Pre-Project Maximum System Load (MW) (used for scaling)
  • Solar Profile Location
  • Solar PV DC Capacity (MW)
  • Line Losses

Outputs:

  • Changes to gas emissions by weight and percentage
  • Modified load profile (after solar project implementation)

First of all, the Pre-Project Load Profile input is scaled to the specified Pre-Project Maximum System Load input via a process referred to as peak scaling. This involves creating a scaling factor that is equal to the Pre-Project Maximum System Load divided by the maximum of the Pre-Project Load Profile, and then multiplying every value of the Pre-Project Load Profile by this scaling factor.

The file loaded based on the selection of the Solar Profile Location gives the hourly power output (Watts) of a 4 kW nameplate capacity panel in the given city. This comes from the National Renewable Energy Laboratory's (NREL) System Advisor Model (SAM) using typical meteorological year (TMY) data. See the section below for more information on SAM. Each hourly value is then scaled based on the user input Solar PV DC Capacity (MW) by dividing by 4000 and multiplying by the Solar PV DC Capacity. This results in a scaled hourly solar generation profile.

The modified load is simply the Pre-Project Load Profile minus the scaled hourly solar generation profile. The change in load also affects the line losses: The changes in losses are calculated as a constant factor (see the Additional Transmission and Distribution loss changes section for values and sources) times the difference between the baseline unmodified load and the modified load with solar included. The application of the line losses produces the final modified load - both it and the unmodified load are scaled up to AVERT region size via the aforementioned peak scaling factor and run through the AVERT engine. The difference in emissions is then scaled back to project size, and the scaled difference represents the estimated impact due to the photovoltaic project.


Algorithm Flowchart
This flowchart illustrates the methodology used in the Photovoltaic Solar Generation module, discussed above.

National Renewable Energy Laboratory's (NREL) System Advisor Model (SAM)

In order to determine photovoltaic output for a system based on location, NREL's SAM was used. Detailed information on SAM can be found here: https://sam.nrel.gov/. It is important to note that SAM is regularly updated, so values used by the EQT may differ slightly from the current version even with the same inputs.

For each city listed, the "Photovoltaic (PVWatts)" performance model was used, with the "Residential (distributed)" financial model. For each city listed in the EQT, a SAM simulation was run. The preferred data set used was typical meteorological year 2 (TMY2). No default modeling parameters were changed besides the selection of location. This included leaving in a 20 degree tilt and 180 degree Azimuth. The final output time series used for EQT calculations is the "AC inverter power." As noted in the module description above, this is scaled to proper project size.

To get a more accurate power output for your specific project, it is recommended that you download SAM and tune project specific parameters to optimize your project. Alternatively, NREL has a web-based version of PVWatts located here: http://pvwatts.nrel.gov/. Reminder: custom solar input files fed into the EQT need to be in Megawatts.

Back to the Top

Advanced Metering Infrastructure (AMI)

This module estimates CO2 reductions from the reduction of truck rolls for meter reading that comes from installing advanced metering infrastructure (AMI).

There are three possible input types that the user can provide:

  • By Fuel Consumption
  • By Miles per Gallon (MPG)
  • By Number of Customers

By Fuel Consumption:

This input type only takes in the annual fuel consumed due to meter reading, and uses the following estimation for CO2 per gallon: 8.887 * 10^-3 metric tons CO2 per gallon of gasoline. This comes from the Environmental Protection Agency (EPA): http://www.epa.gov/cleanenergy/energy-resources/refs.html.

The metric tons CO2 is then converted to US tons (short tons) by the following conversion factor: 1 metric ton = 1.10231131092 US tons. This conversion factor comes from Google's unit converter: https://support.google.com/websearch/answer/3284611?hl=en.

By Miles per Gallon (MPG)

This input type takes in the total annual meter reading miles traveled, and the average fuel economy (in miles per gallon) of the meter reading fleet. Total fuel consumption is determined by the number of miles divided by the average fleet fuel economy.

The following estimation for CO2 per gallon is used: 8.887 * 10^-3 metric tons CO2 per gallon of gasoline. This comes from the Environmental Protection Agency (EPA): http://www.epa.gov/cleanenergy/energy-resources/refs.html.

The metric tons CO2 is then converted to US tons (short tons) by the following conversion factor: 1 metric ton = 1.10231131092 US tons. This conversion factor comes from Google's unit converter: https://support.google.com/websearch/answer/3284611?hl=en.

By Number of Customers

This input type takes in the number of customers who have their meter read annually, the customers per mile (number of meters read for each mile driven), number of meter reading trips taken per year, and the average fuel economy of the meter reading fleet.

Total fuel consumption is determined by multiplying the number of customers by the number of trips per year, and dividing this value by the customers per mile times the average fleet fuel economy.

The following estimation for CO2 per gallon is used: 8.887 * 10^-3 metric tons CO2 per gallon of gasoline. This comes from the Environmental Protection Agency (EPA): http://www.epa.gov/cleanenergy/energy-resources/refs.html.

The metric tons CO2 is then converted to US tons (short tons) by the following conversion factor: 1 metric ton = 1.10231131092 US tons. This conversion factor comes from Google's unit converter: https://support.google.com/websearch/answer/3284611?hl=en.


Algorithm Flowchart
This flowchart illustrates the methodology used in the Advanced Metering Infrastructure module, discussed above.

Back to the Top

Conservation Voltage Reduction (CVR)

This module estimates the change in emissions by weight and percentage for a technique known as conservation voltage reduction (CVR), in which distribution feeder voltage is minimized to reduce end use voltage as much as possible while still staying within standards outlined in ANSI C84.1 (plus or minus 5% at the point of delivery). Conservation voltage reduction can be performed year-round, and that is what this module implements.

Inputs:

  • AVERT analysis region
  • Pre-Project Load Profile (hourly, in MW)
  • Pre-Project Maximum System Load (MW) (used for scaling)
  • CVR Region
  • CVR Load Reduction (%)
  • Line Losses (transmission only)

Outputs:

  • Changes to gas emissions by weight and percentage
  • Modified load profile (after CVR project implementation)

First of all, the Pre-Project Load Profile input is scaled to the specified Pre-Project Maximum System Load input via a process referred to as peak scaling. This involves creating a scaling factor that is equal to the Pre-Project Maximum System Load divided by the maximum of the Pre-Project Load Profile, and then multiplying every value of the Pre-Project Load Profile by this scaling factor.

The actual possible reductions that can be achieved vary hourly based on many factors, including environmental factors such as temperature. Hourly reduction values utilized by this module come from a study performed in GridLAB-D, which accounts for many factors when modeling what reductions are actually possible. Please see the section below for more information. The hourly reduction values are an aggregated average percentage based on many different feeders from the user specified CVR Region.

This hourly reduction is then scaled based on the user input CVR Load Reduction. The scaling factor is determined by dividing the user input CVR Load Reduction by the mean hourly CVR value from the specified CVR Region. Each hourly CVR value is then multiplied by this scaling factor.

To obtain the post CVR modified load, each hourly scaled unmodified load value is multiplied by the corresponding scaled CVR value, and this product is subtracted from the scaled unmodified load value.

The change in load also affects the line losses: The changes in losses are calculated as a constant factor (see the Additional Transmission and Distribution loss changes section for values and sources) times the difference between the baseline unmodified load and the modified load with CVR included. The GridLAB-D data already includes distribution line losses, thus only transmission line losses are selectable in this module. The application of the line losses produces the final modified load - both it and the unmodified load are scaled up to AVERT region size via the aforementioned peak scaling factor and run through the AVERT engine. The difference in emissions is then scaled back to project size, and the scaled difference represents the estimated impact due to the CVR project.


Algorithm Flowchart
This flowchart illustrates the methodology used in the Conservation Voltage Reduction module, discussed above.

GridLAB-D Study

GridLAB-D is a power distribution system simulation and analysis tool developed at Pacific Northwest National Laboratory. For detailed information, explore here: http://www.gridlabd.org/.

The following study provides the basis for the EQT's CVR analysis: http://www.pnl.gov/main/publications/external/technical_reports/PNNL-19596.pdf. Several taxonomy feeders for each CVR region included in the study were utilized for each climate region. This information was then aggregated to be utilized by the EQT. (consider revising last sentence)

For each region and taxonomy feeder within that region, 15 minute increment real power values (with and without CVR implemented) were averaged to create hourly real power values for each feeder by averaging the appropriate four 15 minute increment power levels. (revise previous sentence) Next, the hourly difference between feeder power levels with and without CVR was taken. These hourly differences were then converted to percent differences (can be positive or negative). Finally, the average of all taxonomy feeder hourly percent reductions within the region was taken. This average hourly reduction is what is used by the EQT in calculating reductions (described above).

Back to the Top

Smart-Charging

This module estimates the difference in emissions between charging electric vehicles (EVs) in an uncoordinated manner, or a coordinated manner. Coordinated utilizes a valley filling algorithm to shift charging load to periods of low demand.

Inputs:

  • AVERT analysis region
  • Pre-Project Load Profile (hourly, in MW)
  • Pre-Project Maximum System Load (MW) (used for scaling)
  • Uncoordinated Charging Profile
  • Number of EVs
  • Line Losses

Outputs:

  • Changes to gas emissions by weight and percentage
  • Uncoordinated load profile (after adding uncoordinated EV charging load)
  • Coordinated load profile (after coordinated EV charging project implementation)

First of all, the Pre-Project Load Profile input is scaled to the specified Pre-Project Maximum System Load input via a process referred to as peak scaling. This involves creating a scaling factor that is equal to the Pre-Project Maximum System Load divided by the maximum of the Pre-Project Load Profile, and then multiplying every value of the Pre-Project Load Profile by this scaling factor.

Next, an uncoordinated EV load is created based on the Uncoordinated Charging Profile selected by the user. This file contains aggregated average hourly charging rates per electric vehicle in kW, and is separated into weekday and weekend charging. These values come from the EV project, described below. These hourly values are repeated, concatenated, and scaled based on the Number of EVs input by the user to form an hourly uncoordinated EV only load. The overall uncoordinated load is simply the hourly sum of the Pre-Project Load Profile, and the uncoordinated EV only load.

Next, the coordinated charging needs defined. Eligible charge times are defined to be 5 p.m. until 8 a.m. The last day of the year simply allows charging over the course of the entire day to eliminate the fencepost issue. For each 24 hour period, the total EV charging energy required by the uncoordinated EV only load is determined by summing charge energy for that period. This energy is then filled into the lowest periods of demand between 5 p.m. and 8 a.m. (of the next day) via a bisection method. However, additional power levels added to the unmodified load from this charging energy cannot exceed the number of EVs times user input maximum power. If an hour's new load level exceeds this power, it is reduced to the maximum allowable power, and disallowed from future bisection. The difference in energy from the original bisection and the maximum allowable power is recorded, and bisection is then performed again, absent any hours that exceeded the maximum allowable power condition. This occurs until no values violate the maximum power.

After the coordinated charging profile has been determined, the change in load also affects the line losses: The changes in losses are calculated as a constant factor (see the Additional Transmission and Distribution loss changes section for values and sources) times the difference between the uncoordinated load and the coordinated load. The application of the line losses produces the final coordinated load - both it and the uncoordinated load are scaled up to AVERT region size via the aforementioned peak scaling factor and run through the AVERT engine. The difference in emissions is then scaled back to project size, and the scaled difference represents the estimated impact due to coordinated versus uncoordinated EV charging.


Algorithm Flowchart
This flowchart illustrates the methodology used in the Smart-Charging module, discussed above.

The EV Project

The EV Project collects and analyzes data to characterize electric vehicle use. More detailed information can be found here: http://energy.gov/eere/vehicles/avta-ev-project. The EQT leverages data from the EV Project to create uncoordinated electric vehicle charging profiles based on different charging schemes and incentives.

EV Project infrastructure data from 2012 (Quarter 3 and Quarter 4) and 2013 (Q1-Q4) were aggregated to create typical weekday and weekend charging days for every city involved in the EV Project. The EQT used home charging data only. The EV Project reports a "typical" weekday and weekend charging profile constructed from median charging values from the entire quarter. These median values come in 15 minute increments, and are combined to construct a typical day. To further aggregate the data, the EQT does the following: These quarterly median home charging loads are averaged into hourly loads (by combining the appropriate four 15 minute intervals), converted to kW, and divided by the number of participating electric vehicles to establish a per vehicle charging load. Next, the hourly per vehicle load from each quarter is averaged together to create an overall final hourly per vehicle load.

These final hourly per vehicle weekday and weekend loads are created for each city in the EV Project. City loads are further averaged together based upon similar charging profiles and grouped into four charging profiles: flat-rate, time of use (TOU) 1, TOU 2, and TOU 3 charging profiles seen in the EQT module. The city groupings are illustrated in the table below:

Charging Profile

Cities (or state)

Flat-Rate

Chattanooga

Chicago

Dallas/Ft. Worth

Houston

Knoxville

Memphis

Nashville

Philadelphia

Washington State (not a city)

Washington, D.C.

TOU 1

Los Angeles

Phoenix

TOU 2

Atlanta

Oregon (not a city)

Tucson

TOU 3

San Diego

San Francisco

Back to the Top

Demand Response (DR)

This module estimates the change in emissions due to the implementation of a cooling-season, air-conditioner-based demand response program. Demand response programs encourage reduction in energy consumption through increased energy prices during high-load periods.

Inputs:

  • AVERT analysis region
  • Pre-Project Load Profile (hourly, in MW)
  • Pre-Project Maximum System Load (MW)(used for scaling)
  • Number of Critical Peak Period (CPP) Days per Cooling Season
  • Peak Period Start Time
  • Peak Period Stop Time
  • First Cooling Month
  • Last Cooling Month
  • DR Program Participants with Central AC (%)
  • Temperature Profile Location
  • Flat Rate ($/kWh)
  • On-Peak Rate ($/kWh)
  • CPP Rate ($/kWh)
  • Demand Response Fade Factor

Outputs:

  • Changes to gas emissions by weight and percentage
  • Modified load profile (after DR project implementation)

The demand response program is implemented through the use of the Brattle Group’s PRISM model with fade and rebound/snapback effects added. This model implements a two-tier pricing structure with a limited number of CPP days during the cooling season. No “enabling technology” was implemented for this version of the PRISM model.

The realization of the PRISM model used in this analysis is based on a spreadsheet version found here: edisonfoundation.net/iei/Documents/PRISM_Suite.xls. The number of inputs required by this implementation is relatively large and to simplify the EQT user experience a few methods of estimating these inputs were developed.

To determine the pre-DR energy consumption by customers with and without central air-conditioning (CAC), offline GridLAB-D simulations (http://www.gridlabd.org) are run for a limited number of locations throughout the United States. Each simulation calculates the hourly energy used by residential customers for an entire year, grouped by CAC ownership. Note that this simulation does not implement the demand response program, it is simply used to calculate the energy consumption, taking into account the weather-induced energy requirements.

Using the timing inputs provided by the user for the demand response program, the energy consumed by customers with and without central air-condition during on-peak and off-peak periods of the cooling season days. To tabulate the energy consumed during CPP days, results from the GridLAB-D simulation are examined to find the days with the highest peak daily load. Based on the number of CPP days specified by the user, that specific number of days are classified as CPP and the energy used during those days is tabulated separately as specified in the PRISM spreadsheet.

In addition to the energy consumption, the PRISM model also requires an estimate of the amount of cooling required during the cooling season, expressed as average cooling degree-hours per hour on a baseline line of 72 ‘F. These values are calculated based on the user selected temperature profile (by default, the temperature profile associated with the location used in the GridLAB-D simulation but optionally a separate user input). With the user-provided definition of the cooling season and on-peak times, the average difference in on- and off-peak cooling requirements are calculated for CPP and non-CPP days. Similarly, the same is calculation is made for the daily average cooling requirement for both CPP and non-CPP days.

The last offline calculation made is for the off-peak rate. To ensure that the revenue derived by the utility under the demand response program is equal to that prior to the program being implemented, only three of the four rate inputs are free for the user to define: flat-rate (prior to the demand response program being implemented), CPP, and on-peak. The fourth, off-peak rate, is calculated to ensure the demand response program is revenue neutral.

Using these calculated values based on the GridLAB-D simulation, it is possible to complete the PRISM elasticity values calculations and find the percentage change in load during the on-peak, off-peak and CPP periods. These changes in load are directly applied to the user-defined Pre-Project Load Profile, reducing the load during the on-peak periods and increasing it during off-peak periods. Load values outside the cooling season are not adjusted.

The PRISM model assumes constant changes in load during on-peak periods but other research has shown the effects tend to diminish as the event lengthens due to re-equilibration of the cooling provided by the air-conditioner (at its higher set-point) and the outside air temperature. To estimate this behavior, a constant percentage reduction in effectiveness is added, the constant being defined by the “Demand Response Fade Factor” user input. Adding this effect creates an exponential decay or fading in the load reduction; if the on-peak period is long enough or the fade factor high enough, the load will return to its pre-event level before the end of the on-peak period.

When the on-peak period ends and air-conditioner set-points drop, a large increase in load is typically seen as most or all of the air-conditioners in the system simultaneously begin re-cooling their respective residences. This “rebound” or “snapback” effect is not a part of the PRISM model and has been added to the EQT. The amount of energy in the rebound event is defined to be equal to the amount of energy conserved during the on-peak period. This total energy is allocated in the two hours following the end of the on-peak period: 75% of it is added to the load in the first off-peak hour and the remaining 25% is added to the load in the second off-peak hour. This allocation is found based on results found in the following study: http://www.pnnl.gov/main/publications/external/technical_reports/PNNL-20772.pdf


Algorithm Flowchart
This flowchart illustrates the methodology used in the Demand Response module, discussed above.

Back to the Top

Custom

The custom module is by far the simplest module of the EQT, yet allows sweeping flexibility in using the underlying AVERT driven load to emissions mapping. A user simply inputs a pre and post smart grid project load profile, and NOx, SO2, and CO2 are reported.

Inputs:

  • AVERT analysis region
  • Pre-Project Load Profile (hourly, in MW)
  • Smart Grid Modified Load Profile (hourly, in MW)
  • Line losses

Outputs:

  • Changes to gas emissions by weight and percentage

The change in load between the Pre-Project Load Profile and Smart Grid Modified Load Profile affects the line losses: The changes in losses are calculated as a constant factor (see the Additional Transmission and Distribution loss changes section for values and sources) times the difference between the baseline pre project load and the post-project load. The application of the line losses produces the final modified load - both it and the unmodified (pre-project) are run through the AVERT engine, outputting criteria pollutant reductions.


Algorithm Flowchart
This flowchart illustrates the methodology used in the Custom module, discussed above.

Back to the Top

Future Modules

The EQT team recognizes that the current set of modules does not provide full coverage of available smart grid technology, or all specific use cases for technologies provided. Please scroll to the bottom of the Projects tab for a sampling of new modules we hope to add. If you have specific ideas that may be a good fit for the EQT framework, please contact us.

Back to the Top