This software was developed for analysis by U.S. Department of Transportation staff of potential fuel economy requirements.
This software uses input files containing detailed information regarding vehicles manufactured for sale in the United States and creates output files containing similarly detailed information regarding such vehicles. If input files containing information in any way (e.g., based on entitlement under 5 U.S.C 552 to confidential treatment) protected from disclosure to the public are used, some output files created by this software must also be protected from disclosure to the public.
The CAFE Model software is a U.S. government work not subject to copyright pursuant to 17 USC 105; however, some of the third-party works used by the software are subject to usage agreements, as described below.
The button controls in the application toolbar of the CAFE Model use images from the Glaze Icon Set (version 0.4.6, released on 3/06/2006) obtained from http://www.notmart.org. All icons and/or images within the Glaze Icon Set are distributed under the GNU Lesser General Public License (LGPL), version 2.1. The version 2.1 of the GNU LGPL may be obtained from http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html.
Some of the compiled or object code used by the CAFE Model was obtained from third-party sources. Specifically, the code for randomizing the forecast data for the average fuel prices, which is executed as part of the CAFE Model’s uncertainty analysis, makes use of the inverse of the beta cumulative probability density function contained within the Meta.Numerics library (in particular, the InverseLeftProbability method of the BetaDistribution class is used). The Meta.Numerics library (version 2.0.0, released on 4/06/2011) was obtained from http://www.meta-numerics.net and is distributed under the Microsoft Public License (Ms-PL). The latest version of the Ms-PL may be obtained from http://www.microsoft.com/en-us/openness/licenses.aspx.
If users of the CAFE model have any questions about this notice, please contact the current administrators of the CAFE Model project.
Installation and System Requirements
The CAFE Compliance and Effects Modeling System (abbreviated: CAFE Model) runs on IBM-compatible computers using the Microsoft® Windows operating system. Although the software does not have strict hardware requirements, beyond what is needed to run the operating system, a 1 GHz or faster Intel compatible processor, with at least 2 GB of physical memory (RAM) is strongly recommended. The software has been developed and tested on computers using Windows XP/7 and Windows Server 2003/2008, but may operate properly on machines using older versions of Windows (e.g., Windows 2000), or newer versions (e.g., Windows 8), as long as a compatible Microsoft® .NET Framework is installed.
The CAFE Model software uses Microsoft® Excel to read input files needed for modeling. As such, Excel must be installed on the system. The software also uses the Microsoft® .NET Framework, version 3.5. If the Framework is not already present, it must be installed. Instructions for downloading and installing the .NET framework are available on the Internet at http://www.microsoft.com/download/en/details.aspx?id=22.
Based on the characteristics of machines used in the development of this software, the following table provides a summary of system requirements:
Table 1. CAFE Model System Requirements
|
512 MB RAM (2 GB recommended) |
|
10 MB hard drive space for installation (additional disk space will be required during runtime) |
|
Microsoft® Windows XP/Vista/7 Microsoft® Windows Server 2003/2008 |
|
Microsoft® .NET Framework 3.5 |
|
Microsoft® Office 2003 or later |
Once the system requirements have been met, the latest version of the CAFE Model may be obtained by contacting NHTSA or Volpe Center staff.
The current version of the software is packaged in a way that does not require installation. To operate the model, place the “CAFE Model.exe” file on the desktop and execute it[1].
CAFE Model Graphical User Interface
The CAFE Model Graphical User Interface (GUI) provides users with a set of tools necessary to set up and run multiple modeling test scenarios, which are commonly referred to as CAFE Model sessions. Each CAFE Model session can be configured independently, each with its own set of model inputs and settings. Once configured, the session may be saved for future runs, or executed immediately.[2] When the model runs, the system displays the progress of the compliance modeling process in each session’s window.
The model GUI consists of two primary screens: the main CAFE Model window and the Modeling Settings window. The CAFE Model window is used for managing the modeling sessions, while the Modeling Settings window is used to configure them.
To run the modeling system, click on the CAFE Model executable file located on the desktop. When the application launches, a Warnings dialog box is displayed (Figure 1). The user must read and understand the warnings listed prior to using the modeling system.
Figure 1. Warnings Dialog Box
After clicking the OK button in the Warnings dialog box, the main CAFE Model window, described below, opens.
The main CAFE Model window (Figure 2) is used to create, configure, and manage CAFE modeling sessions. The main window also controls the model operation, allowing users to start and stop modeling simulation.
Figure 2. CAFE Model Window
When the model first starts up, most of the menu items and toolbar icons are disabled, until a new session is created, or an existing one is opened.
The model GUI is operated using a simple, easy to use file-menu (Figure 3), with most commonly used shortcuts also available on the model toolbar (Figure 4). For user convenience, most of the menu entries may also be controlled using keyboard shortcuts.
Figure 3. CAFE Model File Menu
Figure 4. CAFE Model Toolbar
Some of the most commonly used file menus are:
· File > New Session: Creates a new CAFE Model Session and displays the Modeling Settings window to the user.
· File > Open Session: Opens an existing CAFE Model Session.
· File > Close Session: Closes the active CAFE Model Session.
· File > Save Session: Saves the active CAFE Model Session.
· File > Start Modeling: Begins the modeling process for the active CAFE Model Session.
· File > Stop Modeling: Suspends the modeling process of the active CAFE Model Session.
· File > Exit: Exits the CAFE Model. If any of the modeling sessions are still opened, they will be closed prior to exiting the model.
· View > Modeling Settings: Displays the Modeling Settings window, where all modeling options and settings may be configured.
· View > Optimization Settings: Displays the Manage Optimization window, where additional options for Optimization modeling can be configured.
· View > Monte-Carlo Settings: Displays the Manage Monte-Carlo window, where additional options for Monte-Carlo modeling can be configured.
· View > Output Location: Opens the Windows Explorer and browses to the location where the output files and reports of the active session are saved.
The Modeling Settings window contains multiple panels for configuring all of the runtime options available to the model. The user can operate this window to set up a new session, or modifying an existing one, before starting the modeling process. Each of the available configuration panels is outlined in the sections below.
General Compliance Settings Panel
The General Compliance Settings panel (Figure 5) is used to specify what type of modeling the user would like to run. Each model is tailored to different type of analysis, using its own set of assumptions and configuration settings. Presently, four model types are available:
· Standard Compliance Model: The Standard Compliance Model is the default mode of operation for the CAFE modeling system. This model type is used to evaluate technology costs and benefits in response to the required CAFE standards defined in the modeling scenarios.
· Compliance Model with EIS: This model type is similar to the Standard Compliance Model, except additional analysis necessary for the Environmental Impact Statement is performed.
· Optimization Model: This model type should be used to perform analysis for optimizing the shape of the required CAFE standard.
· Monte-Carlo Model: The Monte-Carlo Model is a specialized CAFE modeling type, which is used for running customized Monte-Carlo simulations necessary for uncertainty analysis.
Figure 5. General Compliance Settings Panel
On the Input Settings panel (Figure 6), the user can select the input data files for use with the modeling system. To protect Confidential Business Information (CBI), some of the input files may be password protected. The system, therefore, provides an option for users to enter an input password prior to loading such files.
Figure 6. Input Settings Panel (1)
When selecting input files, the model will attempt to verify if an appropriate file was used. If incorrect file path is entered, an error message will be displayed (Figure 7).
Figure 7. Input Settings Panel (2)
The Output Settings panel (Figure 8) is used to configure the location where modeling results will be saved.
Figure 8. Output Settings Panel
The modeling system automatically generates the following eight output files (in CSV format) during runtime:
· Technology Utilization Report: Provides manufacturer-level and industry-wide technology application and penetration rates for each technology, model year, and scenario analyzed. The results are disaggregated by regulatory class, as well as combined over the entire fleet.
· Compliance Report: Provides manufacturer-level and industry-wide summary of compliance model results for each model year and scenario analyzed. The results are disaggregated by regulatory class, as well as combined over the entire fleet.
· Societal Effects Report: Provides industry-wide summary of energy and emissions effects for each model year and scenario analyzed. The results are disaggregated by regulatory class and fuel type, as well as combined across all fuels and over the entire fleet.
· Societal Costs Report: Provides industry-wide summary of consumer and social costs for each model year and scenario analyzed. The results are disaggregated by regulatory class and fuel type, as well as combined across all fuels and over the entire fleet.
· Annual Societal Effects Report: This output file is similar to the Societal Effects Report, except it further disaggregates the results by calendar year.
· Annual Societal Costs Report: This output file is similar to the Societal Costs Report, except it further disaggregates the results by calendar year.
· Vehicles Report: Provides a detailed view of the final state of each vehicle examined by the model, for each model year and scenario analyzed.
·
Optimization Report: Provides functional
coefficients and CAFE levels (required and achieved) for each iteration that
was evaluated during optimization. This output file also contains a brief
snapshot of industry-wide results, per iteration, that aided the model in
picking the optimum levels for each model year that was optimized.
Note: the Optimization Report is only generated when the Optimization
Model is run.
The Runtime Settings panel (Figure 9) provides additional modeling options to further customize the model behavior, beyond what is available in the input files:
· Operate in “Maximum Technology” mode: Specifies that the model should operate in “maximum technology” mode, where each manufacturer is assumed to be unwilling to pay CAFE fines, all vehicle refresh and redesign schedules are ignored, and all technologies are available for application immediately and without being subject to phase-in constraints.
· Allow Voluntary Overcompliance: Specifies that the model should continue to apply technologies after reaching compliance during a given model year, as long as the application of additional technologies is cost effective.
· Allow Credit Transfers and Carry Forward: Specifies whether the model should be able to transfer credits between fleets (PC and LT) within the same manufacturer and model year, and whether the model should be able to carry past credits forward for up to five years within the same fleet and manufacturer.
· Enable the Dynamic Fleet Share Model: Specifies whether the model should dynamically adjust the model year’s PC/LT fleet share based on achieved CAFE levels from the previous year, the PC share from the previous year, and the current year’s fuel prices.
· Merge the Fleet for Modeling: Specifies whether to merge the entire industry into a single large manufacturer before beginning the modeling process.
Some of the options loaded from a parameters input file may be overridden using the Runtime Settings panel as well. If an “override” option is checked off (not selected), a default value from the input file is used. If an override option is checked on (selected), that value will be used in place of what was loaded from the parameters file. In Figure 9 below, the options for overriding the rebound effect and the discount rate are selected, and set to 20% and 7% respectively.
The following options from the parameters file may be overridden:
· Override Fuel Price Estimates: Specifies whether to use the low, average, or high fuel price estimates from the parameters input file. By default, average fuel price estimates are used.
· Override CO2 Estimates: Specifies whether to use low, average, high, or very-high carbon dioxide cost estimates from the parameters input file. By default, average CO2 cost estimates are used.
· Override Rebound Effect: Overrides the Rebound Effect value read in from the parameters file with a user defined value. Valid values are between -1.00 and 1.00.
· Override Discount Rate: Overrides the Discount Rate value read in from the parameters file with a user defined value. Valid values are between 0.00 and 1.00.
· Override Value of Travel Time per Vehicle: Overrides the Value of Travel Time per Vehicle value read in from the parameters file with a user defined value.
· Override Military Security Cost: Overrides the Military Security component of economic costs read in from the parameters file with a user defined value.
· Scale Consumer Benefits During Effects Calculations: Specifies whether the model should scale the private consumer benefits by a specific percentage during the effects calculations. Valid values are between 0.00 and 1.00.
Figure 9. Parameters Overrides Panel
The Manage Optimization window (Figure 10) provides additional options necessary for configuring the system for optimization modeling.
The first set of options determines the type of optimization – that is, which fleet the model should optimize:
· Cars: Forces the modeling system to optimize vehicles regulated as passenger automobiles only. If the market data input file contains any vehicles regulated as light trucks, the value of CAFE standard for those vehicles will be kept at a constant rate throughout optimization.
· Trucks: Forces the modeling system to optimize vehicles regulated as light trucks only. If the market data input file contains any vehicles regulated as passenger automobiles, the value of CAFE standard for those vehicles will be kept at a constant rate throughout optimization.
· Auto-detect: Allows the model to automatically determine whether to optimize passenger automobiles or light trucks. This option is useful if the market data input file contains only one class of vehicles (e.g., cars-only or trucks-only). If the market data file includes a mixed fleet of vehicles (passenger autos and light trucks), this option should not be used.
The next set of options determines the optimization mode the model should use when identifying the optimum value of the CAFE standard:
· Optimize based on maximum Net Benefits: Specifies that the optimization model should optimize the value of CAFE standard based on the difference between the discounted social benefits and technology costs, by maximizing that difference.
· Optimize by minimizing Net Benefits, after the maximum has occurred: Specifies that the optimization model should optimize the value of CAFE standard based on the difference between the discounted social benefits and technology costs, by finding the lowest positive difference after the maximum difference has occurred.
Additional optimization options are:
· Iterations above optimum: Indicates the number of iterations to examine above the initially calibrated shape of the target function, by moving the function upwards in GPM space. Raising the function produces a less stringent value of CAFE standard. Valid values are between 0 and 1000.
· Iterations below optimum: Indicates the number of iterations to examine below the initially calibrated shape of the target function, by pushing the function downwards in GPM space. Lowering the function produces a more stringent value of CAFE standard. Valid values are between 0 and 1000.
· Increment by: Specifies the value by which to increment the target function in GPM space. Valid values are between 0.00001 and 0.1.
· Begin optimizing starting with the specified year: Specifies the first model year to optimize.
Figure 10. Manage Optimization Window
The Manage Monte-Carlo window (Figure 11) provides additional options necessary for configuring the system for Monte-Carlo modeling. During modeling, the system will use an input file specified here, containing trials to use for analysis. Upon completion of the modeling process, the system will automatically generate Monte-Carlo log files.
Figure 11. Monte-Carlo Model Settings Panel
This section provides examples for configuring and running the CAFE Model sessions using various model types.
Example 1 – Configuring for Standard Compliance Modeling
This example demonstrates the steps necessary for configuring the modeling system to perform a regular Compliance Model run.
· Run the CAFE Model by clicking on the CAFE Model executable located on the desktop. Read through the Warnings dialog box, and then click the OK button.
· Select File > New Session to create a new modeling session. The Modeling Settings window appears. Note the errors at the bottom of the window; these indicate that the input files have not been selected.
· On the General Compliance Settings panel, select the Standard Compliance Model as shown in Figure 12 below.
Figure 12. Select Standard Compliance Model
· Click on the Input Settings panel to select the input files to use for modeling (Figure 13). Note that once all the input files have been selected appropriately, the error messages disappear.
Figure 13. Select Input Files
· On the Output Settings panel, select the location for output files (Figure 14).
Figure 14. Select Output Location
· The Runtime Settings panel is not used for this exercise.
· Click the Save button to save the modeling settings and load the input files (Figure 15).
Figure 15. Save Modeling Settings
· Once loading completes, click the Close button to return the main CAFE Model window. A new Compliance Model session, titled “Session 1” has now been created (Figure 16).
Figure 16. New Compliance Model Session Created
· Save the new session by selecting File > Save Session As.... Enter “demo.cmsd” in the dialog box that appears, and click the Save button (Figure 17).
Figure 17. Save New Session
· After the session has been saved, notice the title of the session has changed to “demo” (Figure 18).
Figure 18. “demo” Session Saved
· Select File > Start Modeling to start the compliance modeling process. As the model runs, the progress of the Compliance Model is displayed in the session window (Figure 19).
Figure 19. Modeling Progress from the Compliance Model
· After modeling has completed, the “Modeling Completed!” message appears at the bottom of the main CAFE Model window (Figure 20).
Figure 20. Compliance Model Completed
· Select View > Output Location to open Windows Explorer and browse to the location where model outputs for the “demo” session are saved.
· Exit the session by selecting File > Close Session.
· Exit the CAFE Model by selecting File > Exit, or proceed to the next example.
Example 2 – Configuring for Optimization Modeling
This example demonstrates how the take an existing session created in Example 1 – Configuring for Standard Compliance Modeling, and modify it to run the Optimization Model.
· Run the CAFE Model by clicking on the CAFE Model executable located on the desktop. Read through the Warnings dialog box, and then click the OK button.
· Select File > Open Session to open an existing modeling session. Select “demo.cmsd” in the dialog box that appears, and click the Open button (Figure 21).
Figure 21. Open “demo” Session
· Once the session has been loaded, select View > Modeling Settings to bring up the Modeling Settings window. There select the Optimization Model as in Figure 22.
Figure 22. Select Optimization Model
· Under the Input Settings panel, select a market data file containing data for the light truck fleet only, as well as a scenarios file required for optimization modeling (Figure 23).
Figure 23. Select Scenarios File for Optimization
· The Output Settings and Runtime Settings panels are not used for this exercise.
· Click the Save button to save the updated modeling settings; then click Close, once saving completes.
· Select View > Optimization Settings to bring up the Manage Optimization window, then configure the system for optimization modeling as specified in Figure 24. (**Note: with this version of the model, the system has been modified from using linear/additive stringency increments to multiplicative stringency increments. Hence, setting the increment incorrectly may lead to undesired behavior.)
Figure 24. Configure Optimization Model Settings
· Click the Save button to save the Optimization Model settings; then click Close.
· To prevent overwriting results from the “demo” session, select File > Save Session As... to save the modified session with a new name. For this example, the optimization session was saved as “demo-opt.cmsd” (Figure 25).
Figure 25. Save Modified Session
· Select File > Start Modeling to start the optimization modeling process. As the model runs, the progress of the Optimization Model is displayed in the session window (Figure 26).
Figure 26. Modeling Progress from the Optimization Model
· After optimization modeling has completed, the “Modeling Completed!” message appears at the bottom of the main CAFE Model window. Select File > Exit to exit the model.