Use of IM & NEI Data Trends to Improve Vehicle Emissions Models IM Solutions Salt Lake City April 2014 Jim Lindner
th CRC Workshop Presentations Preliminary Analysis on Long Term Deterioration of Tier 2 Vehicles –EPA/OTAQ- Carl Fulper, James Warila, Connie Hart –ERG- Sandeep Kishan, Meredith Weatherby, Michael Sabisch, Tim DeFries, Cindy Palacios –CDPHE- Jim Sidebottom, Jim Kemper Four State OBD Evap Analysis –EPA/OTAQ- Carl Fulper, Connie Hart, Glenn Passavant, Dave Hawkins –ERG- Sandeep Kishan, Meredith Weatherby, Michael Sabisch, Tim DeFries, Cindy Palacios
Overview GA Data Trends –2007 through 2012 annual report data Evaluation & Sensitivity Analysis of MOVES Input Data –2014 TRB Annual Meeting January 13, 2014
GA Data Trends Multiple years of data ( ) on MY1996+ vehicles –Identify vehicles with both a high failure rate and large number of total inspections as these should be driving the overall OBD failure rate –Individual vehicles that fail in consecutive years would provide information regarding the prevalence of onroad MIL-on
Top 25 M/M/MY Groups with Highest Number of Fails Highest number of inspections used rather than highest failure rate CY –Capture most frequent failures –And also the more common vehicles in the fleet –Initial test only for this filter Other –Track individual vehicles using all tests –Track specific DTCs for M/M/MY groups
Calendar Year 2003 MakeModelMY# Insp # Fail % Fail Rank for # Fail Rank for # Insp FORDEXPLORER4-DR %111 FORDEXPLORER %28 HONDAACCORD %32 FORDTAURUS %436 FORDTAURUS %531 TOYOTACAMRY %613 FORD F150SUPERCABSH ORT %729 HONDAACCORD %86 FORDEXPLORER %95 FORDWINDSTAR %10100
CY03 Top 10- % Fail with # Inspected Rank for # Fail % Fail
Calendar Year 2008 MakeModelMY# Insp # Fail % Fail Rank for # Fail Rank for # Insp FORDEXPLORER %130 FORDEXPLORER4DR %241 HONDAACCORD %32 FORDEXPLORER %417 HONDAACCORD %516 HONDAACCORD %610 HONDAACCORD %71 FORDEXPEDITION %831 TOYOTACAMRY %928 NISSANMAXIMA %10100
CY08 Top 10- % Fail with # Inspected Rank for # Fail % Fail
Calendar Year 2012 MakeModelMY# Insp # Fail % Fail Rank for # Fail Rank for # Insp HONDAACCORD %111 HONDAACCORD %27 HONDAACCORD %332 TOYOTACAMRY %441 HONDAACCORD %540 FORDEXPLORER %650 TOYOTACAMRY %713 FORDEXPLORER %8109 NISSANMAXIMA %975 FORDEXPEDITION %10100
CY12 Top 10- % Fail with # Inspected Rank for # Fail % Fail
Evaluation & Sensitivity Analysis of MOVES Input Data NEI is compiled by EPA every 3 years, covering major pollutants for all sectors and U.S. counties Draft 2011 estimates recently released 2011 is the first NEI relying solely on MOVES for onroad (outside CA) States given option to submit complete emissions, or MOVES County Database (CDB) inputs States are not required to follow EPA’s SIP/Conformity guidance for NEI
Study Objectives CRC A-84: “Study of MOVES Information for the National Emissions Inventory” –Complete CRC Report: Task 1: Evaluate state-submitted data –Task 1a: How do methods states used to gather data compare with EPA best practice? –Task 1b: What is the range of data submitted, and how do they compare to MOVES defaults? Task 2: MOVES sensitivity based on range of state data Task 3: Recommendations for improvement
MOVES CDBs Analyzed (Dark blue = submitted) Texas CDBs provided by TCEQ 30 states provided data in 1 st round (~1,400 counties)
What is Submitted? MOVES County Data Manager County Data Manager (CDM) allows custom input for following parameters, through MS Excel tables Data entered through CDM“Best Practice” Sources Vehicle Miles TravelledHPMS, Travel models Temperature & HumidityMeteorology data Vehicle PopulationRegistration data, fleets Average Speed DistributionTravel models Vehicle Age DistributionRegistration data, fleets Fuel Properties/Market SharesFuel surveys, fuel regulations Road Type DistributionHPMS, Travel models (VMT source) Fuel Technology MixRegistration data, fleets I/M Compliance/Waiver RatesOperating program data & history CDM interface 15
Analysis of Submitted Data Focused on 5 primary inputs project to have the largest impact on annual emissions –VMT, Vehicle Population, Age Distribution, Average Speed, Road Type Distribution Following Examples –Age Distribution –Speed Distribution
Age Distribution Example Distribution of submitted data for Passenger Cars
Speed Distribution: Clusters Cluster analysis of EPA’s Tom Tom data found 13 unique groupings of average speed distribution across roadtype, hour and weekend/weekday MOVES default clusters were similar.
Sensitivity Analysis Evaluate sensitivity to changes in 5 inputs, while holding other inputs constant –Average Speed & Road Type analyzed in conjunction Vary inputs based on spread of data submitted by states in first round of 2011 NEI –10 th percentile inputs –Median –90 th percentile inputs –Inputs chosen from submitted county databases Assess impact on total daily emissions of HC, CO, NOx and PM –Typical July day, Montgomery County, TX
Comparison Across Inputs Change in emissions from 10 th 90 th percentile inputs All source types / clusters
Most influential inputs by source type/cluster >30% Increase >20% Increase
Recommended Improvements NEI Process –More outreach to states that did not provide data, or provided mostly MOVES defaults –Template for documentation –Sharing of best practices between states MOVES Inputs –Assist in developing data sources for states –Mine national databases of local data e.g., vehicle registration databases are compiled nationally –Develop capacity with emerging data sources e.g. telematics datasets for speed, trip data
Acknowledgments GA Repair Data Trends –GA EPD- Pam Earle, Tim Smith, Steve Leydon Evaluation & Sensitivity Analysis of MOVES Input Data –ERG- John Koupal, Timothy DeFries, Cindy Palacios, Scott Fincher, Diane Preusse