Potential Energy Impacts of Connected and Automated Vehicles ...
Potential Energy Impacts of Connected and Automated Vehicles: Opportunities and Approaches ANL: Tom Stephens, Joshua Auld NREL: Jeff Gonder, Yuche Chen ORNL: Zhenhong Lin, Changzheng Liu, Jan-Mou Li UIC: Kouros Mohammadian, Ramin Shabanpour
Automated Vehicle Symposium Breakout Session Presentation July 20, 2016 UIC = University of Illinois at Chicago
1
On-Going Collaborative Research Project This short presentation: • High-level scenario assessment based on literature inputs and rough assumptions • Methodology summary for more detailed and comprehensive analysis (on-going work)
Assessment Structure: Main factors Value to Consumers
Change in Demand
• Generalized cost of personal travel • Mobility for underserved
• Travel by general public • Travel by underserved • Sharing • Mode shifts
Change in Efficiency • Faster/smoother travel • Lower loads • Smaller vehicles • Platooning
Energy
3
Scenarios Description Scenario: Automation Level Vehicle Ownership Ridesharing Efficiency Improvement VMT Demand Impact* CAV Incremental Cost**
AutoTaxi- AutoTaxiConvPartialPartialFullFullAutoTaxi- AutoTaxiRideshare- RidesharePrivate Private-UB Private-LB Private-UB Private-LB UB LB UB LB N/A
Partial
Full
Full
Full
Private
Private
Private
Shared
Shared
No
No
No
No
Yes
N/A
Low
High
Low
High
Low
High
Low
High
N/A
High
Low
High
Low
High
Low
High
Low
N/A
Low
High
Low
High
Low
High
Low
High
*Includes travel time costs (Low time cost leads to high VMT and thus higher energy use) **Includes vehicle purchase cost
UB: Upper bound of energy impact (higher energy use) LB: Lower bound of energy impact (lower energy use) “Partial” = NHTSA Level 1-2; “Full” = NHTSA Level 3-4 Auto taxi = Fully automated vehicle providing transportation as a service
4
Travel Demand May Increase Significantly with Full Automation Total VMT (Trillion) Total PMT (Trillion) Average Occupancy
Total VMT (trillion mi)
10 8 6 4 2 0 -2
2.8 4.65 1.67
3.1 5.26 1.67
2.9 9.5 4.84 14.27 1.67 1.50
3.6 5.67 1.59
9.5 3.6 14.27 5.67 1.50 1.59
9.0 3.1 14.27 5.67 1.59 1.85
Shift from transit, air Repositioning (empty travel) Underserved Faster Travel, Decr travel time cost Ridesharing Base VMT
5
Visualizing Average Consumption Rate/100 mi (and Reduction Source) Total VMT (Trillion) Total PMT (Trillion) Average Occupancy
2.8 4.65 1.67
3.1 5.26 1.67
2.9 4.84 1.67
9.5 14.27 1.50
3.6 5.67 1.59
9.5 14.27 1.50
3.6 5.67 1.59
9.0 3.1 14.27 5.67 1.59 1.85
Average Vehicle's Fuel Consumption per 100 Miles driving (Gallon per 100 miles)
5.0 4.0 3.0 2.0 1.0 0.0 -1.0 -2.0 -3.0 -4.0
6
Total US LDV Fuel Consumption per Year Total VMT (Trillion) Total PMT (Trillion) Average Occupancy
2.8 4.65 1.67
3.1 5.26 1.67
2.9 4.84 1.67
9.5 14.27 1.50
3.6 5.67 1.59
9.5 14.27 1.50
3.6 5.67 1.59
9.0 14.27 1.59
3.1 5.67 1.85
Total US LDV Fuel Consumption (Billion Gallons per Year)
350 300 250 200 150 100 50 0
7
On-Going Collaborative Research Project This short presentation: • High-level scenario assessment based on literature inputs and rough assumptions • Methodology summary for more detailed and comprehensive analysis (on-going work)
Analysis framework: Conceptual calculation flows Establish scenarios of different powertrain and CAV feature adoption rates over time
Quantify potential efficiency and travel behavior impacts of CAV features in different driving situations
2015
Calculate national-level energy use and GHGs from aggregating driving situation impacts by the proportion of national VMT each represents
Transfer regional analyses to estimate national evolution of VMT distributions over time
2040
Aggregate petroleum and GHG impacts of scenarios
9
Estimating CAVs Adoption: Adapt consumer choice model to include CAVs purchase decision • Quantify utility to consumers within different market segments and resulting impacts on ownership and operation decisions • Utility components: stress, energy, time, mobility, productivity • Revise ORNL’s MA3T choice structure to include CAVs – In addition to buy/no-buy a new vehicle, add the options of buying a CAV and using AutoTaxis
Buy Non-CAV Powertrain choice
No Buy CAV
Buy used vehicle Use other modes walk/bike transit shared (taxi, Uber)
Powertrain choice 10
Estimating Impacts on Travel Demand: Use transferability modeling to expand detailed travel simulation results to the national level • Transfer results from transportation system simulations of CAVs in a metropolitan area
11
Aggregate energy/GHG impacts of CAV features nationally:
Quantify different CAV feature fuel economy impacts in different driving situations
Consider the relative proportion of national VMT represented by each driving situation
Calculate national total energy use and GHG emissions by summing VMT for the entire U.S. road network
Preliminary Conclusions • Potential energy and GHG emission impacts from CAVs are uncertain • Previous work is difficult to synthesize into consistent scenarios • On-going DOE-supported simulation and data collection efforts will supply detailed inputs into the presented national-level assessment methodology
Planned Future Work • Further develop and validate expansion aggregation methods and apply these to detailed simulation/data collection results • Estimate potential adoption of CAVs technologies by different population segments • Consider multiple scenarios for analysis – Driverless taxis, with/without ridesharing – Connected vehicles in an urban environment (traffic smoothing) – Connected vehicles on highways (CACC, platooning) 13
Back up slides
14
Details of Initial High-Level Scenario Energy Calculations • Disaggregate the fuel consumption into road type (highway/urban), level of service (congestion/noncongestion). • Search literature to quantify each CAV feature’s energy impact on specific road type/level of service condition under different scenarios. • Re-calculate CAVs’ specific energy impact on applicable road type (urban or highway) and condition (congested or non-congested) into aggregated national average energy impact on overall driving • Estimate the total fuel consumption for average LDV per year with breakdown of CAV features’ impacts.
Assumptions (based on GPRA BaSce) • Average LDV annual mileage, 13350 mile/year • Average LDV fuel economy, 26.9 mi/gal • Gasoline price: $2.93 2010$ / gal VMT % and fuel economy by road type and level of service Highway Congested Highway Non-congested Urban Congested Urban Non-congested
VMT % 18% 27% 22% 33%
MPG 29.7 35.0 21.4 25.2
Based on assumptions from EPA MOVES model national default inputs and EPA fuel economy tag inputs. 16
Platooning
This estimate is much higher than other references. It assumed much smaller gaps between vehicles ( 0.1 to 0.3 vehicle lengths). We assume CAV Lev 1-2 can only achieve half of this benefit, and CAV Lev 3-4 can achieve up to 25% fuel savings.
-30%
-25%
-20%
Assuming platooning benefits occur only on non-congested highway driving, i.e. 27% of total VMT.
-15% Energy Impact %
Ref. 31, NREL, 2014 Ref. 29, USC, 1995 Ref. 28, UC Berkeley, 2012 Ref. 27, Great Britain, 2012 Ref. 26, Sweden, 2010 Ref. 6, MIT Tech. Review, 2011 -10%
-5%
Highway NonCongested
Full Automation CAV Partial Automation CAV
-30%
-25% Full Automation CAV Partial Automation CAV
-30%
-25%
0%
-20%
-15%
Calculate all road type equivalent energy impact by using VMT% and MPG data on slide 24
-20%
-15%
-10%
-5%
0%
All Road Types
-10%
-5%
0% 17
Drive Profile & Traffic Flow Smoothing Ref. 17, UC Riverside, 2013 Ref. 16, VTTI, 2012 Ref. 15, VTTI, 2012 Ref. 9, US RAND, 2014 Ref. 5, SUNY, 2012 Ref. 1, UC Riverside, 2009 -25%
-20%
-15% -10% Energy Impact %
-5%
0%
-10%
-5%
0%
-10%
-5%
0%
Urban Non-Congested Highway Non-Congested
Urban Congested Highway Congested
-25%
-20%
-15%
Full Automation CAV All Road Types Partial Automation CAV
-25%
-20%
-15%
Drive Profile & Traffic Less Intersectio Vehicle Flow Collision Hunting n V2I/I2V Resizing Smoothing Avoid. Platooning Fast Travel for Parking Comm.
Vehicle-Level Energy per Mile Impact Ranges of CAV Features Full Automation CAV Partial Automation CAV
-50% -25%
-20%
-15%
Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV
-10% -5% Energy Impacts %
0%
5%
10% 19
Explanation of Bar Chart Format for Presenting CAV Features’ Energy and Demand Impacts: • Reductions: for visualization, the reduction from the original attributed to each feature moves from above to below the x-axis. • Increments: add on top of top of the original bar. • The final height of the bar (in the positive region only) shows the net fuel consumption including all impacts 5
Average LDV Fuel Consumption (Gallons per 100 mile)
4 3 2 1 0
Original
Reduction
Increment
-1 -2
20
Automated Vehicle Symposium Breakout Session Presentation July 20, 2016 UIC = University of Illinois at Chicago
1
On-Going Collaborative Research Project This short presentation: • High-level scenario assessment based on literature inputs and rough assumptions • Methodology summary for more detailed and comprehensive analysis (on-going work)
Assessment Structure: Main factors Value to Consumers
Change in Demand
• Generalized cost of personal travel • Mobility for underserved
• Travel by general public • Travel by underserved • Sharing • Mode shifts
Change in Efficiency • Faster/smoother travel • Lower loads • Smaller vehicles • Platooning
Energy
3
Scenarios Description Scenario: Automation Level Vehicle Ownership Ridesharing Efficiency Improvement VMT Demand Impact* CAV Incremental Cost**
AutoTaxi- AutoTaxiConvPartialPartialFullFullAutoTaxi- AutoTaxiRideshare- RidesharePrivate Private-UB Private-LB Private-UB Private-LB UB LB UB LB N/A
Partial
Full
Full
Full
Private
Private
Private
Shared
Shared
No
No
No
No
Yes
N/A
Low
High
Low
High
Low
High
Low
High
N/A
High
Low
High
Low
High
Low
High
Low
N/A
Low
High
Low
High
Low
High
Low
High
*Includes travel time costs (Low time cost leads to high VMT and thus higher energy use) **Includes vehicle purchase cost
UB: Upper bound of energy impact (higher energy use) LB: Lower bound of energy impact (lower energy use) “Partial” = NHTSA Level 1-2; “Full” = NHTSA Level 3-4 Auto taxi = Fully automated vehicle providing transportation as a service
4
Travel Demand May Increase Significantly with Full Automation Total VMT (Trillion) Total PMT (Trillion) Average Occupancy
Total VMT (trillion mi)
10 8 6 4 2 0 -2
2.8 4.65 1.67
3.1 5.26 1.67
2.9 9.5 4.84 14.27 1.67 1.50
3.6 5.67 1.59
9.5 3.6 14.27 5.67 1.50 1.59
9.0 3.1 14.27 5.67 1.59 1.85
Shift from transit, air Repositioning (empty travel) Underserved Faster Travel, Decr travel time cost Ridesharing Base VMT
5
Visualizing Average Consumption Rate/100 mi (and Reduction Source) Total VMT (Trillion) Total PMT (Trillion) Average Occupancy
2.8 4.65 1.67
3.1 5.26 1.67
2.9 4.84 1.67
9.5 14.27 1.50
3.6 5.67 1.59
9.5 14.27 1.50
3.6 5.67 1.59
9.0 3.1 14.27 5.67 1.59 1.85
Average Vehicle's Fuel Consumption per 100 Miles driving (Gallon per 100 miles)
5.0 4.0 3.0 2.0 1.0 0.0 -1.0 -2.0 -3.0 -4.0
6
Total US LDV Fuel Consumption per Year Total VMT (Trillion) Total PMT (Trillion) Average Occupancy
2.8 4.65 1.67
3.1 5.26 1.67
2.9 4.84 1.67
9.5 14.27 1.50
3.6 5.67 1.59
9.5 14.27 1.50
3.6 5.67 1.59
9.0 14.27 1.59
3.1 5.67 1.85
Total US LDV Fuel Consumption (Billion Gallons per Year)
350 300 250 200 150 100 50 0
7
On-Going Collaborative Research Project This short presentation: • High-level scenario assessment based on literature inputs and rough assumptions • Methodology summary for more detailed and comprehensive analysis (on-going work)
Analysis framework: Conceptual calculation flows Establish scenarios of different powertrain and CAV feature adoption rates over time
Quantify potential efficiency and travel behavior impacts of CAV features in different driving situations
2015
Calculate national-level energy use and GHGs from aggregating driving situation impacts by the proportion of national VMT each represents
Transfer regional analyses to estimate national evolution of VMT distributions over time
2040
Aggregate petroleum and GHG impacts of scenarios
9
Estimating CAVs Adoption: Adapt consumer choice model to include CAVs purchase decision • Quantify utility to consumers within different market segments and resulting impacts on ownership and operation decisions • Utility components: stress, energy, time, mobility, productivity • Revise ORNL’s MA3T choice structure to include CAVs – In addition to buy/no-buy a new vehicle, add the options of buying a CAV and using AutoTaxis
Buy Non-CAV Powertrain choice
No Buy CAV
Buy used vehicle Use other modes walk/bike transit shared (taxi, Uber)
Powertrain choice 10
Estimating Impacts on Travel Demand: Use transferability modeling to expand detailed travel simulation results to the national level • Transfer results from transportation system simulations of CAVs in a metropolitan area
11
Aggregate energy/GHG impacts of CAV features nationally:
Quantify different CAV feature fuel economy impacts in different driving situations
Consider the relative proportion of national VMT represented by each driving situation
Calculate national total energy use and GHG emissions by summing VMT for the entire U.S. road network
Preliminary Conclusions • Potential energy and GHG emission impacts from CAVs are uncertain • Previous work is difficult to synthesize into consistent scenarios • On-going DOE-supported simulation and data collection efforts will supply detailed inputs into the presented national-level assessment methodology
Planned Future Work • Further develop and validate expansion aggregation methods and apply these to detailed simulation/data collection results • Estimate potential adoption of CAVs technologies by different population segments • Consider multiple scenarios for analysis – Driverless taxis, with/without ridesharing – Connected vehicles in an urban environment (traffic smoothing) – Connected vehicles on highways (CACC, platooning) 13
Back up slides
14
Details of Initial High-Level Scenario Energy Calculations • Disaggregate the fuel consumption into road type (highway/urban), level of service (congestion/noncongestion). • Search literature to quantify each CAV feature’s energy impact on specific road type/level of service condition under different scenarios. • Re-calculate CAVs’ specific energy impact on applicable road type (urban or highway) and condition (congested or non-congested) into aggregated national average energy impact on overall driving • Estimate the total fuel consumption for average LDV per year with breakdown of CAV features’ impacts.
Assumptions (based on GPRA BaSce) • Average LDV annual mileage, 13350 mile/year • Average LDV fuel economy, 26.9 mi/gal • Gasoline price: $2.93 2010$ / gal VMT % and fuel economy by road type and level of service Highway Congested Highway Non-congested Urban Congested Urban Non-congested
VMT % 18% 27% 22% 33%
MPG 29.7 35.0 21.4 25.2
Based on assumptions from EPA MOVES model national default inputs and EPA fuel economy tag inputs. 16
Platooning
This estimate is much higher than other references. It assumed much smaller gaps between vehicles ( 0.1 to 0.3 vehicle lengths). We assume CAV Lev 1-2 can only achieve half of this benefit, and CAV Lev 3-4 can achieve up to 25% fuel savings.
-30%
-25%
-20%
Assuming platooning benefits occur only on non-congested highway driving, i.e. 27% of total VMT.
-15% Energy Impact %
Ref. 31, NREL, 2014 Ref. 29, USC, 1995 Ref. 28, UC Berkeley, 2012 Ref. 27, Great Britain, 2012 Ref. 26, Sweden, 2010 Ref. 6, MIT Tech. Review, 2011 -10%
-5%
Highway NonCongested
Full Automation CAV Partial Automation CAV
-30%
-25% Full Automation CAV Partial Automation CAV
-30%
-25%
0%
-20%
-15%
Calculate all road type equivalent energy impact by using VMT% and MPG data on slide 24
-20%
-15%
-10%
-5%
0%
All Road Types
-10%
-5%
0% 17
Drive Profile & Traffic Flow Smoothing Ref. 17, UC Riverside, 2013 Ref. 16, VTTI, 2012 Ref. 15, VTTI, 2012 Ref. 9, US RAND, 2014 Ref. 5, SUNY, 2012 Ref. 1, UC Riverside, 2009 -25%
-20%
-15% -10% Energy Impact %
-5%
0%
-10%
-5%
0%
-10%
-5%
0%
Urban Non-Congested Highway Non-Congested
Urban Congested Highway Congested
-25%
-20%
-15%
Full Automation CAV All Road Types Partial Automation CAV
-25%
-20%
-15%
Drive Profile & Traffic Less Intersectio Vehicle Flow Collision Hunting n V2I/I2V Resizing Smoothing Avoid. Platooning Fast Travel for Parking Comm.
Vehicle-Level Energy per Mile Impact Ranges of CAV Features Full Automation CAV Partial Automation CAV
-50% -25%
-20%
-15%
Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV Full Automation CAV Partial Automation CAV
-10% -5% Energy Impacts %
0%
5%
10% 19
Explanation of Bar Chart Format for Presenting CAV Features’ Energy and Demand Impacts: • Reductions: for visualization, the reduction from the original attributed to each feature moves from above to below the x-axis. • Increments: add on top of top of the original bar. • The final height of the bar (in the positive region only) shows the net fuel consumption including all impacts 5
Average LDV Fuel Consumption (Gallons per 100 mile)
4 3 2 1 0
Original
Reduction
Increment
-1 -2
20
