TRANSIMS Microsimulator Application for Improving Fuel
Consumption at Urban Corridor
Jaesup LeeUniversity of Virginia & Virginia DOT
Byungkyu “Brian” Park & Jaeyoung Kwak University of Virginia
Presented at the 12th TRB National Transportation Planning Applications Conference, May 17-21, 2009, Houston, Texas
Sponsor
• Federal Highway Administration– Broad Agency Announcement – Project Manager: Brian Gardner, FHWA
Motivations
TRANSIMS
• Extensively developed in 1990s and demonstrated in Dallas/Ft. Worth and Portland case studies
• Integrated activity based modeling and microscopic traffic simulation (cellular automata)
• Open source community
Proposed Research
• Application of TRANSIMS for Sustainable Transportation: mainly focused on Microsimulator
• Microscopic simulator calibration/validation • Integration of the Microsimulator, a fuel
consumption and emission model, and an optimizer
• Demonstrate feasibility via a case study
TRANSIMS Microsimulator
• Explicitly models individual vehicles • Updates vehicle status (e.g., speed and
acceleration) every 1-second
Calibration/Validation ProcedureSee: http://faculty.virginia.edu/brianpark/SimCalVal/
Case Study Network
Charlottesville, VA
Calibration ParametersParameters Unit
PLAN_FOLLOWING_DISTANCE mLOOK_AHEAD_DISTANCE m
LOOK_AHEAD_LANE_FACTOR -LOOK_AHEAD_TIME_FACTOR -
MAXIMUM_SWAPPING_SPEED m/secMAXIMUM_SPEED_DIFFERENCE m/sec
DRIVER_REACTION_TIME secPERMISSION_PROBABILITY %
SLOW_DOWN_PROBABILITY %SLOW_DOWN_PERCENTAGE %MINIMUM_WATING_TIME minMAXIMUM_WATING_TIME min
MAX_ARRIVAL_TIME_VARIANCE minMAX_DEPARTURE_TIME_VARIANCE min
Calibration Validation Procedure
• Experimental Design Approach – Exhaustive search infeasible– Latin Hypercube Sampling method
• Developed 200 sets and made 5 replications for each set
Default vs. Calibrated
Default Parameters
Calibration using anExperimental Design
Validation
Calibrated parameters were validated with untried field data
Achieving Sustainable Transportation: Saving Fuel Consumption…
Why VT-Micro Model?
• Current EPA Mobile uses average link speed…
VT-Micro Model
Fuel/Emission Estimation Procedure
Traffic Signal Timing Optimization
• SYNCHRO for minimizing delay and stops
• Proposed approachesi. Minimizing fuel consumptionii. Minimizing total vehicle-hours-
traveled
Genetic Algorithm Convergence
Performance Evaluation
MOE
Mobility EnergyEfficiency Emission
VHT (hr)
Number of Trips (veh) Fuel
(liter) HC (g) CO (g) NOX (g) CO2 (kg)started complet
edaverage 332.5 6062 5701 814.4 884.6 8064.5 898.7 1876.6STDEV 9.6 38.9 38.6 19.35 17.5 116.3 22.7 44.9
min 318.2 5999 5627 785.53 857.2 7827.2 862.3 1809.7max 370.3 6222 5783 894 956.6 8506.4 996.8 2061.6
SYNCHRO Optimized – Base Case
Note: Results are based on 50 TRANSIMS Microsimulator replications.
Performance Evaluation
Proposed Approach – VHT Minimization
Note: Results are based on 50 TRANSIMS Microsimulator replications.
MOE
Mobility EnergyEfficiency Emission
VHT (hr)Number of Trips (veh) Fuel
(liter) HC (g) CO (g) NOX (g) CO2 (kg)started completed
average 247.2 6222 5980 663.2 759.2 7540.6 738.5 1525.8
STDEV 2.6 0.14 17.7 6.4 6.5 57.8 7.6 14.9
min 242.6 6221 5935 651.6 748.1 7439.2 725.0 1498.9
max 254.6 6222 6013 680.6 776.2 7675.8 760.2 1566.2
Performance Evaluation
Proposed Approach – Fuel Consumption Minimization
Note: Results are based on 50 TRANSIMS Microsimulator replications.
MOE
Mobility EnergyEfficiency Emission
VHT (hr)Number of Trips (veh) Fuel
(liter) HC (g) CO (g) NOX (g) CO2 (kg)started Completed
average 243.76 6220.5 5962.6 648.9 749.0 7434.8 713.6 1492.1
STDEV 3.6 3.1 21.0 7.7 7.57 63.1 8.8 17.9
min 237.3 6207 5908 635.6 735.5 7308.4 698.7 1461.6
max 258.0 6222 6011 677.5 776.1 7616.8 744.5 1558.3
Comparison of VHT
Comparison of Fuel Consumption
Concluding Remarks
• Calibration/Validation is necessary for TRANSIMS Microsimulator to properly reflect field condition
• Successfully integrated VT-Micro emission estimation module, an GA-based optimizer, and Microsimulator
Concluding Remarks (cont’d)
• The integrated approach improved fuel consumption and emission over state-of-the-practice tool (i.e., SYNCHRO)
• More efficient computation is required for a large scale optimization