Embed Size (px)
Citation preview
Use of Truck GPS Data for Travel Model ImprovementsTalking Freight Seminar April 21, 2010
22
Presentation OverviewTruck GPS Data
• Vehicle Speed Comparison
• Grocery Store Trip Generation
• Discussion
33
Speed Differences on FreewaysTruck GPS Data
Truck Speeds
Trip Generation
Discussion
Trucks Travel Slower than Cars
• Acceleration
• Deceleration
• Speed Maintenance
Modeling Implications
• Travel Demand Model
• Air Quality Model
• Analysis
44
Motivation ContinuedTruck GPS Data
Truck Speeds within PSRC’s Modeling Framework
• Trucks speeds are not differentiated
• Overestimation of trucks on freeways
• Underestimation of trucks on arterials
Facility Type Count Volume Locations DifferencePercent Difference
Freeways 552,755 960,468 186 407,714 74%
Arterials 679,012 267,427 581 (411,585) -61%
Total 1,231,767 1,227,895 767
(3,871) 0%
Truck Speeds
Trip Generation
Discussion
55
GPS DataTruck GPS Data
• Collected in 2003 and 2004
• Nearly ½ million observation points
• From 8 trucking firms and 25 GPS units
• 14 routes considered
• No comparable passenger vehicle data available
Truck Speeds
Trip Generation
Discussion
66
GPS Analysis ResultsTruck GPS Data
Midday General Purpose (GP) and Truck Speeds
0
10
20
30
40
50
60
I-405
: fro
m N
E 8th
to N
E 170t
h
I-405
: fro
m I-
5 to
NE 8th
I-405
: fro
m N
E 8th
to I-
5
I-405
: fro
m 1
70th
to N
E 8th
I-5: f
rom 1
84th to
Oliv
e
I-5: f
rom O
live to
SR 5
26
I-5: f
rom O
live to
184
th
I-5: f
rom S
R 526
to O
live
I-90:
I-5
to Iss
aqua
h
I-90:
Issa
quah
to I-
5
SR 167
: 84t
h to
15th
SR 167
: 15t
h to
84th
SR-520
: I-5
to 1
48th
SR-520
: 104
th to
I-5
Sp
ee
d (
MP
H)
GP TruckTruck Speeds
Trip Generation
Discussion
77
Short and Long Vehicle ComparisonTruck GPS Data
Data
• 20-second speed trap data from the Washington State Department of Transportation
• 4 vehicle length classes
• Bin 1: < 26’ (short—i.e. cars)
• Bins 2-4: > 26’ (long—i.e. trucks)
• Only consider observations where either cars or trucks are present
Truck Speeds
Trip Generation
Discussion
88
Short and Long Vehicle AnalysisTruck GPS Data
Comparison of Daily Average Truck GPS and Long Vehicle Speeds
GPS Long Vehicles
Speed Observations Speed Observations
45.8 10 44.8 292
Truck Speeds
Trip Generation
Discussion
99
Short and Long Vehicle AnalysisTruck GPS Data
Time Short Long Difference
Percent Difference
AM Peak 57.1 46.8 10.3 -18%
PM Peak 53.7 42.5 11.2 -21%
Weighted Average Speed DifferencesTruck Speeds
Trip Generation
Discussion
1010
Short and Long Vehicle AnalysisTruck GPS Data
Comparison Over Time
0
10
20
30
40
50
60
70Spee
d (M
PH
)
AM short 00 AM short 06 AM long 00 AM long 06
Truck Speeds
Trip Generation
Discussion
1111
Short and Long Vehicle AnalysisTruck GPS Data
Speed and Car Percent at Select LocationsTruck Speeds
Trip Generation
Discussion
1212
In and Out Lane ComparisonTruck GPS Data
Data
• 5-minute speed trap data from the Washington State Department of Transportation
• Innermost non-HOV lane as proxy for car speeds
• Second to outermost lane as proxy for truck speeds
Truck Speeds
Trip Generation
Discussion
1313
In and Out Lane AnalysisTruck GPS Data
Time Short Long Difference Percent Difference
AM Peak 54.7 48.6 6.1 -11%
PM Peak 54.7 50 4.7 -9%
Weighted Average Speed DifferencesTruck Speeds
Trip Generation
Discussion
1414
In and Out Lane AnalysisTruck GPS Data
0
10
20
30
40
50
60
70Speed (M
PH
)
AM in 00 AM in 06 AM out 00 AM out 06
Comparison Over TimeTruck Speeds
Trip Generation
Discussion
1515
DiscussionTruck GPS Data
0
10
20
30
40
50
60
70Spee
d (M
PH
)AM short AM long AM in AM out
Method ComparisonTruck Speeds
Trip Generation
Discussion
1616
Moving ForwardTruck GPS Data
• The in and out lane method yields the most promising results and is simple to implement
• In 2006 trucks traveled by an average of 10% slower than cars, virtually unchanged from 2000
• Future truck model will account for 10% difference on freeways
• The short and long vehicle comparison could be useful if disaggregate classification data is available
• Continued need for truck count information for model validation
Truck Speeds
Trip Generation
Discussion
1717
GPS Based Trip GenerationTruck GPS Data
Truck Speeds
Trip Generation
Discussion
Data from the Washington State Department of Transportation (WSDOT) and University of Washington
2,500 trucks per day
15 minute reads
> 3,000,000 records per month
Examined data from Fall 2008
Geocode observation to road network
1818
Defining Origins and DestinationsTruck GPS Data
Truck Speeds
Trip Generation
Discussion
Traffic based versus intentional stops
Use 3-minute dwell time to differentiate
One month of data:
• 3,000,000 reads
• 358,000 trips
• 16 mile average trip distance
• 21 minute average travel time
• 34 miles per hour average speed
1919
Grocery Store Trip GenerationTruck GPS Data
Truck Speeds
Trip Generation
Discussion
Truck O/D data related to grocery stores and distribution centers in the Puget Sound
Over 91 days:
• 2,400 trucks
• 22,000 tours
• 215,000 individual trips
• 9 tours per truck
• 0.1 tours per day
• 10 trips per tour
• 2 trips to major grocer
2020
ValidationTruck GPS Data
Truck Speeds
Trip Generation
Discussion
GPS dataset is subset of all trucks
Favorable comparison to interview information (10 to 12 daily trucks)
But half of observed manual counts
2121
Daily Trips by Area TypeTruck GPS Data
Truck Speeds
Trip Generation
Discussion
Land Use Daily Trucks
Metropolitan Cities 12.4
Core Cities 12.1
Larger Cities 8.4
Smaller Cities 6.6
Unincorporated Urban Areas 7.3
Rural 3.9
2222
DiscussionTruck GPS Data
Truck Speeds
Trip Generation
Discussion
Can estimate total trips by referencing truck count data
disaggregated by employment sectors, land use types and times of day
Average trip and tour lengths
Useful for estimating and calibrating aggregate distribution models
Calibrate aggregate trip generation models
Speed data and route choice
Potential for commodity flow model
Improving quality of GPS data
2323
Thank YouTruck GPS Data
Alon [email protected]
Puget Sound Regional Councilhttp://psrc.org/
Truck Speeds
Trip Generation
Discussion
