TransForM—A New Regional Travel Demand

Model Developed for Prince George’s County

Howard Slavin
Srinivasan Sundaram
Andres Rabinowicz

Caliper Corporation
March 2006
Maryland-National Capital Park and Planning
Commision- Prince George’s County
Transportation Planning Section Model Team

Eric Foster
Manfredo Davila
Eric Jenkins
Harold Foster
Project Objectives

• A regional model with greater detail and

accuracy in Prince George’s County
• Conversion to TransCAD
• Modeling Refinements
• Year 2000 Update and Validation
Model Development Tasks

• Conversion of MWCOG model to TransCAD

• Additional Traffic Zones in PGC
• Greater Network Detail in PGC
• Year 2000 Trip Generation Models
• Year 2000 Trip Distribution Models
• Year 2000 Mode Choice Analysis
• Traffic Assignment
• Model Calibration
• Model Validation
Model Conversion

• Pre-validated, expanded model not

available
• Straight conversion of the MWCOG Model
from TP+ to TransCAD performed
• Results replicated fairly closely
• Some discrepancies in transit pathfinding
and model convergence
• Integration of System II network data
accomplished with TransCAD
Model Development Considerations

• Use greater geographic detail

• Retain similar model structure to accommodate
future MWCOG updates
• Make use of Windows, GIS-based software
• Update to a 2000 Base Year using available data
• Use better algorithms for transit pathfinding, trip
distribution, and traffic assignment and achieve
higher convergence and closer calibration
• Consideration of TRB Review Criticisms
TransForM Model Structure
Expanded Zone System

• 2,523 Zones
• 2,476 internal and 47 external
• 885 zones in PGC instead of 381
• Demographics from population synthesis
except for employment which was
disaggregated from MWCOG estimates
TAZ Geography
Highway Network Development

• A new regional network was used that is

geographically aligned over aerial
photography throughout the entire region
• Divided highways are “dualized” and
ramps and connectors were added. The
prior System II network for PGC was also
realigned and augmented.
• New centroid connectors throughout
• The resulting network has 36,272 links
and 22,284 nodes
Highway Network in TransCAD
Transit Network Development

• Used GPS Points from WMATA

• Stop Locations and Schedules
• GIS processing
• Links added to the road network for transit
• Comparison with TP+ routes with manual revisions
• Names matched and transferred
• A separate route system created for rail
• Access and egress links use streets for
walking and roads for park and ride
Transit Network Map
Trip Generation

• Based upon Year 2000 Wave 3 COG Panel

Survey
• Statistical Analysis of Trip Making
• Trip Rates by Jurisdiction
• Cross classification by Worker status, Auto
availability, and #persons >16 in HH
• Application with population synthesis
• Attraction rates from the MWCOG model
• MWCOG latest demographics
Trip Productions & Attractions
PGC Productions & Attractions
Trip Distribution

• Similar Gravity Models with Gamma

Impedance Functions
• Updated Friction Factors consistent with
Panel Survey Trip Lengths
• No K factors for person trips (as yet)
Trip Distribution
Intra- Avg Dur. Avg Dur.
Trip Purpose Trips zonal Pct IZ (Model) (Survey)
Trips Min. Min.

HBW - Inc 1 548,113 33,816 6.2 18.9 19.0

HBW - Inc 2 861,414 54,099 6.4 22.0 21.7
HBW - Inc 3 724,651 37,840 5.2 25.0 25.5
HBW - Inc 4 1,300,217 48,600 3.7 26.9 26.0
HBW – Total
3,755,725 174,355 4.6 26.0 24.3
(with EI and EA trips)
HBS – Total 2,592,345 345,630 13.3 14.0 11.9

HBO – Total 7,084,164 1,310,540 18.5 14.7 15.5

NHB – Total 4,525,409 705,358 15.6 15.0 13.8

MedTrk – Total 287,976 13,242 4.6 27.3 -

HvyTrk – Total 113,365 3,606 3.2 30.2 -

Trip Distribution from the Model
Mode Choice Models

• Retains the Two-stage Logit Models

• Drive alone, Carpool, & Transit
• Split among HOV2, 3, & 4+
• Market segments by auto availability
• Preliminary estimation tests suggested
the coefficients were reasonable
• Constants recalibrated
• Close match to observed mode shares
 Home-Based Work Mode Shares
2_Person 3_Person 4+_Person
Market Segment Drive Alone Carpool Transit
CP CP CP
0 96,141 191,186
0 Veh HH 32,114 32,013 32,013
(0%) (33.5%) (66.5%)
760,828 106,853 257,157
1 Veh HH 83,749 19,151 3,952
(67.6%) (9.5%) (22.9%)
2,031,556 138,538 173,463
2+Veh HH 112,596 17,126 8,815
(86.7%) (5.9%) (7. 4%)
2,792,385 341,553 621,806
All Segments 228,460 68,291 44,781
(74.3%) (9.1%) (16.6%)
Estimated versus observed mode
choice shares

Observed Estimated Observed

Estimated
Purpose Transit Transit Auto-Occ
Auto-Occ Factor
Pct Pct Factor
HBW 16.9 16.6 1.12 1.12
HBS 1.2 1.3 1.23 1.24
HBO 2.1 2.1 1.44 1.43
NHB 2.5 2.8 1.25 1.26
HBW Mode Shares by Origin Area
HBW Mode Shares by Destination
Traffic Assignment
• 5 classes, 3 time periods
• HCM 2000 suggested BPR coefficients by
functional class with some adjustments
• Calibration performed with a relative gap of .001
(80-160 iterations)
• Full model feedback with MSA flow combinations
(skim difference RMSE<.01)
• Recommend production runs to a gap of .01 or
lower for major projects (18-50 iterations)
• Model run time 1.5 hours per loop or 6 hours for 4
loops on this laptop
• More and better counts needed
Regional VMT Estimates

Region Daily VMT (in

thousands)
TOTAL 152,024
DC 8,410
VA 58,545
MD 85,069
Daily Regional Traffic Flow Map
Daily Flows: Beltway, DC, & PGC
AM Flows: Beltway, DC, & PGC
PM Flows: Beltway, DC, and PGC
Comparison of Daily Model Flows & Counts
Pct Num
County FClass RMSE Obs Total Flow Total Count Pct Diff
ALL
Counties ALL Classes 41.69 2086 28,498,288 27,817,345 2.45
ALL
Counties Freeways 15.52 105 7,310,503 7,508,328 -2.63
ALL Major
Counties Arterials 43.99 718 12,212,285 11,011,901 10.90
ALL Minor
Counties Arterials 53.26 671 4,527,832 4,324,570 4.70
ALL
Counties Collectors 89.00 411 1,051,347 1,407,578 -25.31
ALL
Counties Expressways 26.90 125 3,169,006 3,256,646 -2.69
ALL
Counties Ramps 74.42 36 222,974 227,187 -1.85
Comparison of Flows v. Counts for
Prince George’s County

County F Class RMSE Obs. Tot. Flow Tot. Count Pct

Diff
PGC ALL Classes 35.74 992 12,617,466 12,423,373 1.56

PGC Freeways 12.69 46 3,433,705 3,497,144 -1.81

PGC Major Arterials 37.69 238 3,419,712 3,153,986 8.43

PGC Minor Arterials 52.93 468 2,960,056 2,761,817 7.18

PGC Collectors 74.02 118 239,706 336,000 -28.66

PGC Expressways 26.03 97 2,421,836 2,506,410 -3.37

PGC Ramps 36.37 17 138,111 135,566 1.88

Comparison of Flows v. Counts for
Montgomery County

County F Class RMSE Obs. Tot. Flow Tot. Count Pct Diff

Montgomery ALL Classes 25.19 80 2,665,174 2,814,693 -5.31

Montgomery Freeways 15.67 20 1,708,204 1,846,695 -7.50

Montgomery Major Arterials 29.17 32 725,423 708,744 2.35

Montgomery Minor Arterials 69.58 2 17,847 12,600 41.65

Montgomery Collectors 56.14 2 7,020 11,598 -39.47

Montgomery Expressways 28.27 5 121,814 143,435 -15.07

Montgomery Ramps 123.41 19 84,863 91,621 -7.38

Comparison of Counts and Flows
by Time Period
• Similar RMSEs by Time Period
• AM a little worse, PM and Off-Peak closer
than the Daily Flows
• Indicates a Stronger Validation
VMT Comparisons

Region Model Count Ratio

Daily VMT Daily VMT (Mod/Count)
(thousands) (thousands)
Overall 15,795 15,031 1.05
DC 1,338 1,172 1.14
VA 2,498 2,754 0.91
MD (including PGC) 11,959 11,105 1.07
Prince George’s County, MD 5,824 5,776 1.01
PGC Screen lines

Screenline 2
Inside I-95/I-495
concentric ring
Ratio Flow/cnt=.95
RMSE = 28.0
Details on next
slide
Name ab_flow ba_flow ab_count ba_count tot_flow tot_count
MD 650 39,369 39,818 36,500 36,500 79,187 73,000
MD 212 14,171 13,327 10,000 10,000 27,498 20,000
Cherry Hill Rd 10,591 11,177 11,500 11,500 21,768 23,000
US 1 23,799 23,141 27,600 29,200 46,940 56,800
Rhode Island Ave 7,717 6,713 8,500 8,500 14,429 17,000
Cherrywood Lane 6,802 6,403 6,250 6,250 13,205 12,500
MD 201 NB 23,376 25,700 23,376 25,700
MD 201 SB 23,314 27,200 23,314 27,200
MD 193 25,400 24,460 25,000 25,000 49,860 50,000
BW Pkwy NB 27,585 47,500 27,585 47,500
BW Pkwy SB 29,623 47,500 29,623 47,500
Good Luck Rd 6,641 6,461 8,750 8,750 13,102 17,500
MD 450 19,334 18,437 22,250 22,250 37,772 45,000
MD 950 (Garden City) 148 3,000 3,000 148 6,000
US 50 EB 37,152 48,000 37,152 48,000
US 50 WB 0 39,158 39,200 39,158 39,200
MD 704 18,197 16,854 15,000 15,000 35,051 30,000
Ardwick Ardmore Rd 3,567 2,661 4,750 4,750 6,228 9,500
MD 202 29,931 26,728 30,000 30,000 56,659 60,000
Arena Dr 4,088 3,290 4,500 4,500 7,378 9,000
MD 214 34,814 36,504 36,500 36,500 71,318 73,000
Ritchie-Marlboro Rd 8,622 7,782 6,500 6,500 16,404 13,000
Darcy Rd 1,628 764 2,750 2,750 2,392 5,500
MD 4 25,770 30,172 32,000 32,000 55,943 64,000
Suitland Pkwy EB 20,005 10,000 20,005 10,000
Suitland Pkwy EB 9,840 10,000 9,840 10,000
Forestville Rd 4,969 10,281 7,750 7,750 15,250 15,500
Suitland Rd 11,006 10,635 10,050 10,050 21,641 21,000
Auth Rd 3,235 1,611 2,125 2,125 4,846 4,250
MD 5 33,522 34,249 32,500 32,500 67,771 65,000
Temple Hill Rd 53,192 67,200 53,192 67,200
MD 414 NB 25,285 24,000 25,285 24,000
MD 414 SB 24,148 24,000 24,148 24,000
Livingston Rd 11,795 12,237 8,000 8,000 24,031 16,000
MD 210 21,494 21,592 12,500 12,500 43,086 25,000
I-295 27,811 37,501 27,811 37,501
I-295 36,359 37,501 36,359 37,501
Total 649,390 459,365 698,877 476,575 1,108,755 1,176,852
Model User Interface
Next Steps

• Sensitivity Tests
• Further checking of networks and results
• Further minor adjustments and calibration
• Completion of Final Report & User’s Guide