1. Integrating Mode Specific Attributes into the Transit Pathfinder in a Manner Consistent with the Multimodal Demand Model
Richard Walker and Bud Reiff, Portland Metro
Ben Stabler, Parsons Brinckerhoff, Inc.
2009 TRB Planning Applications Conference

2. Motivation
New Starts – FTA encourages integration of mode choice time parameters and path choice variables to ensure consistency in user benefit calculations
Pathfinder will choose most efficient transit submode (LRT, streetcar, bus) using criteria based upon system attributes
There are many pathfinder levers in VISUM – gives the analyst more control in replicating real-world decision process

3. Project Goal
Closely integrate transit mode choice time coefficients and path choice parameters
Mode choice
bring more explanatory power into the time variables, thus reducing (but not totally eliminating) the need for the modal preference to be expressed solely in the constant
modal preference as a constant is hard to capture in the path choice
Path choice
bring mode choice coefficient values into path choice

4. Strategy
Improve the modeling of the choice between the transit submodes
Differentiate time effects for walk, wait, transfer, and in-vehicle time – variation due to:
transit stop amenities
transit submode characteristics
Use VISUM to replicate the differentiated time parameters into the transit path choice

5. Transit Stop Amenities
Traditional stop
Small waiting area
Near curb and busy street
Presence of shelter unlikely
Little or no schedule information

6. Transit Stop Amenities
Improved stop
Shelter
Small platform
Schedule information signs

7. Transit Stop Amenities
Transit Center
Large protected platform
Shelters
Message signs
Vendors
Transit supervisors
Well lit
Good accessibility to adjoining neighborhoods

8. Hypothesis and Implementation
Characteristics of transit stop influence perception of walk time to transit and wait time (initial and transfer) for a transit vehicle
Implementation
Classify transit stop nodes in network
Classification influences walk and wait time perception, thus path choice
Incorporate the weighted perception value in mode choice

9. Network Implementation Concept
Node classified as Transit Center
Walk and wait time – less onerous than traditional stop
Line 1
Line 2
Node classified as Traditional Stop
Walk and wait time – more onerous than Transit Center

12. Strategy
Improve the modeling of the choice between the transit submodes
Differentiate time effects for walk, wait, transfer, and in-vehicle time – variation due to:
transit stop amenities
transit submode characteristics
Use VISUM to replicate the differentiated time parameters into the transit path choice

13. Transit Vehicle Attributes
Local service vehicle
Frequent stops
Route “uncertainty”
Exhaust
Noise
Travel time uncertainty due to operation in mixed traffic

14. Transit Vehicle Attributes
Fixed route circulator
Frequent stops
Route certainty (“branding”)
Exhaust free
Quiet
Travel time uncertainty due to operation in mixed traffic

15. Transit Vehicle Attributes
High capacity transit
Fewer stops
Route certainty (“branding”)
Travel time certainty
Quiet
Smooth starts and stops

16. Hypothesis and Implementation
Transit vehicle characteristics influence the perception of travel time
Implementation
Classify transit vehicles in the network
Classification influences in-vehicle travel time perception, thus path choice
Incorporate the time spent in each vehicle type into the mode choice model
Weight each by differences in perception

17. Network Implementation Concept
Transit vehicle classified as HCT
In-vehicle time parameter – less onerous than a local service vehicle
Line 1
Line 2
Transit vehicle classified as Local Service Vehicle
In-vehicle time parameter – more onerous than HCT

20. Measuring Time Perceptions
Stated preference survey currently underway
PB, RSG, Bradley
Will measure perceived time relativity between stop types and between vehicle types
Target of 900 completes
~ 75% transit users and ~ 25% auto users
Computer-assisted experiment combining text, photos, skims and GPS to create realistic choices
Findings will be applied within the model

21. Transit Mode Choice Utility Expression
Utl = f(demographics, urban form, costs, time, modal pref)
Time and modal preference:
c1a(walk – trad stop) +
c1b(walk – imp stop) +
c1c(walk – transit center) +
c2a(wait – trad stop) +
c2b(wait – imp stop) +
c2c(wait – transit center) +
c3a(IVT – local) +
c3b(IVT – circulator) +
c3c(IVT – HCT) +
modal preference
Walk times influenced by classification of transit stops
Wait times influenced by classification of transit stops
IVT by mode is retained in pathfinder and expressed in utility
Preference value by submode and weighting it by the respective IVT by OD

22. Will the Approach Work? Metro Test Case In-vehicle time
LRT: 80% of model estimation value
Streetcar: 90% of model estimated value
Bus: 100% of model estimated value
Transit stop wait and walk time
100% for all stop types
node classification feature recently incorporated into VISUM
next test will use this enhancement

23. Transit Paths – Are They Reasonable?
Before: In-Vehicle Time (1, 1, 1)  95% Rail
After: In-Vehicle Time (1, 0.9, 0.8)  100% Rail
Pioneer Courthouse Square  Lloyd Center (PM Peak)

24. Transit Paths – Are They Reasonable?
Before: In-Vehicle Time (1, 1, 1)  82% Rail
After: In-Vehicle Time (1, 0.9, 0.8)  97% Rail
Lloyd Center  Pioneer Courthouse Square (PM Peak)

25. Transit Paths – Are They Reasonable?
Before
After
Bus
81%
80%
Rail
18%
19%
Streetcar 1%

26. Conclusions Based on testing so far… Results are promising
Transit submodes can be accounted for in a direct manner
Less data management than traditional methods
Decision processes can be represented in the route pathfinding algorithm

27. Next Steps Stop classification application
Compare algorithm path choices with those revealed from on-board surveys
Link procedure with DASH (dynamic tour-based model under development at PSU – John Gliebe)
Use procedure in the estimation process using new household survey data ( )