Regional Bicycle Demand Model: In Use Today in Portland

1. Regional Bicycle Demand Model:In Use Today in Portland Bill Stein, Metro TRB Transportation Applications Conference Reno, Nevada – May 9, 2011

2. Presentation overview • Attributes of Portland’s regional bicycle network • Reasons for and features of our bicycle model • Bicycle utility incorporated into regional model • Bicycle path choice / assignment • Next steps

3. Central Portland bicycle network

4. Bicycle boulevards • Low-volume, low-speed streets optimized for bike travel • Treatments include traffic calming/diversion, signage and pavement markings, and intersection controls • Through movements for bikes prioritized over other modes

5. The problem • Portland policymakers have great interest in learning what types of projects will increase bicycling. • For most trip purposes, our previous regional model used only distance as the factor influencing bike mode choice. • A tool was sought for evaluating impacts of future bike projects on mode share and route choice.

6. Project support • Portland State University • 2007 GPS survey of bicyclists’ route choices • Dr. Jennifer Dill • Development of bicycle route choice model • Dr. John Gliebe, Joe Broach • PTV America • Software support • Chetan Joshi

7. Survey participants’ home locations

8. What our bike model does • Models bicycle travel on all streets—plus multi-use paths—in the Portland-Vancouver metro area • Path choice is based on full origin-to-destination consideration (not link-to-link) • example: percent of route on bicycle boulevard • Bicycle utilities feed into regional demand model • Robust bicycle assignments, with multiple display options • Models 24-hour average weekday bicycle flows

9. Special considerations • The model currently runs on “satellite” bicycle networks which took substantial effort to build • The model is only as good as our validation data • Run time is not trivial: millions of paths are individually analyzed several times • Bicycle-transit interactions are not addressed • Path choices are based on route preferences of current bicyclists

10. Bicycle utilities in mode choice equations • Commute (HBW + College) • Mode-specific constant • Coefficients on distance and land use mix • With bike model: coefficient on bicycle utility • Non-commute (all other purposes) • Mode-specific constant • Coefficient on distance • With bike model: coefficient on bicycle utility

11. Variables affecting bicycle utility • Proportion of route on off-street paths, bike boulevards, bike lanes • Proportion of route on links with grade > 2% • Turns, traffic signals, stop signs per mile • Traffic volumes of on-street travel and opposing links at left turns • Bridge bike facility type • Distance • Commute or non-commute trip

12. Bicycle utility & distance coefficients • Bicycle utility coefficient • Bicycle distance coefficient

13. Bicycle mode choice constants

14. Bicycle model application • Skim creation • Route search based on impedance function • Returns single, best utility path between zones • Demand model • Bicycle utility added to mode choice equations • Assignment • Generates up to nine alternative routes per zone pair based on path utilities • Optimum paths are determined using a stochastic assignment process

15. 2007 bicycle productions per square mile

16. 2007 bicycle attractions per square mile

17. 2007 bicycle mean distance (miles)

18. 2007 bicycle mode share

19. Seven paths generated for my commute

20. 2007 modeled bicycle commute flows

21. 2007 modeled bicycle commute flows westbound on the Hawthorne Bridge

22. Next steps • Through mid-summer 2011: further testing of model (validation and sensitivity tests) • Beginning late summer: model will be available for RTP, MTIP, air quality, and corridor projects • Mode choice parameters will be updated when 2011 household travel survey data are available

23. Questions? Bill Stein Metro 503-797-1855 Bill.Stein@OregonMetro.gov