Diese Präsentation wurde erfolgreich gemeldet.
Wir verwenden Ihre LinkedIn Profilangaben und Informationen zu Ihren Aktivitäten, um Anzeigen zu personalisieren und Ihnen relevantere Inhalte anzuzeigen. Sie können Ihre Anzeigeneinstellungen jederzeit ändern.

That Bike is Too Heavy: Merging Bicycling Physics, Human Physiology and Travel Behavior

60 Aufrufe

Veröffentlicht am

Alex Bigazzi, University of British Columbia

Veröffentlicht in: Bildung
  • Als Erste(r) kommentieren

  • Gehören Sie zu den Ersten, denen das gefällt!

That Bike is Too Heavy: Merging Bicycling Physics, Human Physiology and Travel Behavior

  1. 1. That bike is too heavy: Merging bicycling physics, human physiology, and travel behaviour Dr. Alex Bigazzi Assistant Professor University of British Columbia Department of Civil Engineering & School of Community & Regional Planning Portland State Transportation Seminar, May 2019
  2. 2. Bicycle physics Active travel behaviour Energy & physiology 2
  3. 3. 3?
  4. 4. 4 Walking and cycling activity Travel-related outcomes Environment Traveler Energy Routing & networks Health & safety assessmentsSimulation/ operations models Travel demand models New mobility policies Equipment Speed + energy preference & capability
  5. 5. 5 Power = 9.8𝑚𝑚 𝐺𝐺 + 𝐶𝐶𝑟𝑟 𝑣𝑣 + 0.5𝑚𝑚∆𝑣𝑣2 + 0.6𝐶𝐶𝑑𝑑 𝐴𝐴𝑓𝑓 𝑣𝑣3 Mass Rolling parameter Drag parameter Energy = Power x Time Physical attributes SpeedRoad grade
  6. 6. 6 Physical attributes of real-world urban cyclists
  7. 7. 7 DEVELOPMENT OF A FIELD BICYCLE COAST- DOWN TEST • Builds on previous methods + wind & grade • Additional sensors, more complex equation • Precision similar to past indoor tests
  8. 8. 8 INTERCEPT SURVEY • Intercepted cyclists in Vancouver, summer 2016 • 9 locations in a variety of contexts (residential, waterfront, downtown, university) • 18 weekdays, noon to early evening • Questionnaire & physical testing • Travel habits • Preferences & attitudes • Socio-demographics
  9. 9. 9 SURVEY SAMPLE • 648 participants (signed consent form), 625 complete questionnaires • Demographics comparable to Metro Vancouver cyclists in a 2011 TransLink Household Travel Survey
  10. 10. 10 RESISTANCE PARAMETERS (557 CYCLISTS) Wider and higher range than in the existing literature, which focusses on sport cyclists
  11. 11. 11 HIGHER RESISTANCE PARAMETERS FOR LESS SPORT- ORIENTED CYCLISTS Negative correlations • Tire pressure & 𝐶𝐶𝑟𝑟 • Tire pressure & 𝐴𝐴𝑓𝑓 𝐶𝐶𝑑𝑑 (via bike type) Positive correlations • Tire width & 𝐶𝐶𝑟𝑟 (via pressure) • Tire width & 𝐴𝐴𝑓𝑓 𝐶𝐶𝑑𝑑 (via bike type) • BMI, age, overall mass & 𝐴𝐴𝑓𝑓 𝐶𝐶𝑑𝑑 Correlations shown significantly different from zero at p<0.05
  12. 12. 12 PHYSICAL TYPOLOGY M, H, R cyclists increasingly, significantly: • Cycle year-round • Ride faster • More weekly physical activity • More commute cycling • More “strong & fearless” & less “interested but concerned”
  13. 13. 13 BUT differences were modest and the ranges within clusters were wide Also, questionnaire data (socio-dem, habits) poorly predict physical attributes
  14. 14. 14 PEOPLE WITH LIGHTER, “BETTER” CYCLING EQUIPMENT ALSO CYCLE MORE BUT WILL BETTER EQUIPMENT MOTIVATE MORE CYCLING?
  15. 15. 15
  16. 16. 16 APPLICATIONS OF PHYSICAL PARAMETERS Signal timingPollution in halation modeling
  17. 17. 17 Collecting naturalistic cycling GPS data for speed and energy analysis
  18. 18. Smartphone-based GPS data collection • Off-the-shelf app • 1-sec GPS logging • Heart rate pairing • Direct data sharing 18
  19. 19. 19 SUMMER 2017 GPS BICYCLE TRAVEL STUDY …generated 15,000 km of cycling data over 2,300 trips (8% e-bike)… 150 participants from 12 municipalities around Metro Vancouver… … cycling at an average of 17 km/hr. 7% 28% 41% 18% 4% 1% <10 kph 10-15 kph 15-20 kph 20-25 kph 25-30 kph >30 kph47% 17% 14% 11% 8% 2% WORK ERRAND LEISURE SCHOOL OTHER EXERCISE
  20. 20. Sample characteristics 20 0% 20% 40% 60% 80% 100% 0 20 40 60 Age (years) This sample 2016 intercept survey 2011 household travel survey 0% 20% 40% 60% 80% 100% This sample 2016 intercept survey 2011 household travel survey Percent of sample <$25k $25k-$50k $50k-$75k $75k-$100 $100k-$150k >$150k
  21. 21. 21 DATA CLEANING AND PROCESSING Lat/Long data from GPS Stop periods Cumulative/ net moved distance Separate trips Raw speed >2 minutes Retained trips >1 minute >5 km/hr Filtered speed Difference Exclude >1.6x neighbors Digital Elevation Model Raw grade Filtered grade Exclude >10% Interpolate up to 4 sec gaps Kernel smoothing Difference Grade Speed Acceleration
  22. 22. Processing GPS data for high-res analysis • Isolating travel observations (removing stops) • Map matching • Filtering & smoothing 22
  23. 23. Road Grades 23
  24. 24. Road Grade Recommendations • After map-matching… • Identify elevated links (OpenStreetMaps flag) • Non-elevated links – Interpolate highest-resolution Digital Elevation Model (DEM) • Elevated links – Process cloud-point LIDAR data, if available – <100 m, assume constant-grade – >100 m, seek out design drawings or other data source 24
  25. 25. Using the GPS travel data – Speed choice modelling & e-bikes 25
  26. 26. Speed choice and speed-energy trade-off • Utility-based speed choice model with time, energy, and control factors • Parameters – Physical attributes (coast-down test + physiology models) – Marginal Rate of Substitution (MRS) between time and energy • Calibration of MRS requires observation of an equilibrium “cruising” speed 26
  27. 27. Predicted effects of e-assist on speed 27 Baptista et al., 2015
  28. 28. 28 IDENTIFYING CRUISING EVENTS IN GPS DATA • 15-45% of observations • Cruising speeds different from average speeds • Regression on grade o HIGH inter-person variability Simple Binning Time-series clustering #1 Time-series clustering #2 Time-series clustering #3
  29. 29. 29 CONSISTENT WITH SELF- ASSESSED SPEED LEVEL
  30. 30. Ongoing work – After data processing study diversion 30 • Speed/energy preferences (MRS) – Sensitive to data accuracy (especially grade) – Intra- and inter-person heterogeneity (trip purpose, weather, age, etc) • Significantly improves route prediction? – e.g., explains heterogeneity in “cost” of hills
  31. 31. 31 EFFECTS OF E- ASSIST ON SPEED DYNAMICS • ~30% faster • Greater speeds, grades, and speed + grade dynamics • Important to compare similar trips & travellers
  32. 32. 32 2x motive energy o ½ higher speed & grade o ¼ greater speed dynamics o ¼ greater resistance parameters
  33. 33. Future & ongoing work • Energy validation with heart rate data from survey • Field tests with new mobility devices (e-scooters, etc) and controlled e-assist levels – Energy validation with breath measurements • Transferability to other cities • Validation with (e)bikeshare system data • Application of speed choice model in bicycle microsimulation 33
  34. 34. reactlab.civil.ubc.ca alex.bigazzi@ubc.ca Acknowledgments: REACT Lab students, Natural Sciences and Engineering Research Council of Canada (NSERC), Social Sciences and Humanities Research Council of Canada (SSHRC)

×