Project Meeting 2018.05.25

Stop models

• Trip_destinations completed
  • Ran into the problem where a work tour with walk mode has a university stop, but there are no universities located within walking distance and so the model crashes
  • Solution borrowed from DaySim - re-run earlier models with new random numbers for failing choosers since the number of occurrences is small and therefore the statistical results are insignificant
  • Reworked code to handle missing trip types (atwork, subtour, school, work, etc.) for small samples, such as just one household
  • Fixed the destination fail - mark failed trips and skip them, setting subsequent trip’s origin to previous trips dest (version 0)
  • Fixed the destination fail - retry failed trips (and their leg-mate cohorts) rerunning both trip_purpose and trip_dest on failed trips with new random numbers (version 1). If any trips still failing after MAX_ITERATIONS, then fall back to version 0
• Tracking failure by iteration so we can see the impact; let's share the numbers from the full run
• Can now run just a select list of households through the models as a result
• Will start implementing trip timing and mode choice next week

Documentation / release

• Still working on updated documentation
• Added a flow chart of the example TM1 sub-models and an image map with links to sub-model configuration files
• Should be done soon

Project admin

• I revised the draft phase 4 amendment
• Report from @partners about next amendment