Denvers ActivityBased Model Project: Status Report

1
Denvers Activity-Based Model Project Status
Report

• Erik Sabina, P.E., DRCOG
• Thomas Rossi, Principal, Cambridge Systematics
• TRB Transportation Planning Applications
  Conference, Daytona Beach, Florida, May 6-9, 2007.

2
Status

• Software spine, end to end
• Some draft components integrated
• Model estimation data complete
• Point demographic data
• Transit skim comparison to survey data
• Model estimation underway
• Integration of PopSyn in progress
• Packaging GISDK code for GUI call

3
Approach

• Services and components

4
Comparison to Other Approaches

• Choice structure similar to Sacramento
• Minimize risk its worked before
• Differences
• TAZ to point level (Sacramento uses TAZ to parcel
  to point)
• Software approach very different
• More work in-house

5
IRM Model Structure
Zonal Socioeconomic Data(Households,
Employment, Parcels)
AccessibilityMeasures
Land Use Model
List of Personswith Characteristics
Logsum Variables
Long-Term Models
Person Day Models
List of Person Tours,Extra Stop Purposes
Tour Models
Tours by Purpose withDestination, Mode,
Time-of-Day
NetworkLevel-of-Service
Trip Models
Auto and Transit Person andVehicle Trip Tables
by Time-of-Day
Highway and Transit Assignment
Highway and Transit Networks
6
Key Lessons Selling Your Project

• Sell, sell, sell (to everyone you can think of!)
• A project champion is indispensable (and
  several champions is better)
• Develop project goals, and stand on them

7
Key Lessons Project Management

• Project managers need to be oxymorons
• Flexible but insistent
• Demanding but patient
• Cost conscious, but willing to spend to get it
  right
• You wont complete your project over everyones
  dead bodies
• Evaluate your strengths and weaknesses

8
Key Lessons Teamwork

• Pick your project partners carefully
• Youll be working closely with them for a long
  time
• Big team may be better, will be slower (and more
  expensive)
• Doing a lot in-house is slow
• But important to your understanding
• And fun, if you like that sort of thing

9
DRCOG IT Three components Software spine GUI,
Database, communications GISDK elements
Cambridge Systematics IT Seven components Design
consultation
CS Estimation team Seven components DRCOG
oversight/assistance
DRCOG Estimation team Three components, plus
PopSyn
CS Calibration Team By component Overall model
DRCOG Calibration Team Calibration data Model run
support
10
Key Lessons Project Design

• Simplicity!
• Produce products in a stream
• Rebuilding a trip-based model is hard, too
• Make your best schedule estimate, and double it
• Think carefully about project goals Vision
  Phase

11
Key Lessons Software

• Custom software takes a long time to produce!
• A good IT group really helps
• Youll need to program better
• Definitely do iterative software development

12
Key Lessons Data

• Estimation data takes a long time to produce!
• Be VERY nice to your sources of disaggregate data

13
Key Challenges

• Getting the team to agree on approach
• Do enough work in-house (but not too much)
  paying now vs. paying later
• Designing a model (and application software) that
  can be operated
• By us, by consultants, by planning partners
• Dealing with staff turnover

14
More Key Challenges

• Drinking from the information fire hose
• Data, software, choice models, etc.
• Knowing when to stop at each step (anything can
  be improved)
• Tying into EVERYTHING

15
Things You Should Do Differently

• Think, but dont over-think at the start
• Think over your collaborative ambitions
• Start developing estimation data now (I mean
  today)
• Have consultants review it as you go
• Decide in advance the acceptability of options
  (trade-offs) and what youre willing to pay to
  choose a superior option

16
Contact Information

• Erik Sabina, esabina_at_drcog.org, 303-480-6789
• Tom Rossi, trossi_at_camsys.com, 617-354-0167