A New Performance Measurement Tool

1
Probe Vehicle Data

• A New Performance Measurement Tool

Rick Schuman, INRIX Transpo 2008 September 24,
2008
2
What Can We Do With This?
3
We Did This
http//scorecard.inrix.com
4
INRIX National Traffic Scorecard

• 2007 Annual Report of traffic conditions
• Major highways in top 100 metropolitan US markets
• Covering 200M people
• Specific road segment performance analysis
• Hour of day, day of week
• 30K segments, 47K miles
• First version released in June
• Recurring congestion focused
• Not weighted by volume or lane miles
• Extensions, expansions possible
• Reliability, coverage, weighting, causation, etc.
• Periodic updates planned

5
Scorecard Objectives

• Credible data
• Demonstrate the value of this kind of data to
  assist in performance analysis
• Enrich ongoing policy debates
• Another way of examining congestion, trends, etc.
• Lay the foundation
• Periodic ongoing public reports
• Offer data to analysts to go deeper (nationally,
  regionally)
• Public feedback on results (and what isnt
  analyzed) guide future directions

6
The Road Network Analyzed
7
Regions in Florida Analyzed

• 9 Regions CBSAs
• gt 14M Population
• 3050 road miles

8
Results

• Two Types
• Congestion levels (regional, national, hourly,
  daily, etc.)
• Bottlenecks (specific road segment detail)
• National Results
• Drive Time Travel Time Index
• Day/Hour Travel Time Index
• Metro-to-Metro Comparisons
• Overall congestion
• Travel Time Index
• Nations worst bottlenecks
• Metropolitan Summaries
• Congestion compared to 2006
• Congestion compared to Worst city (LA)
• Worst Hour congestion
• Worst bottlenecks

9
Components for Analysis

• Average Speed
• Each hour/day of week
• Monday 8-9am
• 168 values for each road segment
• Reference Speed
• Uncongested/free flow speed
• One value for each segment
• Capped at 65 MPH

10
Methodology Basics

• Congestion
• For each segment, each hour
• If avg speed lt ref speed, then adds to congestion
• Amount added based on how slow and how long the
  segment
• Sum up all congestion factors in a metro hourly
  congestion total
• Peak Period M-F 6-10am, 3-7pm
• Bottlenecks for each segment, each hour
• If avg speed lt ½ ref speed then hour
  congested
• Avg speed while congested intensity of
  congestion
• Score is hours congested intensity of
  congestion

11
Example Road Segment

• Where
• Orlando
• Road
• I-4
• Direction
• Eastbound
• Segment
• Exit 35 (Kaley Ave)
• Length
• 0.48 miles

12
I-4 EB _at_ Exit 35 (Kaley Ave)

• 2007 Reference Speed 58 MPH
• 2007 Hourly Average Speeds (MPH)
• Speeds above or near reference speeds 10pm to 5am
  (not shown)

13
Congestion Analysis

• Focus on Peak Days/Hours (Green)
• Calculate Travel Time Index (TTI) for each Peak
  Hour
• These TTIs building blocks for all congestion
  results

14
Bottleneck Analysis

• Analyze all 168 hours
• Reference Speed 58 MPH
• Congested Hour lt 29 MPH (in red)
• Total Hours Congested 41
• Average Speed While Congested 18.7 MPH
• TTI while Congested 58/18.7 ? 3.10
• Bottleneck score 413.10 ? 123.41
• 123.41 becomes basis for comparison with all
  other segments

15
Congestion Results
16
National Summary Factoids

• 2007 Natl Travel Time Index 1.133
• Random trip on network 13.3 extra time needed
  in peak hours
• 1.9 increase from 2006
• National Factoids (Drive Times)
• Worst Day Friday 14
• Worst commute Friday PM 20
• Worst hour Friday 5-6 PM 26
• Worst Morning Wed AM 13
• Best Day Monday 11
• Best Commute Friday AM 9
• Best Hour Friday 6-7 AM 6
• Best Afternoon Monday PM 11

17
National TTI By Hour/Day
Travel Time Index (TTI)
Hour of Day
18
Metropolitan Congestion Rank

• Ten Most Congested Regions Nationally

19
Metropolitan Congestion Rank

• Where do Florida regions rank?

20
Metropolitan Travel Time Index Rank

• Ten Highest TTIs Nationally

21
Metropolitan Travel Time Index Rank

• TTIs in Florida

22
Worst Congested Hour Rank

• Nationally

• Florida

23
Why results might differ from other analyses?

• Coverage
• Mainline highways only, no ramps, arterials, etc.
  
• Speed focus, not volume, lane-miles weighted
• Varying network coverage in regions
• TTI scores likely lower than most studies
• Scorecard includes greater fringe metro coverage
• Reference speed used vs. flat free flow speed
  (e.g. 60 MPH)
• Definition of metro areas could be different
• E.g., some reports include Baltimore with DC
  metro region

24
Bottlenecks
25
Top 1000 National Bottlenecks (Red)
26
Where the top 1000 bottlenecks are
25 of nations worst 1000 bottlenecks in 2007 in
Florida
27
Nations Worst Bottlenecks
28
Florida Metropolitan Summaries
29
Florida Metropolitan Summaries (2)
30
Florida Metropolitan Summaries (3)
31
Florida Metropolitan Summaries (4)
32
2008 Results?
33
2008 End of an Era?

• Fuel Prices per Energy Information Administration
• 2007 to 1H2008 2.85 to 3.50 (23)
• Traffic Volumes Urban Interstates per FHWA
• 1H2007 to 1H2008 -2.2
• Urban Congestion Reporting TTI per FHWA
• 2Q2007 to 2Q2008 1.287 to 1.242 (-3.5)
• What does the probe data say?

34
Scorecard 1H2008 Analysis

• National Travel Time Index
• 2007 to 1H2008 1.133 to 1.098 (-3.0)
• 96 of 100 markets had drop in congestion
• Other 4 had increases of under 0.3
• Worst 1000 bottlenecks
• 21 less total congestion in 1H2008 than 2007
• Thursday and Friday 5-6pm in dead heat for most
  congested hour
• TTI 1.193
• National peak hour TTIs down every hour/day

35
Congestion down all hours, days from 2007

• 2007 TTI By Hour by Day

TTI Difference 2008 vs. 2007

• 1H 2008 TTI By Hour by Day

36
1H2008 Metropolitan Congestion Rank

• Top 10 Nationally

37
1H2008 Metropolitan Congestion Rank

• Florida

38
1H2008 Travel Time Index Rank

• Nationally

39
1H2008 Travel Time Index Rank

• Florida

40
Summary
41
Summary

• Keeping score is possible and increasingly easy
• More markets, more miles, more discrete
  segments/time slices previously possible
• Output of results can occur quicker than
  everlike quarterly financial reports?
• Data open to much interpretation
• Butquick, obvious conclusions
• Even in a slowing economy with high fuel prices,
  congestion is a pervasive problem
• Los Angeles and New York far ahead of pack in
  terms of scale of slowdowns
• Points more to how bad it is there not how good
  it is in other places
• Bottlenecks are everywhere not just at big
  interchanges in the largest cities
• No margin for error in many cities
• If an accident or bad weather, these roads become
  hopeless quickly
• 2008 improvementsthe deficit analogybetter,
  but still awful

42
Probe Vehicle Data
A New Performance Measurement Tool
Rick Schuman, INRIX rick_at_inrix.com 407-298-4346
http//scorecard.inrix.com