Development and sensitivity
testing of alternative mobility
metrics in a regulatory context
John Gliebe, RSG, Inc.
James Strathman, Portland State University
Steven Tuttle, RSG, Inc.
Myra Sperley, Oregon DOT Research Section
Prepared for:
TRB Planning Applications Conference
7 May 2013
Acknowledgments
•
•
This work was funded by an Oregon DOT Research (SPR 716)
The authors are grateful for the contributions of the
following individuals:
•
•
•
•
•
•
•
•
•
•
•
•
Amanda Pietz,ODOT Research
Sam Ayash, ODOT TPAU
Terry Cole, ODOT Region 2
Kathryn McGovern, PSU
David Ruelas, PSU
David Boyd, TAC
Jazmin Casas , TAC
Brian Gregor, TAC
Douglas Norval, TAC
Lidwien Rahman, TAC
Michael Rock, TAC
Mark Vandehey, TAC
2
Background
• Oregon Highway Plan’s (OHP) mobility policies guide planning and
programming by Oregon Department of Transportation (ODOT).
• ODOT has land use change review responsibilities under the
Transportation Planning Rule, as adopted by the state’s Land
Conservation and Development Commission.
• A single volume-to-capacity (v/c) metric currently supports OHP
mobility policies and may be the basis for requiring mitigation.
Sometimes this stops the project.
• Critics of the single facility-based v/c measure charge that it is
focused too narrowly on operational objectives.
• In many cases, adherence to this standard has undermined
community economic development, compact growth, and non-auto
mode share objectives.
• Numerous alternative performance measures have been suggested
that would better capture these concerns; however, many of them
are difficult to predict as an outcome of a particular land use
change proposal.
3
Objectives
• Demonstrate the potential use of alternative mobility
metrics for evaluation of large-scale land use change
proposals
– Related to goals found in the Oregon Highway Plan promoting
non-SOV travel and efficient land use patterns
• Explore how these metrics co-vary with each other
and V/C
– Variation across inputs
– Variation across spatial dimensions
• Provide information for consideration of metrics by
policy boards or as part of transportation system
planning (TSP) process
4
Case Study Methodology
• Chose a representative land use scenario for model
based analysis
– Previously analyzed by ODOT without pending decisions
• Northgate Lifestyle Center proposal – Medford
–
–
–
–
Centrally located
Served by transit
Near highway interchanges
Semi-mixed use
• Analyzed “build” and “no build” scenarios
– 2010 Opening Year
– 2025 Future Year
• Sensitivity tests on alternative futures
– Fringe growth
– Scaled up development
– Conserved growth
5
Criteria for Selection of Metrics
•
•
•
•
•
•
•
•
Provides evidence of a change in travel activity that related
to an OHP policy (e.g., promoting non-motorized travel
modes)
May be theoretically or empirically linked to land use, socioeconomic, or transportation system inputs
Robust over a range of inputs values
Can be forecast using established methods and data
Set of metrics should be complementary, avoid redundancy,
offer a range of perspectives
Set of metrics should represent all travel modes and markets
Set of metrics should include both facility-specific and areawide measurements
Should not include direct measurement of non-travel activity
– “Second-order effects” that results from travel-activity
– E.g., economic impacts, safety impacts, environmental impacts
6
Metrics Selected
•
•
•
•
•
•
•
Network wide V/C
Total vehicle hours of travel time
Person hours of travel time
Average person trip travel time
Trip length distributions
Mode shares
Regional accessibility to employment/shopping
– By Auto, Transit and Walk
Aiemp   E j  f cij 
j J
• Local accessibility to employment/shopping (20-min.
neighborhood)
– By Auto, Transit and Walk
7
Other Metrics Considered
The study team’s review of literature revealed a long list
metrics to consider. Some of the more noteworthy metrics
that we rejected for this study, included…
– Land use variables related to urban form, street connectivity,
lane miles of bike and pedestrian facilities
 Why? Existence value not easily quantified in terms of travel
behavior. Focus should be on the traveler response.
– Reliability indices – planning time index, buffer time index, 95th
percentile travel time
 Why? Difficult to forecast and attribute to a facility (area-wide
measures). Ambiguous implications—very high congestion—
reliably congested.
– Congestion duration, queuing, recurring delay
 Why? Impossible to forecast with static network assignment
models. Need DTA.
8
Study Area
• 219,300 square foot
office park
– professional services and
light industrial uses
• 417,500 square feet
retail shopping space
• 167,000 square foot
business park
• Intra-development
Trolley
Trolley following Central Ave
9
Study Area Cities and TAZ System
Northgate
Site
10
Network Model
Rogue Valley MPO
(RVMPO)
Model Version 2, using
JEMnR platform
– Supplied by ODOTTPAU
– Converted from EMME/2
to EMME/3
– 759 TAZs, 8671 links,
3016 nodes
– 3 TAZs comprise the
Northgate development
Northgate
Site
11
Study Districts Used for Analysis of Spatial Focus
Concentric Study
Districts
–
–
–
–
Site TAZs
Approx. 1 mile out
Approx. 4 miles out
Entire region
12
Classifications of Trips by District
Used to establish spatial focus
1. If either the origin or the destination of a trip belonged to
one of the TAZs on the map shown as District 1, then the trip
was considered to belong to District 1.
2. If either the origin or the destination of a trip belonged to
one of the TAZs on the map shown as District 2, inclusive of
District 1, then the trip was considered to belong to District
2.
3. If either the origin or the destination of a trip belonged to
one of the TAZs on the map shown as District 3, inclusive of
Districts 1 and 2, then the trip was considered to belong to
District 3.
4. All trips were considered to be part of District 4. For
example, a trip with a trip end in District 1 will also be
included in the tabulations for Districts 2, 3 and 4.
13
Network Wide V/C Change Analysis
Example: 2025 Baseline vs. Northgate Scenario
14
Travel Time Metrics
Total Network Travel Time
2010
Baseline
2025
Northgate % Change
Baseline Northgate % Change
Auto/Truck Vehicle Miles (VMT)
1,742,599
1,750,526
0%
2,109,860 2,118,955
0%
Auto/Truck Vehicle Hours (VHT)
67,232
67,552
0%
80,681
81,061
0%
Transit Trip Miles
3,629
3,520
-3%
4,049
3,945
-3%
Transit Trip Hours
3,152
2,992
-5%
3,600
3,450
-4%
Person Hours of Travel Time
2025
Mode
Baseline by Study District
1
2
3
4
Walk
50
1,272
Bike
5
120
742
11
269
0
17
Walk to Bus
PnR Bus
7,329 11,134
Northgate by Study District
1
2
3
4
404
1,491
7,167 10,854
1,067
43
147
744
1,615
2,433
90
316
145
184
0
16
Percent Change
1
2
3
4
708%
17%
-2%
-3%
1,064
703%
23%
0%
0%
1,578
2,377
726%
18%
-2%
-2%
139
177
0%
-6%
-4%
-4%
Drive Alone
251
3,823 23,851 32,397 1,915
5,159 24,278 32,666
662%
35%
2%
1%
Drive w Pasg.
Passenger
204
225
3,581 19,826 26,762 2,096
3,999 21,136 28,682 2,393
5,052 20,212 26,945
5,641 21,470 28,751
929%
962%
41%
41%
2%
2%
1%
0%
All
747 13,081 74,645 102,660 6,941 17,823 75,588 102,834
830%
36%
1%
0%
15
Trip Length Distributions
Example: 2025 Baseline vs. Build by Study District
70%
70%
District 1 Trip Length Distribution:
2025 Base vs 2025 Northgate
60%
50%
40%
40%
Share
Share
50%
30%
District 2 Trip Length Distribution:
2025 Base vs 2025 Northgate
60%
30%
20%
2025 Base
20%
2025 Base
10%
2025 NG
10%
2025 NG
0%
0%
0 to 2
2 to 4
4 to 6
6 to 8
0 to 2
8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20
2 to 4
4 to 6
6 to 8
8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20
Miles
Miles
70%
70%
District 3 Trip Length Distribution:
2025 Base vs 2025 Northgate
60%
50%
40%
40%
Share
Share
50%
30%
District 4 Trip Length Distribution:
2025 Base vs 2025 Northgate
60%
30%
20%
2025 Base
20%
2025 Base
10%
2025 NG
10%
2025 NG
0%
0%
0 to 2
2 to 4
4 to 6
6 to 8
8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20
Miles
0 to 2
2 to 4
4 to 6
6 to 8
8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20
Miles
16
Modes
Mode Shares
2025
Mode
Baseline by Study District
1
2
3
4
Northgate by Study District
1
2
3
4
Percent Change
2
3
1
4
Walk
1%
3%
5%
6%
1%
2%
4%
6%
-16%
-21%
-6%
-4%
Bike
0%
1%
1%
1%
0%
0%
1%
1%
-2%
-9%
-1%
-1%
Walk to Bus
1%
1%
1%
1%
0%
0%
1%
1%
-16%
-4%
-3%
-2%
PnR Bus
0%
0%
0%
0%
0%
0%
0%
0%
0%
-33%
-6%
-4%
Drive Alone
36%
33%
35%
35%
30%
32%
35%
35%
-18%
-3%
0%
0%
Drive w Pasg.
Passenger
28%
33%
29%
34%
28%
31%
27%
30%
31%
37%
30%
35%
28%
31%
27%
30%
9%
13%
2%
3%
1%
0%
0%
0%
100% 100% 100% 100%
0%
0%
0%
0%
All
100% 100% 100% 100%
Trips by Mode
2025
Mode
Baseline by Study District
1
2
3
4
Northgate by Study District
1
2
3
4
1
Percent Change
2
3
4
Walk
107
3,526
32,773
58,103
911
3,902
31,147
55,538
749%
11%
-5%
-4%
Bike
29
676
4,599
7,455
282
861
4,598
7,403
881%
27%
0%
-1%
Walk to Bus
40
625
4,320
7,053
333
843
4,266
6,919
742%
35%
-1%
-2%
0
47
416
491
0
44
396
470
0%
-7%
-5%
-4%
2,622
41,628 243,231 351,213
21,708
56,383 246,798 352,448
728%
35%
1%
0%
Drive w Pasg. 2,044
Passenger
2,384
36,158 191,229 270,611
42,185 213,363 301,944
22,377
27,020
51,725 194,983 271,714 995%
60,773 217,293 302,377 1033%
43%
44%
2%
2%
0%
0%
40%
1%
0%
PnR Bus
Drive Alone
All
7,225 124,845 689,931 996,869
72,633 174,530 699,481 996,869
905%
17
Regional Accessibility Baseline Build Scenario
Total
Households
Transit
Auto /
Highway
Walk
18
Local Accessibility
Measuring the 20-minute neighborhood
2010 Study District
Retail
Work
Mode
Auto
1
2
3
2025 Study District
4
1
2
3
4
3%
3%
3%
3%
3%
3%
3%
3%
19%
19%
1%
1%
17%
16%
1%
1%
Walk
122%
24%
2%
2%
109%
22%
2%
1%
Auto
12%
12%
12%
11%
10%
10%
10%
9%
Transit
40%
64%
9%
7%
37%
58%
7%
6%
210%
44%
7%
6%
183%
40%
6%
5%
Transit
Walk
For example: if your spatial focus is limited to District 3, then the
Northgate scenario would result in a 7% increase in access to retail
shopping opportunities (employment) in 2010, using the 20-minute
neighborhood concept.
Assumptions: walk speed 3 mph, bike speed 9 mph
19
Sensitivity Tests
• Relocating the
Development to a
Fringe Area
• Scaling Up the
Development
– (2X and 5x)
• Conserved Growth
– no net gain in total
employment
– Subtracted
Northgate
employment from
elsewhere
20
Lessons Learned
• Fringe Growth
– Lower impact on surrounding transportation facilities
– Fewer total trips attracted, but nearly all auto
– Net V/C, PHT, Average Person minutes, Regional accessibility,
number of trips by mode and study district capture differences
• Scaled Up Development
– More dramatic positive and negative changes
– Many more local trips, and many more regional trips—offsetting
impacts
– Net V/C, PHT, Average Person minutes, Regional accessibility,
number of trips by mode and study district capture differences
• Conserved Growth
– Shows how a new regional center will draw trips away from
other neighborhood locations
– Net impacts may be negative or positive (negative mostly in this
case)
– Net V/C and regional accessibility capture differences best
21
Regional Accessibility Conserved Growth Scenario
Total
Households
Transit
Auto /
Highway
Walk
22
Lessons Learned
•
•
•
•
The geographic distance at which one measures land use
change impacts is important—affects attenuate further from
the source of change. Not surprising, but important for
regulatory usage.
At the regional level, all modeled scenarios led to slight
increases in auto travel and slight net reductions in non-auto
travel.
The concentration of a large amount of commercial
development in a single location has non-linear increasing
effects on trip attractions.
Because the model system is production constrained and
because the build scenarios assumed only an increase in
employment, without increases in households and workers,
scenarios involving an increase, decrease or change in
location of employment due to the Northgate development
all produced the same number of total trips for the region.
23
Assessment of Metrics
• Network-wide V/C Changes
– Best for showing direct impacts and can show offsetting effects
if evaluated network wide
– Does not explain why changes occur where they do
• Total Network Travel Time and Distance
–
–
–
–
Theoretically nice for portraying total network impacts
Not sensitive enough to local changes---too aggregate
Potentially misleading—hides problems
Lacking in insights
• Total Person Hours of Travel Time
– Captures both increased trip lengths and mode shifts together
– Potentially misleading (e.g., walk time increase may be
beneficial)
– Misses out on external markets and trucks
24
Assessment of Metrics
•
Average Person Trip Lengths & Trip Length Distributions
– Nice to show changes in average trip lengths
– Does not provide enough insight on underlying behavior
– Potentially misleading—regression to the mean
•
Mode Shares
– Percentage shares can be misleading due to small magnitudes of some
modes
– Number of trips by mode and total trips are useful as diagnostics, but
difficult to use in a standardized way
•
Regional Accessibility
– Good for showing benefits of travel differentiated by mode
– Needs to be put into context of households (or whoever benefits)
•
Local Accessibility (20-minute neighborhood)
– Very little regional variation for small areas (need to resize buffer)
– Arbitrary buffer, misleading treatment of trips within buffer
25
Recommendations for Further Consideration
• Network-wide V/C Budget
– Familiar measurement concepts
– May be extended to include V/C “budget”
– Improved V/C on some facilities would offset worsened V/C on
others in mitigation negotiations
– Requires precise measurements of V/C using network models
that can portray pluses and minuses
• Regional Accessibility
–
–
–
–
–
Closely related to economic benefits calculations
May be derived precisely from econometric formulations
Should be weighted by households or other beneficiaries
Could be simplified and standardized
TBD: form of impedance functions, spatial units
26
Methodological Recommendations
•
•
•
Limitations of trip-based modeling and static network assignment
are “exposed” in this type of analysis.
Activity-based models would respond more appropriately because
discretionary, secondary stop making would vary based on
accessibility (not production constrained). Tour-based travel
paradigms might respond differently, as well.
Dynamic Traffic Assignment (DTA) would enable us to consider
additional mobility metrics related to reliability, e.g., recurring
delay, duration of congestion, and queuing.
27
Questions and Answers
For more information:
John Gliebe, RSG 802-295-4999