shrp2 c10a

18
SHRP2 C10A Final Conclusions & Insights TRB Planning Applications Conference May 5, 2013 Columbus, OH Stephen Lawe, Joe Castiglione & John Gliebe Resource Systems Group

Upload: howie

Post on 14-Jan-2016

31 views

Category:

Documents


3 download

DESCRIPTION

SHRP2 C10A. Final Conclusions & Insights. TRB Planning Applications Conference May 5, 2013 Columbus, OH Stephen Lawe, Joe Castiglione & John Gliebe Resource Systems Group. C10A Project Objectives. Current models are limited - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: SHRP2 C10A

SHRP2 C10A

Final Conclusions & Insights

TRB PlanningApplications ConferenceMay 5, 2013Columbus, OH

Stephen Lawe, Joe Castiglione & John GliebeResource Systems Group

Page 2: SHRP2 C10A

2

C10A Project Objectives

Current models are limited Not sufficiently sensitive to the dynamic interplay

between travel behavior and network conditions Unable to represent the effects of policies such as

variable road pricing and travel demand management strategies

Advanced model systems can better represent demand changes and network performance

Peak spreading, mode choices, destination choices Capacity and operational improvements such as signal

coordination, freeway management and variable tolls, TDM

Page 3: SHRP2 C10A

3

C10A Model System

Model components exchange information in asystematic and mutually dependent manner

C10A model components Daysim “activity-based” model TRANSIMS network simulation model MOVES

C10A linked model system implemented in both Jacksonville, FL and Burlington, VT

“Linked” not “Integrated”

Page 4: SHRP2 C10A

4

How are the model system components linked?

Daysim activity-based model provides travel demand to TRANSIMS network simulation model

Minute-by-minute Parcel-to-parcel Detailed market segments (toll/notoll, trip-specific VOT) 1 hour to simulate 1 million people on laptop, ½ hour on server

TRANSIMS provides information on network performance by time-of-day, as detailed as:

10 minute skims Activity locations ~50 VOT classes in assignment

“Studio” controls model system execution and equilibration

Page 5: SHRP2 C10A

5

Application Considerations

Different policy questions require different methods for running the model system

Disaggregate framework Supports more detailed analysis Extracting, managing and

interpreting results is straightfoward

Volume of information is significant

Simulation variation Not an issue for activity-model Significant issue in network

simulation

Planning & Operations

Planning

Operations

Page 6: SHRP2 C10A

6

Conclusions

Integrated model system is more sensitive to a wider range of policies produces a wider range of statistics of interest to

decision-makers

Level of effort required to effectively test different types of improvements varied widely

Debugging the model system, and individual scenarios was the greatest challenge

Must have willingness to investigate and experiment

Page 7: SHRP2 C10A

7

Additional C10 Insights

Examples of sensitivity tests Linkage vs integration Equilibration and convergence Consistency

Page 8: SHRP2 C10A

8

Freeway Tolling: Demand Impacts

Trips shift out of peaks and midday and into evening and early AM

Higher tolls increases the magnitude of this shift

Time shifting varies by purpose

Work trips shift into early AM and out of AM peak

Social/recreation trips shift significantly out of peaks and primarily into the evening

0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

03:0

0

04:0

0

05:0

0

06:0

0

07:0

0

08:0

0

09:0

0

10:0

0

11:0

0

12:0

0

13:0

0

14:0

0

15:0

0

16:0

0

17:0

0

18:0

0

19:0

0

20:0

0

21:0

0

22:0

0

23:0

0

00:0

0

01:0

0

02:0

0

Work & Soc/Rec Trips by Time of DayBASE-WORK

PRICING_5-WORK

BASE-SOCREC

PRICING_5-SOCREC

-4000

-3000

-2000

-1000

0

1000

2000

3000

4000

03:0

0

04:0

0

05:0

0

06:0

0

07:0

0

08:0

0

09:0

0

10:0

0

11:0

0

12:0

0

13:0

0

14:0

0

15:0

0

16:0

0

17:0

0

18:0

0

19:0

0

20:0

0

21:0

0

22:0

0

23:0

0

00:0

0

01:0

0

02:0

0

Difference in Trips by Time of Day

PRICING_3PRICING_4PRICING_5

Page 9: SHRP2 C10A

9

Travel Demand Management

“Flexible Schedule” scenario Asserted assumptions about:

Fewer individual work activities Longer individual work durations Aggregate work durations

constant

Target: Fulltime Workers

0

1

2

3

4

5

6

7

8

Dur

ation

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

10.0

0

11.0

0

12.0

0

13.0

0

14.0

0

15.0

0

% o

f Tou

rs

Work Tour Duration Distribution

Original

Adjusted

Tours by Purpose (Fulltime Workers)Original Adjusted Adj/Orig

Work 94,408 78,472 0.83School 115 140 1.22Escort 8,070 9,023 1.12Pers Bus 13,519 16,848 1.25Shop 10,531 12,938 1.23Meal 3,817 3,842 1.01Soc/Rec 13,076 14,360 1.10Workbased 27,949 23,211 0.83Total 171,485 158,834 0.93

Page 10: SHRP2 C10A

10

Linkage vs Integration

Establishing linkages, not true integration C10 goal of working with the existing tools and

capabilities Integration may require more fundamental

reformulations “Demand” vs “Supply Models

Demand models as “planning models” – most build schedule a priori, and don’t reflect time-dependency throughout the day

DTA as “dynamic models” Mathematical formulations and behavioral theory

Lack of unifying behavioral theory Differences in formulation and foundations between

demand and supply models. Mathematical formulations should follow behavioral

theory

Page 11: SHRP2 C10A

11

Linkage Challenges

Equilibration & Uniqueness Unclear how to address within the context of

complex simulation tools Relevance to linked, advanced demand and

supply models Relevance to reality?

Need to consider multiple metrics Gap Consistency Stability

Practical issues of network supply runtime

Page 12: SHRP2 C10A

12

Convergence Testing

Convergence Necessary to ensure usefulness

of model system Given the same inputs, will the

model system produce the same outputs?

Can significantly influence the conclusions drawn

Network and system convergence

Extensive testing of different strategies

Network temporal resolution Successive iteration feedback Subselection

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

-

10,000

20,000

30,000

40,000

50,000

60,000

70,000

80,000

90,000

100,000

2 6 10 14 18 22 26 30 34 38 42 46 50 54 58 62 66 70 74 78 82 86 90

Gap

s

VM

T --

VH

T --

Prob

lem

s

Assignment Iteration (N)

TEST_1.13_sqrt-1-over-N (G=3)

MSim problems

Router problems

Select Link Vol

VHT (EQUI) / 10

TripGap

RelGap

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

-

10,000

20,000

30,000

40,000

50,000

60,000

70,000

80,000

90,000

100,000

2 6 10 14 18 22 26 30 34 38 42 46 50 54 58 62 66 70 74 78 82 86 90

Gap

s

VM

T --

VH

T --

Prob

lem

s

Assignment Iteration (N)

TEST_1.2_5min (G=3)

MSim problems

Router problems

Select Link Vol

VHT (EQUI) / 10

TripGap

RelGap

Page 13: SHRP2 C10A

13

Lessons Learned: Application

Level of convergence can significantly influence the conclusions drawn from alternative analyses.

Page 14: SHRP2 C10A

14

Consistency

Convergence not meaningful if there are egregious inconsistencies

Temporal Spatial Typological

Example: demand model employs trip-segmented VOT, but then a single VOT used in network model

Activity models (typically) (Relatively) coarse temporal resolution Typological detail

Dynamic network models (typically) Temporal detail Coarse typological resolution

Page 15: SHRP2 C10A

15

Temporal Consistency

Even if consistent in structure or resolution, there can still be issues with outcome consistency

Ensure that the detailed schedules produced by the DaySim model are maintained in the TRANSIMS network model

Inconsistencies are inevitable – how to resolve

Maintain activity durations or departure times?

Allow supply model to reschedule

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

3:00

4:00

5:00

6:00

7:00

8:00

9:00

10:0

0

11:0

0

12:0

0

13:0

0

14:0

0

15:0

0

16:0

0

17:0

0

18:0

0

19:0

0

20:0

0

21:0

0

22:0

0

23:0

0

0:00

1:00

2:00

Total Schedule Difference by Time-of-Day (Daysim Only)FIXED DEPARTURES: NO FURTHER ADJUSTMENTS

'3.10

'3.20

'3.30

'3.40

'3.50

'3.60

'3.70

'3.80

'3.90

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

3:00

4:00

5:00

6:00

7:00

8:00

9:00

10:0

0

11:0

0

12:0

0

13:0

0

14:0

0

15:0

0

16:0

0

17:0

0

18:0

0

19:0

0

20:0

0

21:0

0

22:0

0

23:0

0

0:00

1:00

2:00

Total Schedule Difference by Time-of-Day (Daysim Only)FIXED DEPARTURES: NO FURTHER ADJUSTMENTS

'3.10

'3.20

'3.30

'3.40

'3.50

'3.60

'3.70

'3.80

'3.90

Base

Spatial Detail

Page 16: SHRP2 C10A

16

Usual work location

Auto ownership

Person-day tour generation

Exact number of tours

Work tour time of day

Work tour mode

WB subtour generation

School tour mode

Other tour destination

Other HB tour time of day

Other HB tour mode

Intermediate stop generation

Intermediate stop location

Trip time of day

0% 20% 40% 60% 80% 100%

significant differenceinsignificant differencenot estimable

Estimated difference between Tampa and Jacksonville coefficient estimates% of coefficients by type of choice model

Transferability

Page 17: SHRP2 C10A

17

alt-specific constant

person characteristic

household characteristic

day-pattern characteristic

tour/trip characteristic

impedance measure

land use measure

time schedule measure

logsum from lower model

0% 10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

significant differenceinsignificant differencenot estimable

Estimated difference between Tampa and Jacksonville coefficient estimates% of coefficients by type of variable

Transferability

Page 18: SHRP2 C10A

18

Future Efforts

Reconsideration of the fundamental “demand-supply” linkage

How can models be more tightly integrated? Can integrated solution methods be defined? Does equilibrium exist in reality, and if not what are the

implications?

How can advanced models be implemented and applied most effectively?