FHWA C10 DynusT+CT-RAMP Integration on ARC and MORPC

C10 Projects
CT-RAMP-DynusT Integration
Analysis of results, model system
performance, and readiness for
practice
1ABM/DTA integrated models in Ohio and Atlanta, June 7, 2017

OVERVIEW OF 2 PROJECTS
Commonality and differences between ARC and MORPC applications
ABM/DTA integrated models in Ohio and Atlanta, June 7, 2017 2

2 Parallel applications
Columbus, OH (MORPC)
• 1.4M population
• 2,000 TAZs
• 18,000 MAZs
• 10,000 links
• CT-RAMP2 ABM
• DynusT daily simulation of
6M vehicles
Atlanta, GA (ARC)
• 5.0M population
• 5,873 TAZs
• No MAZs currently
• 50,000 links
• CT-RAMP1 ABM
• DynusT daily simulation of
20M vehicles

Hardware Information
• ARC:
– RAM: 448GB of RAM – New computer will have 1TB
– Processors: 3.2 GHz CPU, 48 physical cores
• MORPC:
– RAM: 256GB – Considering upgrading to 1TB
– Processors: Dual 18 Intel Xeon E5 2699 2.3GHz 36 core processors
• Model runs faster without hyper-threading (i.e. run only 36 threads)
• WSP:
– ARC runs:
• 128 GB RAM
• 2 x Intel® Xeon® CPU E5-2670 0 @ 2.60GHz
– Each CPU with 8 physical, 16 virtual cores
– MORPC runs:
• 128 GB RAM
• 2 x Intel® Xeon® CPU E5-2667 0 @ 2.90GHz
– Each CPU with 6 physical, 12 virtual cores

Highlights on ABMs
• CT-RAMP1 (ARC):
– Fully disaggregate micro-simulation of HHs & persons
– Tour-level consistency but based on primary destination,
stops inserted later
– Intra-household interactions
– Temporal resolution of 30 min (required post-processing
for DTA)
• CT-RAMP2 (MORPC):
– Multi-destination tour formation and combinatorial mode
choice
– Continuous time (was designed to work with DTA
seamlessly)

Setting of Individual Schedule
Adjustment Module (iSAM)
• Mandatory activities (work and school):
– High penalties for arriving late or departing early
– Longer travel times are accommodated:
• By departing from home earlier in outbound direction
• By arriving back home later in inbound direction
• Non-mandatory activities:
– Relatively low penalties for arriving late or departing early
– Substantial penalties for shortening duration
– Longer travel times are accommodated:
• In flexible ways depending on the tour structure
• By stretching the active time window at the expense of in-home
activities
• Possible to explore scenarios with more flexibility for work
– (Ohio is asking about schedule flexibility in the HTS)

Current Performance
Model component MORPC ARC
PopSyn (P) 2 hours (MAZs) 1 hour (TAZs)
Core ABM iteration (A) 4 hours (multi-threaded) 6 hours (distributed)
DynusT iteration (D) 30 min 3 hours
Trajectory processing (T) 2 hours 6 hours
Schedule adjustment (S) 10 min 30 min
Targeted production run
P+3A+3T+3×3S+3×3×5D
2+3×4+3×2+9×0.2+45×0.5 =
42 hours
1+3×6+3×6+9×0.5+45×3
= 177 hours
Implemented full runs for this
report for 1 scenario
3A+3T+3×3S+3×3×3D
3×4+3×2+9×0.2+27×0.5 = 34
hours
Implemented internal loop for
this report for 1 scenario 3S+40D
3×0.5+40×3=122 hours
Comparable DTA run for this
report w/fixed demand for 1
scenario 40D
40×3=120 hours

DTA Runtime Sensitivity to Congestion
• Different from ABM, DTA runtime per iteration is
sensitive to network conditions:
– Base scenario average ARC DynusT runtime per iteration:
• Simulation: 100 min
• Routing: 85 min
– I-85 Bridge closure scenario average ARC DynusT runtime
per iteration:
• Simulation: 240 min
• Routing: 100 min
• Application of internal loop with schedule adjustment
helped improve runtime:
– I-85 Bridge closure DynusT runtime became close to Base

CT-RAMP2 specifics
• Common features of most ABMs in practice:
– Tour as main unit of modeling (travel generation)
– Micro-simulation of individual persons & HHs
– Explicit intra-HH interactions
– Enhanced spatial resolution (MAZ)
• Unique features of CT-RAMP2:
– Explicit activity generation followed by tour formation
(long-term mobility pattern changes)
– Continuous time and seamless integration with DTA
– Consistency of individual patterns within time-space
constraints (towards AgBM)

CT-RAMP2 blocks
1. Population synthesizer w/university sub-model
2. Long-term mandatory activity arrangements
3. Mid-term mobility attributes
4. Special events
5. Activity generation & tour formation
6. Tour TOD
7. Tour & trip mode choice
8. Within-tour time allocation and trip departure

Cube Interface of MORPC ABM
Demo for Lima

Key Features of CT-RAMP2 Essential
for Integration with DTA
• High level of temporal resolution:
– 15 min for tour schedules
– Continuous for trip departure time
• High level of spatial resolution for individual trip loading:
– 2,000 TAZs broken into 18,000 MAZs
• Seamless translation of person trips into vehicle trips:
– Extended set of intra-household interactions and explicit
modeling of joint travel
– Identification of auto driver and passenger roles for joint trips
• LOS manager that works with Accumulated Data Bank of
Individual Trajectories (ADIT) instead of static skims

DynusT Set Up & Network Information

• Cube network was saved to shape file and
imported into DynusT
• MORPC - Cube junction files were available for
controlled intersections. These files were
used to generate DynusT intersection files.
• ARC – Synchro intersection files were used to
generate DynusT intersection files.

• Time dependent lanes can be set up as “work zones”

• Can specify:
– Start and end time by minute
– Speed limit changes (if any)
– Queue discharge rate (saturation flow rate) per lane
– Capacity reduction rate (code max cap and reduce)

Integration Layer
Components
• External Loop 1
w/mining
individual
trajectories LOS
• Internal Loop 2
w/individual
Schedule
Adjustment
Module (iSAM)
17
Microsimulation ABM
(CT-RAMP)
Microsimulation DTA
(Dynus-T)
List of
individual
vehicle
trips &
tours
Individual
trajectories
for the
current list of
trips
ISAM (Individual Schedule
Adjustment Module):
Consolidation of individual
schedules (inner loop 2 for
departure time corrections)
Sample of alternative origins,
destinations, and departure times
Individual
LOS
estimates for
potential trips
List of
individual
person trips
& tours by
mode
Transit microsimulation
(Fixed LOS)
List of
individual
transit
trips &
tours
ADIT
(Accumulated
Database of
Individual
Trajectories):
LOS Mining
Integration Layer 2
Loop1Loop2
Travel
“stress”
evaluation
measures
Sample of
households
for re-
modeling all
choices
Integration Layer 1
ABM/DTA integrated models in Ohio and Atlanta, June 7, 2017

Essence of Each Loop
• External Loop 1:
– Generates activity patterns & schedules
– Uses individualized LOS through trajectory mining
• Internal Loop 2:
– Simulates activity patterns
– Adjusts schedules for realistic trip chain loading
– Uses individual trajectories
– Evaluates “stress” measures

ARC SCENARIOS
Results, analysis, and performance of internal loop

4 Scenarios
• Base DynusT with fixed demand
• Base DynusT+iSAM (schedule adjustment)
• I-85 Bridge closure DynusT with fixed demand
• I-85 Bridge closure DynusT+iSAM (schedule
adjustment)

Regional DTA Network Development
• Mesoscopic is a Scale, DTA is a Method
• Start DTA Network Development with:
• a Commercially Available Street Centerline File (NAVTEQ)
• www.openstreetmap.org is another option
• Start with your Regional Travel Demand Model Network
• Network Consolidation (Black Links were Merged)

ARC Regional DTA Network
with Signalized Intersections
http://arcg.is/290ivBS

NAVTEQ DTA Signalized Intersections
NAVSTREETS - Green
Additional - Red

Atlanta Metro Signalized Intersections

Atlanta Metro Ramp Meters in DTA Network

Atlanta Metro Regional DTA
Estimation & Calibration
Traffic Flow Models
GP
HOV
26

Coordinated Signals on Major Atlanta Corridors
Synchro UTDF

Atlanta RTOP Corridors (34)
Regional Traffic Operation Program
SR10D
SR10G
SR12
SR120
SR138E
SR138S
SR139
SR140G
SR141N
SR141S
SR154A
SR154D
SR155N
SR155S
SR20
SR237
SR280
SR38E
SR3NA
SR3NC
SR3S
SR42
SR5
SR6
SR85
SR8A
SR8CF
SR8D
SR8DF
SR8G
SR92&14
0R
SR92C
SR9N
SR9S

State Route - 42 South Bound
INRIX vs. DTA Speeds

Synchro Files & ABM-DTA Integration
• Intersection Geometry (Left & Right Turn Lanes)
• Traffic Signal Timing & NEMA Phasing
• Green Times, Amber times, Offsets

Overall Scenario Comparison
Scenario Average trip time Average delay Unfinished trips
Base DynusT
w/fixed demand
25.81 min 2.90 min 0
Base DynusT
w/iSAM
24.63 min 2.64 min 0
I-85 Bridge closure
DynusT w/fixed
demand
33.89 min 4.24 min 38,728
I-85 Bridge closure
w/iSAM
30.99 min 3.73 min 26,151

Base DynusT w/Fixed
Demand

Base DynusT w/Fixed
Demand: Convergence
Inner
Loop
Iter Run type
DynusT
UE iters
Upper limit vehicle
departs
Simulation
horizon
iSAM LP
Failures
Vehicles Still in the
Network Relative Gap
minutes clock
First
UE iter
Final
UE iter First UE iter Final UE iter
N/A cold 30 1620 3:00 AM 2000 N/A 774404 0 0.8409773 0.1104254
N/A cold 40 1620 3:00 AM 2000 N/A 774404 0 0.8409773 0.117388

Base DynusT+iSAM

Base DynusT+iSAM:
Convergence
Inner
Loop
Iter Run type
DynusT
UE iters
Upper limit vehicle
departs
Simulation
horizon
iSAM LP
Failures
minutes clock
First
UE iter
Final
0 cold 15 1620 3:00 AM 2000 0 774404 0 0.8409773 0.1438067
1 warm 15 1620 3:00 AM 2000 0 0 0 0.1585612 0.1058591
2 warm 5 1620 3:00 AM 2000 0 0 0 0.1155581 0.1099977
3 warm 5 1620 3:00 AM 2000 0 0 0 0.1178765 0.1075283
iSAM improves DTA convergence as expected

Base DynusT+iSAM: Adjustment
of Departure Time to Work

I-85 Bridge Closure DynusT
w/Fixed Demand

I-85 Bridge Closure DynusT
w/Fixed Demand: Convergence
Inner
Loop
Iter Run type
DynusT
UE iters
Upper limit vehicle
departs
Simulation
horizon
iSAM LP
Failures
minutes clock
First
UE iter
Final
NA cold 30 1620 3:00 AM 2000 0 741658 41013 0.8399832 0.305708
NA cold 40 1620 3:00 AM 2000 0 741658 38728 0.8399832 0.312108
Convergence w/fixed demand under critical network conditions is problematic

I-85 Bridge Closure
DynusT+iSAM

I-85 Bridge Closure
DynusT+iSAM: Convergence
Inner
Loop
Iter Run type
DynusT
UE iters
Upper limit vehicle
departs
Simulation
horizon
iSAM LP
Failures
minutes clock
First
UE iter
Final
0 cold 15 1620 3:00 AM 2000 0 741658 37381 0.8399832 0.3215604
1 warm 15 1620 3:00 AM 2000 726 31360 25272 0.3594274 0.2693154
2 warm 5 1620 3:00 AM 2000 415 21806 26065 0.2974289 0.2903202
3 warm 5 1620 3:00 AM 2000 415 24161 26151 0.314918 0.2907585
iSAM improves DTA convergence as expected although more schedule relaxations needed

I-85 Bridge Closure DynusT+iSAM:
Adjustment of Departure Time to Work

Base DynusT vs. Base DynusT+iSAM:
Research Questions
• What schedule adjustments would be
necessary?
• How would schedule adjustments affect
convergence?
• How would schedule adjustments affect the
network loading and other performance
measures?
• How would schedule adjustments affect
validation against traffic counts?

Traffic Counts (ADT) Correlation Analysis –
Base DynusT with Fixed Demand

Traffic Counts (ADT) Correlation
Analysis – DynusT+iSAM

Base DynusT w/fixed demand Base DynusT w/iSAM
•Minimal rescheduling needed for base
•Convergence somewhat better with iSAM

ARC Base Network
Daily Volume Difference: iSAM vs. Fixed Demand

ARC Base Network
Daily Volume Difference%: iSAM vs. Fixed Demand

ARC Base Network
AM Volume Difference: iSAM vs. Fixed Demand

ARC Base Network
AM Volume Difference%: iSAM vs. Fixed Demand

ARC Base Network
PM Volume Difference: iSAM vs. Fixed Demand

ARC Base Network
PM Volume Difference%: iSAM vs. Fixed Demand

Observations on Impact of Schedule
Adjustments for Base Scenario
• Overall minor differences in assigned volumes at daily
level:
– Some local impacts on certain facilities in congested
corridors are stronger
• Much more substantial impact in AM period:
– Minor rescheduling allows for accommodation of more
traffic in the most congested corridors
• Less prominent effect in PM period:
– Departure from work is largely fixed in current setting of
iSAM
– Rescheduling and spreading of non-work trips allowed for
accommodation of somewhat more traffic in congested
corridors

Base DynusT vs. Bridge Closure
DynusT: Research Questions
• How would DTA with fixed demand perform
under critical network conditions?
• How realistic would be the results?
• How substantial is the network redundancy in
Atlanta to resolve the capacity drop by means
of route choice only?

Base DynusT w/fixed demand I-85 Bridge closure DynusT w/fixed demand
•Rescheduling of departure time is not allowed
•Huge change in volume profiles with many vehicles stuck in the network
•Probably unrealistic simulation of bridge closure

Base DynusT+iSAM vs. Bridge Closure
DynusT+iSAM
• How would DTA with schedule adjustments
perform under critical network conditions?
• How realistic would be the results?
• How substantial is the network redundancy in
Atlanta to resolve the capacity drop by a
combined effect of route choice and
departure time choice?

Base DynusT+iSAM I-85 Bridge closure DynusT+iSAM
•Substantial departure time shift observed
•Substantial change in volume profiles but with fewer vehicles stuck in the network
•More realistic simulation of bridge closure but may still need more schedule adjustments

Bridge Closure DynusT vs. Bridge
Closure DynusT+iSAM
• How would schedule adjustments make user
response to critical capacity drop more
realistic?
• How would schedule adjustments affect DTA
convergence under critical conditions?

I-85 Bridge closure DynusT w/fixed demand I-85 Bridge closure DynusT+iSAM
•Substantial departure time shift and peak spreading observed
•Substantial change in volume profiles but with fewer vehicles stuck in the network
•More realistic simulation of bridge closure but may still need more schedule adjustments

ARC I-85 Bridge Closure
Daily Volume Difference: iSAM vs. Fixed Demand

Daily Volume Difference%: iSAM vs.Fixed Demand

PM Volume Difference: iSAM vs. Fixed Demand

PM Volume Difference%: iSAM vs. Fixed Demand

Observations on Model Response to
Bridge Closure
• DTA with fixed demand creates an unrealistic
pattern with infeasible levels of congestion:
– Rerouting by itself is not enough due to limited
network redundancy
• Allowing for schedule adjustments makes
simulation more realistic:
– Many departure time shifts to earlier or later
hours to avoid the highest congestions

MORPC SCENARIOS
Results, analysis, and performance of internal & external loops

Important Research Questions when LOS Skims
are Replaced with Individual Trajectories
• Would the trajectories from several DTA
iterations be enough to cover the need for LOS
for ABM?
• How good would be the match between the
individual trips and trajectories?
• Do we still need aggregate skims to fill the gaps?
• How different are travel times from DTA
compared to static assignment?
• Would the ABM-DTA integrated model require a
complete recalibration compared to standard
ABM?

Trajectory Coverage Stats
68
TOD
Aggregation level
1 2 3 4 9 Total
Before 6 83.0% 12.5% 0.2% 0.1% 4.2% 100.0%
6-10 61.3% 5.6% 19.5% 7.6% 5.9% 100.0%
10-15 93.4% 5.7% 0.1% 0.1% 0.7% 100.0%
15-19 66.6% 5.9% 17.2% 6.2% 4.2% 100.0%
After 19 92.0% 6.9% 0.1% 0.0% 0.9% 100.0%
Total 77.7% 6.1% 9.6% 3.6% 3.0% 100.0%
ABM/DTA integrated models in Ohio and Atlanta, June 7, 2017
1. MAZ to MAZ, 5 minute congested resolution
2. MAZ to MAZ, 15 minute congested
resolution
3. TAZ to TAZ, 15 minute congested resolution
4. TAZ to TAZ, 60 minute congested resolution
9. Use TAZ to TAZ skim

Travel Time Differences by aggLevel:
Trajectory-Skim, min
0%
5%
10%
15%
20%
25%
30%
35%
40%
aggLevel1
aggLevel2
aggLevel3
aggLevel4

Travel Time Differences by TOD:
Trajectory-Skim, min
0%
5%
10%
15%
20%
25%
30%
35%
40%
early
am
midday
pm
night

Example of Analysis of Time Budgets

IR-670 EB Hard Shoulder
Running Lane
3:30PM – 6:30PM

MORPC Diurnal Profiles:
Base DynusT w/Fixed Demand

MORPC Diurnal Profiles:
IR-670 HSR Lane w/Fixed Demand

Daily Volume Difference:
HSR vs. Base w/Fixed Demand

DynusT+iSAM
• Next two slides compare two network
scenarios with schedule adjustments allowed:
– HRS reduced average travel time
– Less schedule adjustments needed
– Overall very similar profiles

Impact of DTA on Mode Choice
• Useful constrained exercise included
equilibration of the following 3 components:
– ABM mode choice only
– iSAM
– DTA
• It provides a pure impact of substitution of
static LOS skims with DTA trajectories:
– Trip list by all modes stays the same
– Mode switches can be analyzed at individual level

Iter. 1 (Trajectories) vs.
Iter. 0 (Skims)
1=SOV
2=HOV2-
driv
3=HOV3-
driv
4=HOV-
pass.
5=WT-
Local Bus.
6=KNR-
Local Bus.
7=PNR-
Local Bus.
8=WT-
Express
bus
9=KNR-
Express
bus
10=PNR-
Express
bus
14=Walk 15=Bike 16=Taxi
17=School
bus
1=SOV 3,093,373 7 4 2,672 4,255 524 644 283 108 311 5,306 398 2 - 14,514
2=HOV2-driv 6 794,805 - 546 452 144 119 41 20 41 796 50 - - 2,215
3=HOV3-driv 7 - 677,268 344 151 63 28 9 2 4 863 21 - - 1,492
4=HOV-pass. - - - 1,710,546 65 27 1 7 1 2 1,507 8 - - 1,618
5=WT-Local
Bus.
- - - 3 82,035 - - - - - 2 - - - 5
6=KNR-Local
Bus.
- - - - - 8,012 - - - - - - - - -
7=PNR-Local
Bus.
12 - 4 1 7 - 4,037 1 - 1 1 - - - 27
8=WT-Express
bus
- - - - - - 1 3,542 - - - - - - 1
9=KNR-
Express bus
- - - - - - - - 875 - - - - - -
10=PNR-
Express bus
- - - - 4 - - 1 - 1,189 - - - - 5
14=Walk - - - - - - - - - - 410,000 1 - - 1
15=Bike - - - - - - - - - - - 18,845 - - -
16=Taxi - - - - - - - - - - - - 33,956 - -
17=School bus 78 13 5 42 3 1 2 - - - 34 7 - 272,338 185
Gain 103 20 13 3,608 4,937 759 795 342 131 359 8,509 485 2 -
Loss
Mode 1
Mode 0

Mode Gain and Loss by Switching from
Static Skims to Dynamic Trajectories
(20,000)
(15,000)
(10,000)
(5,000)
-
5,000
10,000
Mode Gain & Loss: Iteration 1 vs. Iteration 0
Gain
Loss

% Mode Gain and Loss
(iter. 1 vs. iter. 0)
-25%
-15%
-5%
5%
15%
25%
35%
% Mode Gain & Loss: Iteration 1 vs. Iteration 0
Gain
Loss

Iter. 2 (Trajectories) vs Iter. 1
(Trajectories)
1=SOV
2=HOV2-
driv
3=HOV3-
driv
4=HOV-
pass.
5=WT-
Local Bus.
6=KNR-
Local Bus.
7=PNR-
Local Bus.
8=WT-
Express
bus
9=KNR-
Express
bus
10=PNR-
Express
bus
17=School
bus
1=SOV 3,091,144 8 3 567 443 102 96 52 28 59 868 91 - 15 2,332
2=HOV2-driv 6 794,471 - 110 38 29 20 2 3 7 131 5 - 3 354
3=HOV3-driv 2 - 677,060 51 21 13 12 - - 2 114 6 - - 221
4=HOV-pass. 2,505 498 326 1,710,564 7 9 1 - 1 - 229 1 - 13 3,590
5=WT-Local
Bus.
4,070 425 146 30 82,289 - 7 - - 4 1 - - - 4,683
6=KNR-Local
Bus.
508 141 63 22 - 8,037 - - - - - - - - 734
7=PNR-Local
Bus.
610 112 26 2 1 - 4,080 1 - - - - - - 752
8=WT-Express
bus
273 40 9 2 - - 1 3,558 - 1 - - - - 326
9=KNR-
Express bus
97 17 2 1 - - - - 889 - - - - - 117
10=PNR-
Express bus
293 41 4 2 - - 1 2 - 1,205 - - - - 343
14=Walk 4,304 684 754 1,360 1 - - - - - 411,397 1 - 8 7,112
15=Bike 360 49 17 4 - - - - - - - 18,900 - - 430
16=Taxi 2 - - - - - - - - - - - 33,956 - 2
17=School bus 9 4 2 7 - - - - - - 5 2 - 272,309 29
Gain 13,039 2,019 1,352 2,158 511 153 138 57 32 73 1,348 106 - 39
LossMode 1
Mode 2

Mode Gain and Loss
(Iter. 2 vs. Iter. 1)
(15,000)
(10,000)
(5,000)
-
5,000
10,000
15,000
Mode Gain & Loss: Iteration 2 vs. Iteration 1
Gain
Loss

% Mode Gain and Loss
(Iter 2. vs. Iter. 1)
-25%
-15%
-5%
5%
15%
25%
35%
% Mode Gain & Loss: Iteration 2 vs. Iteration 1
Gain
Loss

Iteration 0 vs Iteration 2
1=SOV
2=HOV2-
driv
3=HOV3-
driv
4=HOV-
pass.
5=WT-
Local Bus.
6=KNR-
Local Bus.
7=PNR-
Local Bus.
8=WT-
Express
bus
9=KNR-
Express
bus
10=PNR-
Express
bus
17=School
bus
1=SOV 3,104,096 4 4 735 627 119 126 62 39 77 1,869 129 - - 3,791
2=HOV2-driv 1 796,472 - 159 64 32 27 3 6 7 243 6 - - 548
3=HOV3-driv 6 - 678,397 68 26 13 14 - - 2 224 10 - - 363
4=HOV-pass. - - - 1,711,720 44 13 - 5 1 - 376 5 - - 444
5=WT-Local
Bus.
- - - 3 82,035 - - - - - 2 - - - 5
6=KNR-Local
Bus.
- - - - - 8,012 - - - - - - - - -
7=PNR-Local
Bus.
8 - 4 1 1 - 4,049 - - - 1 - - - 15
8=WT-Express
bus
- - - - - - - 3,543 - - - - - - -
9=KNR-
Express bus
- - - - - - - - 875 - - - - - -
10=PNR-
Express bus
- - - - - - - 2 - 1,192 - - - - 2
14=Walk - - - - - - - - - - 409,999 2 - - 2
15=Bike - - - - - - - - - - - 18,845 - - -
16=Taxi - - - - - - - - - - - - 33,956 - -
17=School bus 72 14 7 36 3 1 2 - - - 31 9 - 272,348 175
Gain 87 18 15 1,002 765 178 169 72 46 86 2,746 161 - -
LossMode 0
Mode 2

Mode Gain and Loss
(15,000)
(10,000)
(5,000)
-
5,000
10,000
15,000
Mode Gain & Loss: Iteration 0 vs Iteration 2
Gain
Loss

% Gain and Loss
-25%
-15%
-5%
5%
15%
25%
35%
% Mode Gain & Loss: Iteration 0 vs Iteration 2
Gain
Loss

Observations on Impact of DTA on
Mode Choice
• Overall a well-calibrated ABM does not suffer a stress
from switching to DTA
• No substantial recalibration needed
• Most shifts are from auto modes to transit and non-
motorized in the 1st iteration:
– More extreme congestion for certain auto trips compared
to static skims
• The opposite equilibration shift from transit and non-
motorized modes to auto in the 2nd iteration:
– Relative congestion relief in the second DTA application

Convergence

Observations on convergence
• Schedule consistency and stability are improved over
internal iterations and also between the global
iterations although each global iteration (ABM) starts
with a “stress” due to a new demand
• Stressed schedules are improved over internal
iterations but not between the iteration 0 and 1 where
the main change of LOS (trajectories vs. skims) occur
• More global iterations needed to analysis convergence
but it is time taking and was not ready by the webinar
time

DISSEMINATION OF PRODUCTS
Results, analysis, and performance of internal & external loops

Conclusions
• Deep integration of ABM and DTA is feasible:
– Already practical for regions under 1M
• Many additional new avenues:
– Moving towards AgBM
• Runtime is an issue:
– Integration layer adds only a little
– DTA and ABM constitute major time-consuming
components, especially DTA for large regions

Two Ways How Products of this Project
Could be Used in Practice
Complete system transfer
• MPO will have to adopt and
provide all inputs for:
– CT-RAMP2
– DynusT
• Complete integration layer will
be transferred:
– iSAM
– ADIT
– Data exchange
• Given CT-RAMP2 and DynusT
are up and running:
– Possible in 3 months under
100K
Modular use of integration layer
• MPO can have different starting
components:
– Their own ABM, not necessarily CT-
RAMP
– Their Own DTA, not necessarily
DynusT
• Main integration layer
components are generic and
transferable:
– iSAM
– ADIT
• System integration and data
exchange between 4 components
will have to be implemented:
– Possible in 6 months under 300K

Coming Soon - Lima
• Ohio’s implementation was developed with
transferability in mind (Cleveland, Columbus,
Cincinnati/Dayton)
• A test model was made for Lima
– Mostly because it runs faster and hence is easier to
test and debug
• Lima will be made available (for free!) this fall
– Base implementation requires Cube and
DynusT/DynuStudio
– ABM can read TransCad and Emme files as well, but
requires a different interface/set-up

Strawberry Rhubarb Pie
Crust
1 stick butter (slightly softened), cut into
pats
135g flour (~1 cup)
1 tsp. sugar
½ tsp. salt
1.5 Tbsp. ice water
Put first 4 ingredients into the bowl of a
stand mixer. With the dough hook, run the
mixer on medium-low speed until the
butter is mostly cut into the flour. Add the
ice water and continue mixing until the
dough comes together into a ball. Wrap in
plastic wrap and refrigerate.
Filling
~2 c. chopped rhubarb
~2 c. sliced strawberries
Scoop and a half of sugar (~2/3 cup)
Medium scoop of flour (~1/3 cup)
Some vanilla (~1 tsp.)
5 pats of butter
Stir first 5 ingredients together, making sure
that everything gets coated in sugar/flour.
Roll out the dough disk in a generous
amount of flour and slide it into a pie pan.
Add filling. Dig butter pats into the filling.
Crumble the topping onto the top. Bake at
425 degrees for 15 minutes. Reduce heat
to 375 degrees and bake for another 30
minutes.
Topping
1 c. flour, 1 c. brown sugar, 1 Tbsp. cinnamon,
1 tsp. allspice or nutmeg, ½ c. melted butter
Mix dry ingredients well. Add melted butter
and mix with a spoon until it’s all moist.

Contacts
• Guy Rousseau
– GRousseau@atlantaregional.com
• Rebekah Straub Anderson
– rebekah.anderson@dot.ohio.gov
• Peter Vovsha
– Peter.Vovsha@wsp.com
• Robert Tung
– robert.tung@metropia.com

FHWA C10 DynusT+CT-RAMP Integration on ARC and MORPC

More Related Content

Similar to FHWA C10 DynusT+CT-RAMP Integration on ARC and MORPC (20)

Recently uploaded (20)

FHWA C10 DynusT+CT-RAMP Integration on ARC and MORPC