Evaluation of the Effectiveness of Potential ATMIS Strategies Using Microscopic Simulation
-
Upload
kadeem-cook -
Category
Documents
-
view
21 -
download
1
description
Transcript of Evaluation of the Effectiveness of Potential ATMIS Strategies Using Microscopic Simulation
Evaluation of the Effectiveness of Potential ATMIS
Strategies Using Microscopic Simulation
Lianyu Chu, Henry X. Liu, Will Recker
PATH ATMS Center @ UC Irvine
Steve Hague
Traffic operations, Caltrans
Presentation overview
• Background
• Calibration
• ATMIS strategies
• Evaluation studies
• Conclusions
Background
• Caltrans TMS master plan• ATMIS Strategies
– Incident management– Adaptive ramp metering– Adaptive signal control– Traveler information system– Combination / integrated control
I-405 Study network
Scenario description
• northbound of freeway I-405 is highly congested from 7:30 to 8:30 AM
• The merge area of SR-133 and I-405 (on the northbound I-405) is the location where incidents happen most frequently
• Shoulder incident: causes the speed of passing vehicles to be 10 mph for the first ten minutes and 15 mph thereafter
• purpose: evaluate under incident scenario
Calibration: data preparation
– Arterial volume data / cordon traffic counts– Freeway loop detector data– Travel time data– Reference OD matrix (from OCTAM model)– Vehicle performance and characteristics data– Vehicle mix by type
Calibration procedure
• Assumptions– Driver behaviors distribution (awareness and
aggressiveness): normal distribution
– Traffic assignment method: stochastic assignment
• Adjustment of route choice pattern• OD estimation
– Adjustment of the total OD matrix
– Reconstruction of time-dependent OD demands
• Parameter fine-tuning
Adjustmentof route choice pattern
• Route choices: – determined by stochastic assignment, which
calculates shortest path based on speed limits– not affected by traffic signals and ramp
metering (PARAMICS)
• How to adjust:– Adding tolls to entrance ramps– Decreasing the speed limit of arterial links
OD estimation
• an under-defined problem, finding an optimal point in a huge parameter space using limited measurement data
• Our method: two-stage approach– estimation of total OD matrix– profile-based time-dependent OD demands
Total OD matrix (I)
• Reference OD matrix from OCTAM– OCTAM: social-economic data and OD matrix of OC
– sub-extracted OD matrix based on four-step model
– limited to the nearest decennial census year
• Adjustment of the total OD matrix:– traffic counts at all cordon points (i.e. total inbound and
outbound traffic counts )
– balancing the OD table: FURNESS technique
Total OD matrix (II)
• Objective function:– Minimize the difference of estimated traffic flow with
observation – Measurement points: freeway loop stations at on-
ramps, off-ramps and along the mainline freeway, and several important arterial links
– Iterative process: simulation->modify OD->simulation
• overall quality of the calibration: GEH < 5
2/))()((
)()( 2
nMnM
nMnMGEH
simobs
simobs
Time-dependent OD demand (I)
• Most theoretical methods: only apply to simple network
• Our method: profile-based method– Profile: representation of the variation of OD flow
within the whole study time period, which include multiple sample points(16 points)
– Cordon flow (traffic counts): 15-minute interval
– how many vehicles generated from a zone within each interval: profile of the zone
Time-dependent OD demand (II)
• General case: • For any origin i, profile(i, j) = profile(i) , j =1 to N
• Special cases:• If profile can be roughly determined by loop data• If the corresponding OD flow has strong effects on
the traffic condition
– Special OD profiles: • freeway to freeway, • arterial to freeway, • freeway to arterial
Time-dependent OD demand (III)
Destination Origin 1 2 3 4
total_origin (known)
1
2
3
4
Time-dependent OD demand (IV)
0.0%
2.0%
4.0%
6.0%
8.0%
10.0%
12.0%
6:00 6:15 6:30 6:45 7:00 7:15 7:30 7:45 8:00 8:15 8:30 8:45 9:00 9:15 9:30 9:45
Time of day
Per
cen
tag
e o
f to
tal
dem
and
a freeway zone to a freeway zone an arterial zone to an industrial zone
a freeway zone to an arterial zone an artertial zone to a freeway zone
Time-dependent OD demand (V)
• Optimization objectives:– Min (difference between the traffic counts of
simulation and observation over all points and periods)– 85% of the GEH value smaller than 5(during
congestion period: 7:30-8:30AM)
• Iteration is required• Pros: reduction in number of parameter to be
estimated:– 30x30x16 -> 30x16– Totally, 30 profiles in the calibrated model
Parameter fine-tuning
• Link specific parameters• Parameters for the car-following and lane-
changing models• Objective:
– Minimize (observed travel time, simulated travel time)
– Minimize the difference between the traffic counts of simulation and observation over all points and periods
Calibration results (I)
0
100
200
300
400
500
600
700
6:05 6:30 6:55 7:20 7:45 8:10 8:35 9:00 9:25 9:50
405N0.93ml-sim 405N0.93ml-real
0
50
100
150
200
250
6:05 6:30 6:55 7:20 7:45 8:10 8:35 9:00 9:25 9:50
405N1.93ff-sim 405N1.93ff-real
0
200
400
600
800
1000
6:05
6:30
6:55
7:20
7:45
8:10
8:35
9:00
9:25
9:50
405N3.04ml-sim 405N3.04ml-real
0
200
400
600
800
1000
6:05
6:30
6:55
7:20
7:45
8:10
8:35
9:00
9:25
9:50
405N3.86ml-sim 405N3.86ml-real
0
200
400
600
800
1000
6:05
6:30
6:55
7:20
7:45
8:10
8:35
9:00
9:25
9:50
405S3.31ml-sim 405S3.31ml-real
0
50
100
150
200
6:05
6:30
6:55
7:20
7:45
8:10
8:35
9:00
9:25
9:50
133s9.37ml-sim 133s9.37ml-real
Calibration results (II)
Comparison of observed and simulated travel time of northbound I-405
0
100
200
300
400
500
600
6:05
:00
6:20
:00
6:35
:00
6:50
:00
7:05
:00
7:20
:00
7:35
:00
7:50
:00
8:05
:00
8:20
:00
8:35
:00
8:50
:00
9:05
:00
9:20
:00
9:35
:00
9:50
:00
Trav
el t
ime
(sec
)
simulation observation
Calibration results (III)
• The measure of goodness of fit is the mean abstract percentage error (MAPE):
• MAPE error of traffic counts at selected measurement locations range from 5.8% to 8.7%.
• The comparison of observed and simulated point-to-point travel time for the northbound and the southbound I-405, which have the MAPE errors of 8.5% and 3.1%, respectively.
T
tobssimobs tMtMtM
TMAPE
1
))(/))()(((1
ATMIS strategies
• Strategy 1: Incident management– decreasing the response time and clearance time caused
by incidents
• For Caltrans:– no incident management: 33 minutes– existing incident management: 26 minutes– improved incident management: 22 minutes
ATMIS strategies
• Strategy 2: Ramp metering– an effective freeway management strategy to avoid or
ameliorate freeway traffic congestion by limiting vehicles access to the freeway from on-ramps.
• Current implemented ramp metering: fixed-time• Potential improvement: adaptive ramp metering
– local adaptive ramp metering– coordinated ramp metering
ATMIS strategies: ramp metering
• ALINEA: a local feedback ramp metering policy
• maximize the mainline throughput by maintaining a desired occupancy on the downstream mainline freeway.
Downstream detector
On-ramp detector
Queue detector
))(*()(~)( tOOKttrtr R
ATMIS strategies:ramp metering
• BOTTLENECK, coordinated ramp metering• applied in Seattle, Washington State• Two components:
– a local algorithm computing local-level metering rates based on local conditions,
– a coordination algorithm computing system-level metering rates based on system capacity constraints.
– the more restrictive rate will obey further adjustment
• within the range of the pre-specified minimum and maximum metering rates
• queuing control
ATMIS strategies
• Strategy 3: travel information– all kinds of traveler information systems, including
VMS routing, highway radios, in-vehicle equipment, etc.
– pure traveler information system: no traffic control supports
– how to model in PARAMICS: using dynamic feedback assignment
– assumptions: instantaneous traffic information is used for the calculation of the resulting route choice
ATMIS strategies
• Strategy 4: advanced signal control– adaptive signal control, and
– signal coordination
• Actuated signal coordination: – baseline situation: 11 signal intersections in the study
network are coordinated
• Adaptive signal control: – use SYNCHRO to optimize signal timing of those signals
along major diversion routes during the incident period based on estimated traffic flow
Evaluation: Modeling ATMIS strategies
ATMIS components Scena-
rio Scenario description Ramp Metering Signal Control Traveler Information
Incident Management
0 BASELINE 2000 Fixed time Coordinated N/A N/A
1 Non-incident management Fixed time Coordinated N/A 33 mins
2 Existing incident management Fixed time Coordinated N/A 26 mins
3 Improved incident management Fixed time Coordinated N/A 22 mins
4 Local adaptive ramp metering ALINEA Coordinated N/A 26 mins
5 Coordinated ramp metering BOTTLENECK Coordinated N/A 26 mins
6 Traveler information Fixed time Coordinated 5% compliance 26 mins
7 Combination-1 Fixed time Synchro-Adaptive 5% compliance 26 mins
8 Combination-2 ALINEA Synchro-Adaptive 5% compliance 26 mins
Evaluation: MOEs (I)
• MOE #1 system efficiency measure: average system travel time (weighted mean OD travel time over the whole period)
• MOE #2 system reliability measure: weighted std of mean OD travel time over the whole period
ji
jiji
jiji NNTASTT,
,,
,, )(
ji
jiji
jiji NNTStdODTTStd,
,,
,, ))((_
Evaluation: MOEs (II)
• MOE #3 freeway efficiency measure: average mainline travel speed during the whole period and during the congestion period(7:30-9:30)
• MOE #4 on-ramp efficiency measure– total on-ramp delay– average time percentage of the on-ramp queue spillback
to the local streets
• MOE #5 arterial efficiency measure– average travel time from the upstream end to the
downstream end of an arterial and its std
Evaluation: number of runs
N
Y
Original nine runs
Start
Calculating the mean and its std of each performance measure
Is current # of runs enough?
End
Calculating the required # of runs for each performance measure
Additional one simulation run
22/ )(
tN
Evaluation results (I): overall performance
Control strategy ASTT (sec) ASTT Saving (%) std_ODTT (sec) Reliability Increase
(%)
Baseline 271.3 51.7
IM-33 297.0 0.0% 139.6 0.0%
IM-26 293.9 1.0% 130.7 6.4%
IM-22 289.1 2.7% 112.6 19.4%
ALINEA 289.7 2.4% 118.9 14.9%
BOTTLENECK 289.2 2.6% 115.5 17.3%
TI 284.4 4.2% 95.3 31.8%
Combination-1 280.5 5.5% 93.2 33.3%
Combination-2 279.6 5.9% 97.2 30.4% ASTT – Average system travel time Std_ODTT— Average standard deviation of OD travel times of the entire simulation period, which represents the reliability of the network
Evaluation results (II):Freeway performance
Scenario AMTS (mph)
AMTS Increase (%)
peak_AMTS (mph)
Increase of peak_AMTS
TOD (hour)
POQS (%)
Baseline 57.3 50.1 55.1 1.8%
IM-33 50.5 0.0% 37.2 0.0% 55.6 1.9%
IM-26 51.4 1.8% 39.4 6.0% 54.6 2.0%
IM-22 51.9 2.8% 40.0 7.5% 54.0 1.8%
ALINEA 51.6 2.1% 39.8 6.9% 57.6 0.9%
BOTTLENECK 51.9 2.7% 39.7 6.7% 89.1 1.9%
TI 51.9 2.8% 39.9 7.3% 58.0 1.8%
Combination-1 52.2 3.3% 41.0 10.1% 59.5 1.9%
Combination-2 52.3 3.5% 40.6 9.1% 60.0 1.0% AMTS – Average mainline travel speed of the entire simulation period (6 – 10 AM) peak_AMTS – Average mainline travel speed of the congestion period (7:30 – 9:30) TOD – Total on-ramp delay POQS – Time percentage of vehicles on the entrance ramps spillback to surface streets
Evaluation results (III):Arterial performance
Westbound ALTON Scenario ATT (sec) std_ATT
Baseline 515.8 70.3
IM-33 515.5 71.0
IM-26 514.1 68.1
IM-22 512.4 68.1
ALINEA 513.6 67.3
BOTTLENECK 518.3 69.0
TI 518.8 70.2
Combination-1 423.5 51.4
Combination-2 423.2 51.0 ATT – Average travel time Std_ATT – Standard deviation of the average travel time
Evaluation results (IV): IM
• Incident management– fast incident response is of particular
importance to freeway traffic management and control
– To achieve this, comprehensive freeway surveillance system and automatic incident detection are both required
Evaluation results (V): ramp metering
• performance improvement introduced by adaptive ramp metering is minor under the incident scenarios
• If the congestion becomes severe, the target LOS could not be maintained by using ramp metering and the effectiveness of ramp control is marginal
• adaptive ramp metering performs worse than the improved incident management scenario
• BOTTLENECK performs a little bit better than ALINEA in term of overall performance, but, BOTTLENECK causes higher on-ramp delay and spillback.
Evaluation results (VI): TI related scenarios
• traveler information– network topology -- one major freeway segment (I405)
with two parallel arterial streets – traveler information systems can greatly improve
overall system performance• Adaptive signal control:
– shorter travel time along diversion route (westbound ALTON parkway)
• Combination scenarios: perform the best– integration of traffic control & traveler information
Conclusions
• Evaluate the effectiveness of potential ATMIS strategies in our API-enhanced PARAMICS environment.
• Findings:– All ATMIS strategies have positive effects on the
improvement of network performance. – Adaptive ramp metering cannot improve the system
performance effectively under incident scenario.– Real-time traveler information systems have the strong
positive effects to the traffic systems if deployed properly
– Proper combination of ATMIS strategies yields greater benefits.