An Analytical Framework for Managed Lane Facility Performance Evaluation
Xiaoyue Cathy Liu
University of Washington
To Be Presented at the Western ITE Annual Meeting in Santa Barbara, CA
June 26th, 2012
Outline
• General Background• Methodological Framework • Cross-Weave Module• Friction Module• Summary
2
Project Overview – Why Managed Lanes?
• Congestion • A total urban congestion cost of $101 billion and total delay
hours of 4.8 billion hours in 2010
• Vehicle Miles Traveled vs. Roadway Capacity Increase
• Limited Infrastructure Expansion Capability
3
4
Background
• NCHRP 03-96 Project: Analysis of Managed Lanes on Freeway Facilities
• A methodological framework is needed for analyzing freeway facilities with ML and GP lanes operated parallelly• Composition and behavior characteristics difference• Interaction between the two lane groups
5Managed Lane Characteristics
• Operational Strategy: HOV vs. HOT• Separation Type
• Barrier• Buffer• Stripe
Image: I-394 MnPass WA SR 167 HOT
Image: SR 91 Express Lanes
6
Managed Lane Access
Type A: At-Grade Lane-Change Access
Type B: At-Grade Ramp Access
Type C: Grade-Sep. Ramp Access
• The most common ML type is left-concurrent
7Single vs. Multiple Lanes
• Single ML Lane – Inability to pass slow lead vehicle
0
10
20
30
40
50
60
70
80
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Spee
d (m
ph)
Flow (pc/h/ln)
I-405 SB Los Angeles, CA
Sloped
8
Single vs. Multiple Lanes
• Two ML Lanes – Ability to pass slow lead vehicles
0
10
20
30
40
50
60
70
80
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Spee
d (m
ph)
Flow (pc/h/ln)
I-110 NB Los Angeles, CA
Flat
Types of Interaction Between GP and MLs
• Friction Effect - between Adjacent GP and Basic Segments• Poorly-operating GP lanes will have an adverse effect on the
operations of the adjacent ML due to their proximity
• Due to the proximity of GPL and ML traffic, increasing congestion levels on GPLs are proved to have an adverse effect on ML operations, well before the ML demand reaches breakdown levels
• This effect is particularly significant at single lane buffer-separated ML facilities
10
Frictional Effect
Types of Interaction Between GP and MLs
• Cross Weaving Effect – at ML On/Off Ramp Segments• Cross-weaving flows between a GP lane ramp and ML
access may affect GP capacity and speed
Components of the Proposed Method
12
• Method builds upon HCM2010 Freeway Facilities Chapter • But several changes are needed
• Developing managed lane segment types• New speed-flow relationships in the Managed Lanes, including
frictional impact of congestion levels in the GP lanes on ML operations
• Cross-Weave effects across GP lanes
• Implementation to occur in FREEVAL Computational Engine
Step 1: Input DataDemand (GP+ML)Geometry (GP+ML)Time-Space Domain
Step 2: Adjust demand according to spatial and time units established
(if necessary)
Step 3: Compute Segment CapacitiesGP segments using existing methods
ML segments using new/revised methods
Step 4a: Adjust Segment CapacitiesEffects of weather and work zones
Step 4b: Cross-Weave AdjustmentReduction in GP Segment Capacity
resulting from ML Cross-Weave Flows
Step 5: Compute d/c ratiosAll lane groups
Step 7: Adjacent Friction EffectsEvaluate interaction between adjacent
GP and ML lane groups
Step 9: Compute Facility MOEsAssign appropriate level of service
Step 6a: Compute UndersaturatedService Measures
Computed for each lane group
Step 6b: Compute Oversaturated Service Measures
Computed for each lane group
Legend
Existing HCM2010 Step
Existing HCM2010 Step with ML modifications
New Computational Step for ML analysis
Step 8: Lane Group LOSAssign LOS to each lane group considering
all friction and interaction effects
Step 1: Input DataDemand (GP+ML)Geometry (GP+ML)Time-Space Domain
Step 2: Adjust demand according to spatial and time units established
(if necessary)
Step 3: Compute Segment CapacitiesGP segments using existing methods
ML segments using new/revised methods
Step 4a: Adjust Segment CapacitiesEffects of weather and work zones
Step 4b: Cross-Weave AdjustmentReduction in GP Segment Capacity
resulting from ML Cross-Weave Flows
Step 5: Compute d/c ratiosAll lane groups
Step 7: Adjacent Friction EffectsEvaluate interaction between adjacent
GP and ML lane groups
Step 9: Compute Facility MOEsAssign appropriate level of service
Step 6a: Compute UndersaturatedService Measures
Computed for each lane group
Step 6b: Compute Oversaturated Service Measures
Computed for each lane group
Legend
Existing HCM2010 Step
Existing HCM2010 Step with ML modifications
New Computational Step for ML analysis
Step 8: Lane Group LOSAssign LOS to each lane group considering
all friction and interaction effects
Proposed Methodology
Step 1: Input DataDemand (GP+ML)Geometry (GP+ML)Time-Space Domain
Step 2: Adjust demand according to spatial and time units established
(if necessary)
Step 3: Compute Segment CapacitiesGP segments using existing methods
ML segments using new/revised methods
Step 4a: Adjust Segment CapacitiesEffects of weather and work zones
Step 4b: Cross-Weave AdjustmentReduction in GP Segment Capacity
resulting from ML Cross-Weave Flows
Step 5: Compute d/c ratiosAll lane groups
Step 7: Adjacent Friction EffectsEvaluate interaction between adjacent
GP and ML lane groups
Step 9: Compute Facility MOEsAssign appropriate level of service
Step 6a: Compute UndersaturatedService Measures
Computed for each lane group
Step 6b: Compute Oversaturated Service Measures
Computed for each lane group
Legend
Existing HCM2010 Step
Existing HCM2010 Step with ML modifications
New Computational Step for ML analysis
Step 8: Lane Group LOSAssign LOS to each lane group considering
all friction and interaction effects
14
Segment Types for Speed-Flow Curves
15
Quantifying Frictional Effect
vC-FFS=S 1
BP)-(c
BP)-(v CS)-(SBP -SBP=S
2
2
C
C
2fnff BP)-(v C-S=S
Where: S =Speed (mph) FFS = Free Flow Speed (mph) C1 = Linear Coefficient (or slope) v = Flow Rate (pcphpl) c = Maximum Observed Flow (pcphpl) SBP = Speed at Breakpoint (mph) CS = Speed at Capacity (mph) BP = Flow at Breakpoint C2 = Calibration Constant Sf = ML Speed during GP Congestion Snf = Speed of the corresponding non-friction curve for the same flow rate Cf = Friction Curve Constant
Quantifying Cross-Weave Effect
17Results Analysis
18
Capacity Adjustment/Reduction Factor
19
CRF Regression
where CW is the cross-weave flow measured in vph, Lcw-min is the length from the ramp gore to the beginning of BOA measured in ft, and number of GP lanes ranging from 2 to 4
20
Computational Engine: FREEVAL-ML
• Build on FREEVAL-2010• Allow analysis of parallel GP and ML facility with common
lane groups • Development of aggregated and comparative performance
measures
21
22
Summary
• A methodological framework is developed for analyzing freeway facilities with ML and GP lanes operated parallelly
• New modules incorporated in the analysis framework, such as friction module and cross-weave module are quantified via empirical research
• The computational engine is able to evaluate the parallel ML facilities and provide performance assessment to the users
23