Transcript Slide 1
Calibrating Model Speeds, Capacities, and Volume Delay Functions Using Local Data
SE Florida FSUTMS Users Group Meeting
February 6, 2009 Dean Lawrence Munn
Typical Free-flow Speed Lookup Table Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Typical Capacity Lookup Table Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Reasons for Improving Model outputs are more realistic and detailed Model validation process and outcome improved Models are sensitive to more project types Model applications are easier to defend Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Some Recent Projects: Indiana Statewide Travel Demand Model Genesee County, Michigan Model Maricopa Association of Governments Model Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Indiana Statewide Travel Demand Model • Current model covers 87,000 sq miles • Model network has 45,000 links, but only covers Minor Collector or higher • TAZ system has 4700 zones • Extends into neighboring states, includes major metro areas • Contains detailed data Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Indiana Statewide Travel Demand Model Free-flow Speed Assumptions • Speed assumptions have evolved over time • Use real posted speed from roadway data files • Modify posted speed for advisory speeds (no passing, horiz. curves, etc.) • Modify posted speed using data gathered from actual travel speeds • Combine travel speed and intersection delay into composite travel time Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Indiana Statewide Travel Demand Model Free-flow Speed Developed for I-69 Tier 1 EIS Incorporated into ISTDM v 4 Takes estimated posted speeds Adjusts for actual driver behavior Data came from 26 county GPS speed survey This methodology has been implemented in multiple MPO models
Table 1. Free-Flow Speed Estimation Formula Area Type Free-Flow Speed 1, 2 2-lane 2-way undivided highways
Rural 0 .
25 009751 PSPD 2 30 .
03397 Suburban Urban 117 .
640917 PSPD 0 .
0015 0 .
001279 PSPD 25 6 .
25 189 0 .
9437 PSPD 98 .
065483
2-lane 2-way divided highways
Rural 0 .
000017 ( PSPD 19 .
835323 25 72 .
323105 ) 2 Suburban Urban 0 .
019702 1 3 .
180682 PSPD 0 .
857638 84 .
105587 e 41 .
803252 / PSPD 25 0 .
119687 0 .
023365 ln( PSPD ) 1 0 .
373821 PSPD 25
Multilane undivided highways
Rural 0 .
000017 ( PSPD 19 .
835323 25 72 .
323105 ) 2 Suburban Urban 0 .
019702 1 3 .
180682 PSPD 0 .
857638 84 .
105587 e 41 .
803252 / PSPD 25 0 .
119687 0 .
023365 ln( PSPD ) 1 0 .
373821 PSPD 25
Multilane divided highways
Rural 2 .
836165 16 .
0359 PSPD 0 .
8223 0 .
071256 PSPD PSPD 2 Suburban 25 0 .
000071 9 .
061039 ( PSPD ln( PSPD 64 .
166165 ) 2 ) 0 .
000744 PSPD 3 0 .
035258 1 Urban 25 0 .
081714 0 .
016217 25
Full access controlled highways
ln( PSPD ) 1 64.00 67.06 70.21 Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009 Note:
1
Free-flow speeds in mph,
2
PSPD: Posted speeds in mph
Source
: Bernardin, Lochmueller & Associates, Inc., 2004
Condition
25 PSPD 55 PSPD < 25 25 PSPD 55 PSPD < 25 25 PSPD 55 PSPD < 25 25 PSPD 55 PSPD < 25 25 PSPD 55 PSPD < 25 25 PSPD 55 PSPD < 25 25 PSPD 65 PSPD < 25 25 PSPD 55 PSPD < 25 25 PSPD 55 PSPD < 25 25 PSPD 50 50 < PSPD 65 PSPD < 25 25 PSPD 55 PSPD < 25 25 PSPD 55 PSPD < 25 PSPD = 55 PSPD = 60 PSPD = 65 PSPD = 70
Note
No or Partial Access Control No Access Control No or Partial Access Control
Indiana Statewide Travel Demand Model Free-flow Speed Accounting for Intersection Delay Intersection capability developed for I-69 Tier 1 EIS Incorporated into ISTDM v 4 Adjusts travel times to account for signal delays Method transferred to several MPO models Recent MPO models have enhanced capabilities d 0.5C
1 g C 2 PF where, d = delay per vehicle, g = effective green time, C = cycle length, and PF = progression adjustment factor. Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Indiana Statewide Travel Demand Model Development of Model Capacities Computing an HCM compatible capacity from link attributes Originally developed for the Indiana Statewide Model Applied on a link by link basis Documented in NCHRP 358 as a best practice Subsequently implemented in many models Variables Used: Speed Number of Lanes Functional Classification Access Control Median Type Lane Width Shoulder Width Pct. Heavy Vehicles Interchange Density Access Points per Mile Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Indiana Statewide Travel Demand Model Calibrating Model Capacities Adjustment to Capacity for Signal Delay Signal delay is accounted for in capacity instead of VDF Capacity reduction factor is a ratio of flow rates Similar effect as VDFs with signal delay Advantage is simplicity at assignment stage where,
f s f s
a a
* ln( * ln(
FFS
)
b AFFS
)
b
= capacity reduction factor for signal delay, a, b = constants, FFS = free-flow speed, and AFFS = “adjusted” free-flow speed with signal delay. Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Indiana Statewide Travel Demand Model Calibrating Volume-Delay Functions Original work done in 1996-97 at INDOT INDOT calibrated BPR alphas and betas INDOT ATR data covered multiple road classes INDOT had a handful of locations for each class BPR alpha and beta parameters were developed and coded as link attributes Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Genesee County Michigan Model • Current model covers 652 sq miles • Model network has 4300 links • TAZ system has 676 zones • Network contains detailed Michigan Geographic Framework data Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Genesee County Michigan Model Free-flow Speed Assumptions More Advanced GPS Survey Applications Methodology developed for AMBAG model Used for Genesee Model Being applied for Phoenix Model Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Genesee County Michigan Model Free-flow Speed Assumptions Computing Space-Mean Speed from GPS Min T 1 Start Time T n-1 Correspond Model Link Model Link T L T L = Max - Min = GPS Point = GPS Link Layer = Model Node T 1…..
T n = GPS Time T L = Total travel time between two nodes T L = T n - T 1 Max T n End Time Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Genesee County Michigan Model Free-flow Speed Assumptions Computing Space-Mean Speed from GPS – Separating Signals and Mid-Block 400 feet t 1 s d D L Mid Block tmb D T L 400 feet t 2 s d = Signal node t 1 sd = Travel time Signal delay t 2 sd = Travel time Signal delay tmb = Travel time mid-block without Signal delay D L = Total distance of the link between two Signals T L = Total travel time between two Signals Information is used to adjust mid-block free flow speed from posted speed Also used to add travel time for intersection delay Actual GPS data by corridor was used to verify accuracy of Speed-Cap program Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Maricopa Association of Governments Model • Started major model update in 2008, activity based model • Current model covers 11,000 sq miles • Model network has 22,000 links, but only covers Minor Arterial or higher • TAZ system has 2000 zones • Dean was PM on supply-part modeling tasks (network) • Project is on-going Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Maricopa Association of Governments Model Free-flow Speed Assumptions • Obtain travel speeds from GPS survey covering ¾ of road network • Obtain travel speeds from loop detector data (~40 locations) • Test travel speeds from other ITS detectors (radar, passive acoustic) • Used the same methodology used for Genesee, Michigan • Final product was a new speed lookup table Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Maricopa Association of Governments Model Calibrating Model Capacities Using Typical Profiles Capacity Adjustments 2.25
4.50
10.00
Freeways (facility type 1) Variables
Speed No. of Thru Lanes per Direction Lane Width RH Shoulder Width Pct. Hvy. Veh.
1Hr Ideal Flow per Lane f_lateral_clearance f_heavy_vehicles f_lane_width f_number_lanes f_interchange_density 1Hr Capacity per Lane 3Hr Capacity per Lane 6Hr Capacity per Lane 24Hr Capacity per Lane
CBD
1 55 4 12 1 0.22
2250 0.98
0.87
1.00
0.98
0.85
1,605 3,611 7,222 16,048
Outlying CBD
2 60 3 12 6 0.22
2300 1.00
0.87
1.00
0.95
0.90
1,727 3,887 7,774 17,275
Mixed Urban
3 65 3 12 6 0.22
2350 1.00
0.87
1.00
0.95
0.95
1,863 4,192 8,384 18,631
Suburban
4 70 3 12 6 0.32
2400 1.00
0.83
1.00
0.95
1.00
1,895 4,264 8,528 18,951
Rural
5 75 2 12 6 0.32
2400 1.00
0.83
1.00
0.93
1.00
1,849 4,161 8,322 18,493 Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Calibrating Model Capacities Using Typical Profiles
Number of Lanes Adjustment Lanes
1 2 3 4 5
Adjustment
0.93
0.93
0.95
0.98
1.00
Interchange Density Adjustment Lane Width Adjustment Area Type Adjustment
1 2 3 4 0.85
0.90
0.95
1.00
Width
9 10 11 12
Adjustment
0.95
0.98
0.99
1.00
5 1.00
FHWA Functiona l Class
Rur Int Rur Princ Art
HV Prop Default Access Points Per Mile
0.32
0.00
0.18
2.50
Rur Min Art Rur Maj Col Rur Min Col 0.12
0.10
0.05
5.00
7.50
10.00
Rur Local Urb Int Urb Exp Urb Art Urb Maj Col Urb Min Col Urb Local Ramp 0.07
0.22
0.16
0.09
0.07
0.04
0.07
0.22
10.00
0.00
0.00
15.00
10.00
20.00
20.00
0.00
Freeway - Capacity Adjustment Factor for Shoulder Width
Note: Reads number of lanes and speed from Freeway Sheet
Lanes
2 3 4 5 1 0.95
0.96
0.98
0.99
2 1.00
1.00
1.00
1.00
Area Type
3 1.00
1.00
1.00
1.00
4 1.00
1.00
1.00
1.00
5 1.00
1.00
1.00
1.00
Freeway Ideal Capacity per Lane Per Hour Speed Range
0 60 60 65 65 70 70 75
Ideal Flow
2250 2300 2350 2400
Non-Freeway Ideal Capacity per Lane Per Hour Speed Range
0 50 55 60 50 55 60 75
Ideal Flow
1700 2000 2100 2200 Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Maricopa Association of Governments Model Calibrating Model Capacities The Problem of Multi-Hour Time Periods HCM only provides guidance for one hour capacities With uniform temporal distribution, multiply by number of hours Real traffic is not uniformly distributed over time Period-specific directional and peak factors have to be developed 5.00% 4.50% 4.00% 3.50% 3.00% 2.50% 2.00% 1.50% 1.00% 0.50% 0.00% 0 5 10 15 20 25 30 Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Calibrating Volume-Delay Functions Phoenix project used extensive loop detector data Phoenix data limited to freeways and a few arterials Phoenix data allowed some testing of variability by area type Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Calibrating Volume-Delay Functions Freeway Loop Detector Data Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Calibrating Volume-Delay Functions Freeway Loop Detector Data Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Calibrating Volume-Delay Functions Signalized Arterial Loop Detector Data 50 40 30 20 10 0 0 Lane 1 (left lane) Lane 2 (shoulder lane) 50 100
Vol (veh/15min)
150 200 Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009
Calibrating Volume-Delay Functions Curve Parameter Calibration and Evaluation Process
Sample Freeway Section
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0
0.00
0.20
0.40
0.60
0.80
V/C Ratio
1.00
1.20
1.40
HCM max density Freeflow Speed BPR Spiess 45 61.3
Alpha 1.12
7.57
Beta 5.15
1.08
R-squared 0.97
0.96
RMSE 0.41
0.38
1.60
Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009 Actual BPR Spiess
Thank You Any Questions?
Calibrating Model Speeds, Capacities and Volume Delay Functions Using Local Data February 2009