aaron b wilson, eit westech engineering, inc., slc, ut mitsuru saito, phd, pe brigham young...

1

Evaluation of the Effectivness of VASS on Queue Mitigation

in Work ZonesAaron B Wilson, EIT

WesTech Engineering, Inc., SLC, UTMitsuru Saito, PhD, PE

Brigham Young University, Provo, UT

ITE Western District Annual MeetingSanta Barbara, CA

June 24 – June 27, 2012

2

Presentation OutlineBackgroundPurpose and ObjectivesSite SelectionData CollectionMessages Shown on VMSsStatistical Analysis

Defining Slow Traffic During AnalysisResultsConclusion and Recommendations

3

Background More people, more traffic and need for

improvement, meaning many work zones on the highways. Technical and non-technical approaches have been taken.

ITS application to work zone queue mitigation: Many studies performed on variable speed limit (VSL) signs but not many studies on VASS.

First detailed analysis on VASS: Study of VSL effectiveness by Kwon et al. (2007). They used a small warning sign to display recommended speeds.

No study has been done on a VASS system that uses regular-sized VMSs to convey advisory speeds to the drivers

Example of variable message sign on study site (Kwon et al. 2007).

4

Purpose and ObjectivesThe purpose of this project was to evaluate the effectiveness of

a Variable Advisory Speed System (VASS) in work zones to see if it would help mitigate queues.

The three objectives of this study were:Investigate the possible VASS systems to be implemented in UtahSelect a VASS and test it in a long term work zoneConduct a statistical analysis to evaluate the effectiveness of the

selected VASS

5

Site SelectionWork zone used for the

study was a segment of I-15: a widening project between 600 N and 2300 N interchanges on I-15 in Salt Lake City, to add an HOV lane in each direction. (about 3 miles)

5 Microwave sensors (orange dots) and 2 VMSs (green dots), provided by ASTI Transportation Systems. (www.asti-trans.com)

6

Site Characteristics

7

Data CollectionBefore data were

collected starting March 30th 2010

VMSs were turned on from April 27 to June 14 to collect after data.

Weather data was collected by BYU

Data collected by sensors were made available using an ftp site

8

Messages Shown on VMSsXX MPHTRAFFICAHEAD

55 MPHTRAFFICAHEAD

STOPPEDTRAFFICAHEAD

15 to 55 mph with a 5-mph increment

(eg. 42 mph 40 mph)

Location of VMS

1Location of VMS

2

9

Statistical AnalysisNull hypothesis: No difference between before

(with VMS off) and after (with VMS on) data in mitigating queues or improving traffic flows

Alternative hypothesis: VASS reduced queues and improved traffic flow characteristics in the work zone

10

Statistical Analysis (cont.)

Initial analysis planUse volume (throughputs) to evaluate

effectiveness using flow rate and a comparison of before and after data (if volume is higher, then more throughput, and less chance of queue formation)

Evaluate interaction of weather with work zoneEvaluate the same situations in both

before and after data

11

Statistical Analysis (cont.)Actual analysis

Surrogate factors were used to evaluate effectiveness of the VASS due to inconsistencies in volume data Mean speed (if higher, smoother flow, less

chance of queue forming) 15th percentile speed (if higher, closer to

mean, less variation, smoother flow) 85th percentile speed (if higher, closer to

speed limit, smoother flow) Variance in speed (if smaller, less variation,

smoother flow, less crash potential)Weather data was also inconsistent and weather interactions were not analyzed

A sample congestion scene taken by a camera of UDOT at the entry to the work zone on NB I-15, near sensor 4.

12

Statistical Analysis (cont.)Actual analysis (continued)

All factors were evaluated during the evening peak hours (3:00 PM –7:00 PM)

Day Group Mondays Fridays Weekends (Sat – Sun) Workdays (Tue – Thu)

With or without slowdowns

13

Statistical Analysis (cont.):Defining Slow Traffic

0:01:11 2:17:12 4:20:12 6:38:13 8:49:13 10:57:1413:15:1415:20:1517:39:1519:51:160

10

20

30

40

50

60

70

80

90

Volume Average Speed S1 Volume Average Speed S2

Volume Average Speed S3 Volume Average Speed S4

Volume Average Speed S5

Time (military)

Speed (

mph)

Slowdown defined as traffic speeds falling below 50 mph for more than 30 minutes

14

Initial data analysis did not yield convincing results for some combinations of parameters, (e.g. there were fewer slowdowns seen in the before data compared to the number of slowdowns seen in the after data).

Before data sample sizes were smaller than after data Weather factor –it was not feasible to perform meaningful means

tests because there weren’t enough days with bad weather

Eventually the following factors were analyzed with respect to the surrogate parameters for the evening peaks

Sensor location Existence of slow down Day Group: Mondays,

Workdays, Fridays, Weekends Time of day (evening peak)

Statistical Analysis (cont.)

15

Mean Speed at Evening PeakSensor # Slowdown ? Daygroup Significance After n Before n Difference Slowdown ? Daygroup Significance After n Before n Difference

S1 No Friday - - - 68 32 - Yes Friday Yes 59 112 70 16 -11Monday No 68 37 66 32 2 Monday No 67 47 66 32 1Weekend Yes 69 32 71 16 -2 Weekend Yes 65 144 59 38 6Workday No 68 181 67 176 1 Workday No 61 142 61 16 0

S2 No Friday - - - 66 32 - Yes Friday Yes 55 112 67 16 -12Monday No 65 37 64 32 1 Monday No 61 47 63 32 -2Weekend Yes 65 32 68 16 -3 Weekend Yes 56 144 48 38 8Workday Yes 64.7 181 65 176 -0.3 Workday No 57 142 55 16 2

S3 No Friday - - - 67 32 - Yes Friday No 57 96 63 16 -6Monday No 66 37 66 32 0 Monday No 60 47 63 32 -3Weekend No 62 32 62 16 0 Weekend Yes 49 112 43 38 6Workday No 67 181 67 176 0 Workday No 57 142 57 16 0

S4 No Friday - - - 63 32 - Yes Friday No 53 112 60 16 -7Monday No 64 37 62 32 2 Monday No 56 47 58 32 -2Weekend No 60 32 61 16 -1 Weekend Yes 49 144 44 38 5Workday No 63 181 63 176 0 Workday No 54 142 53 16 1

S5 No Friday - - - 62 30 - Yes Friday No 49 112 50 16 -1Monday No 61 37 59 32 2 Monday No 52 47 52 32 0Weekend No 60 32 60 16 0 Weekend Yes 48 144 42 38 6Workday No 61 181 61 176 0 Workday No 49 142 51 16 -2

16

Variance at Evening PeakSensor # Slowdown? Daygroup Significance After n Before n Difference Slowdown? Daygroup Significance After n Before n Difference

S1 No Friday - - - 14.06 32 - Yes Friday No 28.94 112 14.06 16 14.88

Monday No 16.81 37 18.84 32 -2.03 Monday No 17.81 47 15.68 32 2.13

Weekend Yes 11.22 32 9.12 16 2.10 Weekend Yes 17.64 144 26.83 38 -9.19

Workday No 16.08 181 15.76 176 0.32 Workday No 35.16 142 45.02 16 -9.86

S2 No Friday - - - 10.30 32 - Yes Friday Yes 19.62 112 9.06 16 10.56

Monday No 10.63 37 9.86 32 0.77 Monday No 15.92 47 15.13 32 0.79

Weekend No 8.88 32 10.24 16 -1.36 Weekend No 35.52 144 38.81 38 -3.29

Workday Yes 10.30 181 9.86 176 0.44 Workday No 21.81 142 17.98 16 3.83

S3 No Friday - - - 17.22 32 - Yes Friday No 26.01 112 26.52 16 -0.51

Monday No 10.96 37 14.06 32 -3.11 Monday No 25.10 47 28.30 32 -3.20

Weekend Yes 12.32 32 46.24 16 -33.92 Weekend Yes 55.20 144 77.26 38 -22.06

Workday No 12.25 181 12.11 176 0.14 Workday No 28.41 142 21.90 16 6.51



Weekend No 7.84 32 15.52 16 -7.68 Weekend Yes 43.96 144 64.96 38 -21.01

Workday No 8.35 181 8.41 176 -0.06 Workday No 26.73 142 23.23 16 3.50



Weekend No 7.40 32 11.97 16 -4.57 Weekend No 49.98 144 55.65 38 -5.67

Workday Yes 6.05 181 7.34 176 -1.29 Workday No 21.44 142 13.10 16 8.33

17

ConclusionWhen VASS was on, statistical significance of the

difference in mean speeds and variances was seen during the weekend when slowdowns were present.

When VASS was on, speeds were closer to the speed limit of 55 mph near the entry to the active work zone. Speed in the work zone approach seemed to be more stabilized.

“After” speed variances were relatively smaller than “before” speed variances while there were slow downs at all sensors, indicating less variation in speeds and therefore providing smoother flow and contributing to queue mitigation.

18

RecommendationsEquipment installation and configuration may take time;

hence, short term work zones may not be a good candidate for implementing a large scale VASS. VASS is meant for a long-term work zone.

Traffic studies should be performed before implementing VASS to ensure that queues are expected to form regularly to get maximum benefits from VASS.

VASS showed some level of effectiveness for the work zone studied; however, additional studies are recommended to further evaluate the effectiveness of VASS in different work zone layouts.

19

Thank you! Any questions?A 2-page handout is available.

aaron b wilson, eit westech engineering, inc., slc, ut mitsuru saito, phd, pe brigham young...

Documents