Identifying Split Failures Due to Over Saturation (Demand> Capacity)

1
Identifying Split Failures Due to Over Saturation
(Demandgt Capacity)

• Ed Smaglik, Darcy Bullock,Jim Sturdevant Tom
  Urbanik

2
Prevailing View

• A signal phase does not have periods of over
  saturation if an agency never sees or hears about
  problems.

We need tools to tabulate performance metrics
related to over saturation 24-7!
3
Outline

• What is saturation?
• How do we quantify it?
• Yauch v/c ratio
• How can we measure it?
• Fuhrs . collect data for better information..
• How should we look at it?
• What can we do?

4
Indianapolis US 31 116th
5
Saturated or Undersaturated
6
Under saturated split with slack

• Split reduction to improve efficiency

7
Saturated split with no slack

• Unmet demand at termination

8
Cycle Based Binning of qg
9
(No Transcript)
10
v/c ratio
Need served volume available in real time

• Ex.. 30s green on 60s cycle
• Sat flow is 1800 vphpl
• Observed flow on green/amber is 16
• v/c 1.06
• no slack green there
• Served vs. Demand Issue

11
Phase 2, October 18th, 2004
12
Noblesville site, October 18th, 2004 v/c plots
1 2 3 4
5 6 7 8
13
1 2 3 4
5 6 7 8
14
Split reallocation
1 2 3 4
5 6 7 8
October 26th

• October 18th

Phase 5
Phase 6
15

• Can we infer Split Failures by estimating served
  volume to capacity ratios?

16
Split Failures counting the dotsgt1.0
Timing Plan Duration of Operation Split Failures (Before) Number of Phases (Before) Split Failures (After) Number of Phases (After)
AM Peak 3 hr 5 84 5 87
Mid Day 2 hr 1 74 0 74
PM Peak 4.5 hr 16 131 9 131
Off Peak 5.5 hr 17 223 5 226
17
Comments

• Before we can manage during over saturation, we
  need to
• Figure out what we are going to measure
• Determine how we can pull some meaningful
  information out of data
• Short term goal develop tools to enable small,
  strategic human-in-the-loop to adjustmentswithout
  extensive field visits.

18
Challenges

• High Quality Cycle by Cycle Count Data
• Detection Technology
• Collection on Controller
• Quickly extracting meaningful easily understood
  graphs

19
Recommended LT Sensor Installation
20
Questions?
Do we mess with PCEs?
21
Extra
22
(No Transcript)
23
Southbound Count Detectors
24
Results Turning Movements
25
NC1234C
EA1234C
26
Video Detection Noblesville, INData
Collection Cabinet
C Camera Detector Cards D Loop Detector
Cards E Patch Panel

• A Remote Windows Computer for Data Collection
• B Video Multi-plexer

27
Evaluation Location
28
N