Alpharetta Adaptive Signal Evaluation: Driving Factors
Citizens call traffic their #1 issue City Council priority for traffic improvements Bond/Capitol Implementation Funds North Fulton CID project SR 9 Multi-jurisdictional ATMS project
Alpharetta Adaptive Signal Evaluation: Driving Factors
SYSTEM
BENEFITS (Percent change in)INITIAL CAPITAL
COST (per intersection)*Travel Time Delays Stops
ACS-Lite -12% to +7% -38% to +2% -35% to -28% $6,000 to $10,000
OPAC -26% to +10% n/a -55% to 0% $20,000 to $50,000
RHODES -7% to +4% -19% to -2% n/a $30,000 to $50,000
SCATS -20% to 0% -19% to +3% -24% to +5% $25,000 to $30,000
SCOOT -29% to -5% -28% to -2% -32% to -17% $30,000 to $60,000
Alpharetta Adaptive Signal Evaluation: Hypothesis
A well-timed system is a well-timed system
Cycle length- appropriate for volume Split percentages- balanced or prioritized Offsets- provide good progression
Windward Parkway 2007
0
200
400
600
800
1000
1200
1400
07/03 07/05 07/07 07/09
Tra
vel
Tim
e (h
rs/d
ay)
AM EB
AM WB
NN EB
NN WB
PM EB
PM WB
Phase 2 timings:
11% reduction in total delay
Windward Parkway 2009
Phase 3: Volume growth flat
Capacity Improvements-2nd WBL lane open at SR9RTOR allowed at GA 400 ramps
Windward Parkway with Adaptive?
$2000/day
$0.5M/2 yrs
$1200/day
$0.3M/2 yrs
$2500/day
$0.6M/2 yrs
A = ½ b x h $1.4 M/6 yrs
Alpharetta Adaptive Signal Evaluation: Objectives
Prove that system will:– Maintain peak-hour performance– Operate “on-the-fly”– Improve edge-of-peak performance– Provide adequate progression
Show that system will:– Adapt to volume changes over time– Adapt for special events – Adapt to unplanned events
Alpharetta Adaptive Signal Evaluation: Metrics
Travel time runs Cycle failures Queue failures Cycle lengths Peak hour split percentages System settings
Rhythm InSync Adaptive System
Windows based, uses IP communications Video detection- queue density and
occupancy Distributed computing- no central server Can phase in by TOD Price $25,000 per intersection Promised 20% improvement
All phases input to
controller Controller decides
when to start, end
phases based on
coord plan
Traditional Detection
Computer only inputs
calls to phases that it
wants to have on Ped and preempt calls
go to controller and
are serviced normally
InSync Detection
Windward Parkway
Costco
Wal-Mart
Home DepotMarriott
ADP
HP
Marconi
MARTAPark & Ride
HH Gregg Residential
Residential
GA 400North PointParkway
WestsideParkway
SR 9
Residential
Lowes
Kroger Super Target, Frys
Restaurantsand Shops
Residential
ADT: 40,000
Signal Timing Goals
Critical Movements
Metered Movements
Install InSync
PEAK PERIOD GOALS:• Preserve critical movements• Preserve mainline progression• Improve splits for underserved movements on edges of peak
Day 1 PM and Day 2 Noon
• Locked in 140 s cycle• De-prioritized green bands• Restricted side streets• Still too fair to side streets • Not enough progression for NB off-ramp• Other progression OK
Day 3 & 4- Reconfiguratoin
Wanted to force a “known good” configuration Gave Rhythm existing volumes and timings Set system to mimic existing peak coord plans Allowed system to adjust, using existing coord
plans as a basis Better split times throughout peak More time to side streets in beginning of
peaks meant less needed in the middle
Pedestrian Problem
InSync would be calling phase 7 (SBL) while controller was still clearing phase 6 ped
Decided not to runadaptive withoutaccounting for peds
Rhythm installed a modification at 2 intersections with themost ped traffic
Final Configuration
Restored adaptive within constraints Ran system for several months Cycle length ~10 seconds longer in peaks Split times were appropriate and variable Offsets could not maintain equivalent
progression
Things I Liked about InSync
No controller hardware/firmware changes Elegant handling of vehicle calls Recovers from changes faster than coord plan Good video detection Excellent data display Ability to manually control intersection
Rhythm InSync System Evaluation: Objectives
Prove that system will:– Maintain peak-hour performance– Improve edge-of-peak performance– Provide adequate progression– Improve off-peak performance
Show that system will:– Adapt to volume changes over time– Adapt for special events – Adapt to unplanned events
Similar
?
?
Similar
?
~~
The Verdict- InSync
Not seen to handle saturated and over-capacity conditions better than traditional coordination plan
Provides a good video detection system, plus monitoring and other functionality
Can provide a good traffic responsive system
Alpharetta decided to return the equipment and explore other options
Where to Consider InSync
Corridors with:– Less-than-saturated conditions– More protected left-turns– Infrequent re-timing– Need for high variation in splits
Opportunity to manually manage special events
Consider use for video detection, monitoring
SCOOT Adaptive System
Marketed by Siemens/Temple 170 installations worldwide 6 installations with about 200 signals in U.S. Centralized computing Ethernet to serial devices to controllers Uses existing Siemens controllers
SCOOT Adaptive System
Installed at 5 signals on Old Milton Parkway (SR 120) from Ga 400 to North Point Parkway
Most congested corridor in Alpharetta System detection using Sensys
– “Pucks” installed in pavement with repeaters– Significantly easier to install than video or loops
Old Milton ParkwayGA 400 North Point
ParkwayMorris Rd
Siemens HQ
Old Milton/North Point Parkway• PM Peak Hour- 7,000+ vehicles• Have to give OMP enough time to avoid backing through Ga400• Constricted by short turn bays
Ga 400 Interchange• Need to keep lefts clear to avoid gridlock on bridge• Try to keep ramps out of Ga400• Weave section with Morris Rd
SCOOT Pilot Program
Siemens/Temple responsibilities:– Provide all hardware and software– Perform on-site installation and configuration
Alpharetta responsibilities:– Install Sensys detection– Perform before-after studies– Assist with timing implementation
SCOOT Pilot Program
SCOOT Pilot Program
Day 1 of SCOOT control• Incident on Ga 400 SB• Significant increase of exiting vehicles on SB ramp
SCOOT Experiences
Zero complaints from citizens Usually a longer cycle length
– SCOOT: 176 seconds in peaks, 160-176 off-peak– Old plans: 150-160 in peaks, 100-120 off-peak
System has to include side street ped time (?) Does not handle left-turn trap very well Interface is practically incomprehensible Need better error notification Excellent data storage and organization
Cycle Failures at North Point Pkwy
WBL EBL NBL SBL NBT Total
AM Manual 32% 36% 0% 8% 32% 21.6%
SCOOT 56% 12% 0% 40% 4% 22.4%
MiddayManual 36% 8% 0% 12% 36% 18.4%
SCOOT 20% 32% 12% 64% 16% 28.8%
PMManual 8% 0% 52% 100% 96% 51.2%
SCOOT 32% 24% 88% 100% 68% 62.4%
% of cycles where vehicles do not clear in 1 cycle
SCOOT System Evaluation: Objectives
Prove that system will:– Maintain peak-hour performance– Improve edge-of-peak performance– Provide adequate progression– Improve off-peak performance
Show that system will:– Adapt to volume changes over time– Adapt for special events – Adapt to unplanned events
?
?
OPAC
Marketed by Televent 5 installations with 73 signals in U.S. Distributed computing Ethernet based Uses Econolite controllers Can utilize most of the same detectors as
SCOOT
Summary
Looking for a system that can provide good signal timing on-the-fly
Not expecting a magic bullet to end congestion Expecting benefits on day 300, 500,1000… Expecting benefits on unusual days Trying to investigate all options
*Relative* Costs and Benefits
Do Nothing
PeriodicRetiming
FrequentRetiming
Traffic Responsive
SCATS,SCOOT,OPAC
ACS LiteInSync