Synthesis of Variable Speed Limit Signs

1) I–4

2) US 27 West of Fort Lauderdale (two–lane, divided, rural, high speeds)

1) 10.5 miles

2) 3 miles, both directions

1) Mainly automated, but operator input is still used

2) Automated

1) This VSL system has not been very successful, but it is still in operation. There was a study conducted 3 years ago from University of Florida (before/after) that looked at the VSL algorithm and made recommendations for improvement, although Florida DOT has not implemented the changes yet. Overhead signs and more enforcement is needed.

2) The VSL system is effective now. We did not observe much compliance until enforcement was added.

1) Loopdetectors, side–fire radar (volume, speed, occupancy), closed–circuit television cameras (can see weather, but not really using the system for weather conditions)

2) Loop detectors, side–fire radar (added later to detect speed and check compliance rates)

1) We did hold some meetings with the Highway Patrol to get their input and comments. When you have such a large volume of vehicles in the peak hours, it is very difficult to pull over one or two violators, especially without causing a secondary crash. So, the Highway Patrol stayed out of it completely. Florida DOT provided outreach materials (e.g. brochures, websites, announcements, radio ads, etc.), but the outreach was unsuccessful. Some of the lessons learned: we need better (possibly overhead) signs and enforcement partners should be involved to observe better compliance rates.

2) Once the Highway Patrol began issuing tickets, everyone complied.

1) Yes.

2) No.

1) We completely turn off the system to accommodate work zones in the vicinity of the VSL system. The entire corridor is being reconstructed, so work zones are common.

2) There isn't much construction in this area, so work zones are not a problem.

I–35W (18 miles; Burnsville to Minneapolis)

I–94 (10 miles; Minneapolis to St. Paul)

I–35W: Approximate cost to install was $16M. This included the lane control signals and structures. The sensors already existed.

I–94: Approximate cost to install was $10M. This included the lane control signals and structures. The sensors already existed.

I–35W: 155 signs

I–94: 101 signs

VSL worked well for recurring congestion, but did not work well for non–recurring congestion.

One of the VSL roadways is a 5–lane freeway with good sight distance so drivers can see the slowed traffic 1 mile ahead. Roadway design is a factor in the effectiveness of VSL. If MnDOT were to do it again, they would focus VSL in places where sight distance is not good; they would use it in spot locations.

Drivers would not decelerate from 70 mi/h to 45 mi/h because they would get run over; it was unsafe.

VSL should include additional information to help motorists understand the reason for the speed change. For example, messages such as SLOW TRAFFIC AHEAD.

MnDOT has not seen a reduction in crashes.

MnDOT will be using a queue warning system on a construction project on I–94 that uses Doppler radar sensors and PCMSs every 1/2 mile. The system will display the actual speed downstream. So the signs will display XX MPH SPEED 1/2 MILE AHEAD. When the speed is <15 mi/h, the sign will display STOPPED TRAFFIC AHEAD. This system could be more of a replacement for VSL if it is effective. The construction project will be 2 years. The queue warning system will be activated in August 2016 for 3 months an will then be active all next construction season. UMN will do the evaluation after summer 2017.

MnDOT uses ramp metering for roads with closer interchanges and VSL for roads where interchanges are farther apart. What is the cost/benefit for each?

MnDOT emphasizes that there is a difference between speed harmonization and queue warning. They consider queue warning in spot locations.

If the system is being used primarily as a queue warning system, would a dynamic message sign (DMS) or even a static sign with remote flashers be just as effective at less cost? Is speed harmonization effective in urban areas or is it more suited for exurban areas with long distances between ramps? The entire VSL cost needs to be compared with DMS and ramp metering strategies. Full VSL costs must include operations, maintenance, utilities, and end–of–life replacement, all of which are more involved than other strategies. MnDOT has found ramp metering to be the more cost effective.

Yes:

• ACCIDENT AHEAD BE PREPARED TO STOP
• DEBRIS AHEAD DRIVE WITH CAUTION
• DELAYS AHEAD BE PREPARED TO STOP
• MOWING OPERATION AHEAD
• REDUCE SPEED CONGESTION AHEAD

1) Oregon Route OR) 217 Adjacent to I–5 in Portland area

2) OR 213 West of downtown Portland; single location sign at a single intersection, recreational traffic issues, 1st deployment of VSL in Oregon

1) 7 miles

2) 1 intersection

1) Fully automated

2) Semi–hybrid

1) Congestion, Weather

2) Congestion

1) In–road, downstream sensors are used that measure 85th percentile speeds at a 1 minute interval. We pair up the segments of the highway so that the decrease from one speed to the next is no more than 10 mi/h between segments. The current algorithm can be used for advisory and regulatory systems.

2) Single sign

1) Advisory

2) Regulatory

1) There are 7 or 8 segments and a set of signs for each segment. There is a sign over each lane at each location plus VMS sign that further discusses current conditions. There are about 40–50 signs on main line plus "travel time" messages at every intersection as you enter the OR 217 corridor, which accounts for another 30–40 VMS signs. All the signs are full matrix. We could change the system from advisory to regulatory, if needed, at any time.

2) Just one sign

1) Overhead

2) Side–mounted

Total of 9.0 mi

NB = 3.4

SB = 5.6

Yes

70 mi/h: visibility <10 mi / >1320'

50 mi/h: visibility <1320 ft. / >480'

35 mi/h: visibility <480 ft. / > 240'

Yes; sight distance relative to fog conditions.

Signs are located in relation to entrance ramps and MUTCD guidelines.

Yes.

Speed data is available from the Roadway Traffic Monitoring System.

1) I–66 US 50 to I–495

2) I–77 Fancy Gap Area

3) I–95 Express Lanes

1) 12.5 13 mi

2) 15 mi

1) Active (VSL was turned on, then back off after a week of operation. It took 3 months of turning the algorithm to turn it back on again, which was in the middle of January).

2) Planned (end of summer 2016)

3) Active

1) Automated

2) Hybrid

3) Manual

1) Congestion, Work Zones

2) Weather

3) Congestion

1) The algorithm is still being iteratively tweaked.

2) The posted speed primarily depends on the available visibility. We do not want a lot of speed variance, so we are trying to "split the difference" between vehicle's current speeds and the desired speeds.

1) We use a smoothing and trooping algorithm. Dynamic segments are calculated based on current traffic conditions. We look at the slowest speed and then slow down the oncoming traffic into that slower speed.

2) We use a smoothing and trooping algorithm. Dynamic segments exist on this corridor, much like on I–66. We determine where the worst visibility conditions exist, and the speed is set based on the worst case.

1) Rain is not used at all.

2) Rain is not used at all. We want to pursue this in the future, but it is not currently included in the algorithm.

1) Design speeds are much higher than the posted speed limits. Roadway curvature is not included in the algorithm because the roadway is designed for 70 mi/h or more.

2) Sight distance is included in the algorithm since it is a weather–based system.

1) We are still evaluating the effectiveness of the system. It is a congestion–based system, so vehicles can only go but so fast when there is congestion on the roadway. A more important question for evaluating system effectiveness may be "Are we more smoothly transitioning vehicles into different speeds?"

2) The system has not been turned on yet.

1) Compliance is always a question, especially without automated speed enforcement (Europe has this). We have to rely on traditional enforcement and make–do with what we have.

2) We have a challenge due to competing constraints: there is a "safe speed" and then there is actual driver behavior. Sometimes drivers are traveling 20 mi/h over the safe speed. When it comes time to set the speed limit, we don't want to create increased variance, but we also want drivers to travel at a safe speed. This balance can be very difficult.

1) The system was turned on in the middle of September 2015. The total cost was $39 million. This cost included a lot of additional costs for commutations, cameras, infrastructure, gantries, etc. The gantries themselves were probably about $24 million.

1) I–66 has a huge plan set, but it would be hard to track down and hard to interpret.

2) We could see if we can share.

1) Advisory

2) Regulatory

3) Regulatory

1) The system is advisory, so drivers cannot be cited for speeding. The system was implemented more for speed harmonization–related goals. We have different driver attitudes in the United States than in Europe. In the United States, drivers don't slow down unless they see a reason.

2) Enforcement is very challenging. This system drops the speed when there is fog, and we can't have an officer on the side of the roadway when there is heavy fog because it is a safety concern. We are still working through how to enforce the VSL system without endangering the officers. It is possible that enforcement would occur after–the–fact.

1) 21 gantries in each direction and 3–5 signs per gantry

2) 44 signs

1) Overhead

2) Side–mounted

1) Yes. For example, "Congestion ahead."

2) Yes. For example, "Fog ahead."

1) Overhead lane use control signs are used (can either post VSL or green arrow/red "X" indicating lane availability).

2) VSL can be used by operators to reduce the speed limit; however, this is a rural, very low volume area, so work zones don't really cause traffic problems.

1) The challenge in educating drivers with repeated explanations for speed limits and other outreach activities. VDOT employees in the northern region could provide additional insight.

2) We won't know until the system is activated.

1) There are currently no plans to extend the current system. There may be significant geometric changes along I–66, so nobody wants to do anything until we know exactly what the roadway will look like.

2) There are no plans to extend the VSL system.

There are 2 functional groups:

Group 1: Weather / road environmental condition responsive:

• US–2 Vicinity Steven pass
• I–90 Vicinity Snoqualmie pass

Group 2: Congestion / Q–Warn / Speed Transition as part of the ATM corridors

• I–5 Tukwila to Seattle
• I–90 Bellevue to Seattle
• SR 520 Bellevue to Seattle

The VSL system lengths are:

• US–2 Vicinity Steven pass (23 mi over the pass)
• I–90 Vicinity Snoqualmie pass (25 mi over the pass)
• I–5 as part of ATM corridor (8 mi)
• I–90 as part of ATM corridor (10 mi)
• SR 520 as part of ATM corridor (8 mi)

Group 1: Both systems are hybrid, semi–automatic

Group 2: For the ATM corridors, all 3 systems are fully automatic

Group 1: Weather

Group 2: Congestion

Group 1: The display speed is determined from a look–up table. Currently the operator looks–up the table and manually displays the speed.

Group 2: The displayed speeds is determined and adjusted every few minutes by monitoring downstream conditions, calculating the 85th percentile speed, comparing multiple speed values in the corridor, performing smoothing/ transitioning calculations, and displaying speed updates on one or more gantries as needed.

Group 1 has wet pavement conditions as part of the look–up table. Speeds are displayed accordingly.

Group 2 calculates speed based on actual measured downstream condition, hence no need for wet condition adjustment.

Group 1 is in rural setting. The spacing between signs are farther apart. Most of the time, the seed are set for longer segment of the corridor (if not for the entire length). The speed variation tends to be small.

Group 2 is in urban setting and responsive to downstream congestion. Spacing between gantries is approximately mi apart. Variation is tighter at 5 mi increment.

Group 1: Hybrid cut–out LED speed.

Group 2: Full color, full matrix. Speed limits are graphics resided locally in the sign controller.

Group 1: No

Group 2: Yes

Group 1: Well received.

Group 2: Initial reservation during the first few months of deployment. Well received by now after tweaking of algorithm and lower threshold.

Always follow the system engineering process. Let the corridor goals drive the operation needs. Let the operation needs drive the system requirements. Let the system requirements drive the specifications. Do not deploy VSL just because it's "cool".

Credibility is critical. Following a display of a reduced speed limit should be a real condition warranting a reduction of speed.

Some agencies are backing away from VSL because of unexpected consequences.

Buy–in from law enforcement (after VSL is implemented) is an issue.

The expected benefits have not proved out.

WisDOT has looked at the VSL systems in Minnesota and Seattle.

Wisconsin wants to increase capacity; they want speed harmonization and less crashes to increase throughput.

Can you ever get compliance with an advisory system?

Cost/benefit information. They can estimate costs, but want to know what other agencies' actual costs have been. It is also hard to explain to the public the benefits based on the cost.

Information on safety benefits. What have other States seen in terms of safety?

The synthesis report should include best practices on collaboration, cooperation, communication, and outreach.

Information on signage; specifically spacing (how far apart to install the signs).

What threshold do other agencies use for activating weather–responsive VSL systems? Do they activate for mist, flurries, or 1' of snow?

How are messages activated?

What rate of change do agencies use? Do agencies step down speeds by 5 mi/h or something different?

Do agencies use lane–by–lane signage with different speeds or the same speed for all lanes?

Have any public surveys been done? Wisconsin will have a queue warning system this summer which will operate somewhat like an advisory VSL. They will have a survey as part of this project.