Synthesis of Variable Speed Limit Signs

CHAPTER 4. FINDINGS

As a component of active traffic management (ATM), variable speed limit (VSL) systems are subject to a systems engineering process² and are generally operated under a set of rules from a concept of operations document. In addition to planning the technical components of the system, VSL operators must also consider policy implications related to administrative law, case, and law enforcement priorities and policies. The system management includes operations, maintenance, performance monitoring, and coordination with partners, including law enforcement, private roadside assistance services, and external partners who may request VSL implementations for special events. There are several considerations related to systems management that have been learned from past implementations.

PLANNING AND POLICY

Rationale

States have various reasons for implementing traffic control systems (e.g. managing traffic in congested areas and following roadway incidents, altering speeds due to current weather and/ or visibility conditions, controlling traffic surrounding work zones, modifying speeds based on pavement conditions, etc.). A few examples of varying rationales are discussed below.

The Oregon Department of Transportation (DOT) chose to implement VSL along OR 217 due to large crash rates on the roadway (more than 230 crashes/year). Half of the crashes occurred in peak traffic hours, and rear-end collisions accounted for 70 percent of the crashes (Mitchell, 2016).

The VSL system along US 27, a two-lane, divided, rural roadway in Florida, was installed to control high vehicular speeds surrounding a school zone. The goal of the VSL system was to increase safety by better controlling the traffic surrounding the school in both directions.

The Virginia DOT considered utilizing ATM, including VSL, along I-66 to increase safety, decrease congestion, and improve environmental sustainability along the corridor. With these overall objectives in mind, various stakeholders brainstormed specific goals for the system, which included reducing the quantity and severity of collisions, decreasing travel times, increasing system reliability, improving safety surrounding construction zones, enhancing communication tactics to provide vital information to drivers, and lowering vehicle emissions and fuel consumption. (Iteris, Inc., 2011).

The Virginia DOT is currently designing an active traffic safety management system (ATSMS) for I-77 due to two major incidents that occurred because of heavy fog conditions along the corridor. In 2005, 26 people were injured and 1 person was killed when approximately 50 vehicles collided due to extreme fog conditions. In 2010, another incident involving 70 vehicles occurred due to intense fog, where 16 people were injured and 2 people were killed. In addition, the crash in 2010 negatively impacted the economy, costing about $8 million. Following these crash events, the Virginia DOT decided to implement various traffic control methods to improve safety along the roadway by decreasing the magnitude and severity of collisions due to weather conditions (URS, 2012).

While it has not employed VSL systems, the Arizona DOT is currently in the process of designing a VSL system to counteract the State's problem with dust storms. This issue occurs at a specific location in Arizona that is heavily impacted by such storms due to the area's terrain and surrounding land uses. Arizona DOT is hopeful that the VSL system will increase safety along the roadway. If successful, Arizona DOT would consider implementing other VSL systems in more northern areas of the State which are negatively impacted by snow. Further information regarding Arizona's future VSL plans may be found in Appendix B.

Nevada DOT installed a VSL system on US-395 to reduce speeds during high wind events. The VSL is part of a larger wind-warning system and is tied to two road weather information systems (RWIS). One RWIS is located on the northern end of the valley and the other is located on the southern end. High wind speeds have a history of blowing over high profile vehicles on I-580. Therefore, once the wind is high enough, I-580 is closed and vehicles are redirected onto US-395. However, the wind can also affect vehicles on the alternate route, so a VSL system was installed to reduce speeds when warranted by conditions. Typical speed limits on US-395 are either 55 mi/h or 50 mi/h. When one of the RWIS measures a 30 mi/h wind gust, the VSL is activated and all speeds are lowered to 45 mi/h. At least 30 minutes must pass without a 30+ mi/h wind gust measurement from either RWIS before the speed limits can return to 55 mi/h or 50 mi/h.

Initiation Process

One of the first steps in the planning process for the I-66 ATM system was to identify user needs. In order to determine these needs, the Virginia DOT held multiple meetings and forums with various stakeholders to gain their input regarding system design and then summarized these conversations in a Technical Consensus Memorandum. The identified needs were then transformed into overall project goals and objectives that would shape the final design of the system (Iteris, Inc., 2011).

At the beginning of the planning process for the system along I-77, the Virginia DOT identified eighteen specific stakeholders and summarized the roles and responsibilities of each once the system is activated. The roles and responsibilities were categorized as (URS, 2012):

• Responsible (those that work with intelligent transportation system (ITS) devices themselves).
• Accountable (those that can allow or reject operational decisions).
• Consulted (those that provide insight to others in the Responsible/Accountable groups).
• Informed (those that should always be updated and notified of system functionality).

The Virginia DOT outlined the I-77 project goals based on former traffic incidents and previously completed safety studies, which focused on quantifying and categorizing collisions that occurred along the I-77 corridor. Stakeholders and countermeasures were discussed in these past studies; therefore, the Virginia DOT could apply any relevant findings/conclusions from these studies to the I-77 design plan. In addition to defining stakeholders, project goals and objectives were developed, and Measures of Effectiveness were outlined based on the project objectives. A list of potential countermeasures was also developed based on past research studies along I-77. The final, selected countermeasures were determined following various consultations with stakeholders/Virginia DOT employees and further study of the I-77 corridor (URS, 2012).

Overarching Design and Operations Considerations

Currently, the VSL system along the NJ Turnpike is manually operated; however, the system used to be automatic. The automatic system relied on copper inductive loops located in the pavement to gather current traffic data, such as vehicle speed and volume. The automatic VSL system successfully and efficiently managed traffic conditions along the New Jersey Turnpike. However, the automatic system was switched to the current manual system due to the level of maintenance the copper inductive loops required. Any time the roadway was repaved, the inductive loops were damaged and needed to be repaired. Therefore, the system became manual and more reliable sensors (from Sensys) were installed along the roadway. The New Jersey Turnpike Authority noted that the Sensys sensors are very reliable, but they are spaced farther apart than the inductive loops, which creates a slight lag in responsiveness. In addition, sensor reinstallation is still included in all paving contracts to ensure proper sensor functionality. The New Jersey Turnpike Authority is currently working to restore the system's former automatic capabilities in order to increase throughput and efficiency.

Public Outreach

Georgia's first VSL system was activated in September 2014 as a speed management strategy on I-285. Much of the public's reaction was negative, with many believing it to be a new mechanism for generating cash for the State. To help educate the public, the Georgia DOT adopted the slogan of "Slow Down to Get There Faster." They developed an educational video to explain why the VSL system is being implemented, what it is, how it works, and the benefits drivers can expect. The video is posted on a dedicated VSL webpage (Georgia Department of Transportation, 2015) along with other educational information and materials, such as a fact sheet. In addition to the webpage, the Georgia DOT established an email address (VSL@dot.ga.gov) where the public can send their comments.

Minnesota implemented VSL systems on I-35W and I-94, but both are currently turned off. The systems were slow to respond to real-time conditions, which ultimately caused the public to lose trust in the speed limits. Consequently, the systems were turned off and the Minnesota DOT is reevaluating them to make improvements. The Minnesota DOT received very few comments from the public when the VSL were activated. Most questions asked about the meaning of the messages. For example, drivers expressed confusion about whether the displayed speed indicated the speeds ahead or the recommended speed. Of interest, the Minnesota DOT only received two inquiries after the VSL were turned off; both were to ask why the system was off.

Nevada did not use an aggressive public relations campaign before activating the VSL, but a press release was issued. This was due in part to the nature of US 395 serving as an alternate route in a smaller community. Public reaction to the VSL on US 395 has been mixed with positive and negative feedback. The negative responses have primarily been from homeowners because they most often see when issues occur with the signing. For example, hardware problems have caused the signs to go blank. In response to a request by the Highway Patrol, Nevada DOT installed beacons on the VSL signs that flash when the speed changes. Homeowners have complained that the beacons are too bright. To address this issue, the DOT has temporarily disabled the beacons but plans to try a dimmer in the future as a more permanent solution.

When the VSL system was first implemented along OR 217, the Oregon DOT received some feedback indicating driver confusion about speed limit reduction. However, after improving the speed algorithm, public feedback has been positive. The Oregon DOT stressed the importance of public notification about the purpose of the VSL system and why the speed is being reduced.

The Washington State DOT reported that the VSL systems along I-90 (near Snoqualmie Pass) and along US 2 were well-received by the public. However, the public had reservations regarding the VSL systems along I-5, I-90 (Bellevue to Seattle), and SR 520 during the first few months of deployment. But, after refining the algorithm and lowering the speed threshold, the public is now in favor. The Washington State DOT found that during periods of extreme congestion, such as stop- and-go traffic, the public feedback was negative regarding the concept of displaying a system floor threshold speed such as 30 mi/h.

The University of Florida evaluated the VSL system along I-4. Part of their evaluation included a survey that captured driver's opinions of the VSL system. Participant responses indicated that many drivers will not reduce their speed until other drivers begin to slow down as well. Participants also stated that overhead gantries and signing above each lane would be helpful. In addition, survey results showed support for side-mounted VSL signing (Elefteriadou, Washburn, Yin, Modi, & Letter, 2012).

Liability

To date, the team has not identified an agency experiencing issues with liability. Agency processes do include archiving all speed data which can be provided as documentation or evidence of the posted speed limit at any specific time.

Only two VSL systems reviewed noted experience or recommendations for liability issues. Nevada DOT recommended that lawyers should be involved early on in future deployments to evaluate possible tort liability after the VSL implementation on I-80 (Robinson et al. 2002). With respect to the VSL system on I-526 in South Carolina, the system was created due to a court order. A Federal judge ruled that the I-526 Cooper River Bridge construction project must include a low visibility warning system, which included VSL (Goodwin 2003).

DESIGN, DEPLOYMENT, AND STANDARDS

Infrastructure Requirements

Agencies report a variety of infrastructure used to operate their VSL systems. These components may differ based on the function of the VSL. Table 4 shows the fundamental VSL elements for a system operating to manage speeds during congestion, weather, and work zones.

Table 4. Fundamental variable speed limit infrastructure requirements.

Variable Speed Limit Infrastructure Component	Variable Speed Limit Function
	Congestion	Weather	Work Zones
Changeable Speed Limit Signs	✓	✓	✓
Weather/Environmental Sensors		✓
Traffic Speed/Volume Sensors	✓		✓
Communications Equipment to Transmit Data	✓	✓	✓

Signage Type and Placement

The Florida DOT reported approximately 20 VSL signs along the I-4 corridor. At least one sign is posted every mile, and some signs are located in medians while others are side-mounted along the roadway. I-4 also displays various word messages (VMS) in conjunction with their VSL system. Some VSL signs along US 27 are posted in medians while others are side-mounted. No VMS signs are used along US 27.

Georgia's VSL system includes 176 electronic speed limit signs for an interstate corridor that is 36 miles long (inclusive of both directions). Signs are mounted on both sides of the highway in 88 locations and are spaced every ½-mile to 1 ½-mile.

The Minnesota DOT installed 155 signs on I-35W, which is 18 miles long, and 101 signs on I-94, which is 10 miles long. When activated, the State used lane-by-lane overhead signs. Although displays were by lane, each showed the same speed at the same location. The signs are full matrix color CMS that measure 4 feet tall and 5 feet wide. HOT lane signs displayed a white diamond with no speed message. This was enacted to address concerns about displaying different speeds over different lanes, but at the same time not wanting to artificially slow down the speed in the HOT lane. Note that Minnesota did not use word messages in conjunction with the VSL signs.

Due to maintenance issues with the signs, the Minnesota DOT is considering either replacing them or using a single message sign as opposed to lane-by-lane signing. The latter alternative will reduce installation, maintenance, and operations costs. In addition, the Minnesota DOT is evaluating whether a VSL or a simple word message of SLOW TRAFFIC AHEAD is more successful for queue warning in a particular high crash area. It is possible that Minnesota could use the system for spot locations rather than implementing it throughout an entire corridor.

The Nevada DOT system used to lower the speed for trucks so that a long, elevated structure over a canyon can remain open in high winds. Nevada DOT determined the locations of the US 395 VSL signing installations based on locations of intersecting roadways; this permitted trucks to turn off the roadway if extreme winds were encountered, but the bridge remained open to automobile traffic. Signs use embedded LED and are mounted on the right side of the highway.

In addition to the VSL system, VMS are also installed along the New Jersey Turnpike. All VSL signs are posted adjacent to VMS that describe the reason for the speed change, as shown in Figure 1. The following word messages are used to warn drivers of conditions ahead:

• ACCIDENT AHEAD BE PREPARED TO STOP.
• DEBRIS AHEAD DRIVE WITH CAUTION.
• DELAYS AHEAD BE PREPARED TO STOP.
• MOWING OPERATION AHEAD (with tractor image).
• REDUCE SPEED CONGESTION AHEAD.

Note that VMS along the New Jersey Turnpike are also used to notify drivers when the far right lane can be used as a shoulder (red "X") and when it is a travel lane (green arrow) during peak travel periods. VSL and VMS are only displayed on message boards that are within 2 miles of the traffic issue (e.g., lane closing, construction site, congestion, etc.). The New Jersey Turnpike Authority noted that the warning messages and altered speed limits remain relevant to drivers when they are not posted too far in advance.

OR 217 in Oregon has a set of VSL signs for each segment. All signs are displayed overhead above each lane with additional VMS for traffic-related messages. The Oregon DOT estimates 40 to 50 signs on the main line plus 30 to 40 VMS signs that display travel time messages prior to entering OR 217. All of the VMS are full matrix and display messages such as CONGESTION AHEAD at a certain distance upstream of the congestion or information related to current weather conditions.

In the Chattanooga, Tennessee area along I-75, there are 10 signs that are right-shoulder mounted with embedded white LEDs for an interstate corridor that is 9 miles long (inclusive of both directions). There is one display for all lanes and signs are located in relation to interstate entrance ramps and Manual on Uniform Traffic Control Devices (MUTCD) guidelines. The Tennessee DOT uses a FOG SPEED LIMIT word message on the changeable speed limit sign, which is mounted on the shoulder. In addition, the following word messages are activated on VMS in conjunction with the VSL.

• FOG AHEAD TUNE TO 1620.
• REDUCE SPEED TURN ON LOW BEAMS.
• CAUTION FOG AHEAD.
• FOG AHEAD SPEED LIMIT 50 MPH.
• FOG AHEAD SPEED LIMIT 35 MPH.

Along I-90 (near Snoqualmie Pass) in Washington, VSL sign locations vary depending on the roadway geometry at any given point. Some areas have overhead signing while others have side- mounted signs on both sides of the roadway for each direction. US 2 is an undivided highway, and all VSL signs are located on the right-hand side of the roadway. All of the VSL signs on I-90 (near Snoqualmie Pass) and along US 2 are hybrid cut-out LED speed limit signs. VMS are not utilized in either location.

There are 21 overhead gantries along I-66 per direction, and each gantry holds 3 to 5 signs. VMS are used along I-66 to display messages to drivers in addition to the VSL (e.g. CONGESTION AHEAD, etc.). In addition to the VMS and VSL signing, other devices are also provided along I-66, including closed circuit television (CCTV) cameras, ramp metering, lane management devices, etc. (Earnest, 2015).

There will be 44 side-mounted signs along the VSL corridor on I-77 in Virginia. Thirty-six of these signs will be full matrix, VMS that can post speed limit messages and traffic management messages (e.g. FOG AHEAD, etc.). Eight of the signs will be typical variable speed limit signs where the display speed can dynamically change. Additional devices will include CCTV cameras, visibility sensors, etc.

To control everyday traffic along the corridor, I-66 uses signing to indicate lane availability, especially within and surrounding work zones. VMS are used to display information regarding work zones, and green arrows/red "X" symbols over each lane are used to indicate current lane availability. Work zones are not expected to be an issue along I-77 since it is a rural, low-volume roadway.

Integration with Active Traffic Management and/or Road Weather Information Systems

Several VSL systems were either planned as part of a larger ATM system or integrated with existing ATM or RWIS as a source of data or as a shared backbone for hardware and/or software systems.

The VSL system along I-66 in Virginia is part of a larger ATM that also includes VMS that can display other traffic management messages (e.g., CONGESTION AHEAD). The VSL system on the Pennsylvania Turnpike built off an existing ATM system to consolidate operations and reduce cost. The system used a series of RWIS stations to determine fog conditions. Nevada DOT's VSL relies on data from two separate RWIS stations. Once an RWIS measures a 30 mi/h wind gust, the VSL on US-395 is activated to reduce speeds. The VSL signs do not display normal operating speeds until neither RWIS measures a wind gust of 30 mi/h or more for 30 minutes. On the I-215 VSL in Utah, the DOT cited the lack of integration with the existing ATM as a serious obstacle during implementation.

OPERATIONS AND MAINTENANCE

Control Algorithms

VSL system speed control algorithms have been widely studied in both academic papers and evaluation reports. Lu et al. (2014) and Ma et al. (2016) presented comprehensive reviews of advanced algorithms for VSL systems, particularly as components of ATM. While these algorithms have been shown to be effective in simulation studies, they are often too complex to be implemented in the field.

Generally, VSL systems are activated when certain conditions (e.g., volume, occupancy, road surface conditions, or weather conditions) are met; corresponding algorithms will generate new speed limits. Usually, decisions are supported by real-time sensors that can detect current roadway conditions (e.g., traffic, weather, visibility, pavement). The algorithms can differ from project to project. In many cases, the 85th percentile speeds of downstream congested traffic are used directly or indirectly as new speed limits. In other cases where there is no congestion but severe road conditions, such as low visibility, engineers use look-up tables to determine speed limits using pre- determined values based on condition thresholds. Some VSL systems are deployed during major construction projects to slow upstream vehicles for safety purposes; a single reduced speed limit may be set in this case.

Based on their objectives, speed control algorithms can be categorized into two types: 1) speed homogenization projects that focus on improving safety, and 2) multi-objective projects that may strive for improvement of mobility and/or reduction of environmental impacts in addition to speed homogenization. Most systems related to weather, visibility, and work zones fall under the category of speed homogenization, while systems that react to current traffic conditions belong to the category of multi-objective projects.

Two particular challenges of setting variable speed limits were identified during agency interviews. One difficulty was generating speed changes in a way that felt natural to drivers, both at the stage of speed reduction or speed recovery from reduced speed. Oregon DOT advised that it takes multiple iterations to develop a system that reacts naturally enough to reduce negative feedback and to increase compliance. The other challenge was determining how to manage competing interests between the assigned "safe speed" and actual driver behavior, since many drivers travel much faster than the posted speed limit. When calculating a suitable speed limit, the goal should be to display a speed that is safe for travelers but also will not create increased variance. This balance can be very difficult to achieve.

In practice, algorithms differ depending on project objectives, purposes, weather conditions, and the surrounding environment. Algorithms for congestion-focused VSL systems are typically more complex because they need to consider overall effects on corridor traffic conditions instead of simply reducing speed and speed variance for safety. Congestion-focused projects may also be weather responsive if adverse weather conditions exist. Key questions related to dynamic speed limit setting issues include:

• What are the factors to consider, such as volume threshold, occupancy threshold, surface conditions, and 85th percentile speeds? How are they considered?
• Are there any safety issues to slow down traffic if the average speed is considerably higher than posted speed?
• How should maximum and minimum posted speed limits be determined?
• What is the period over which speed statistics are calculated?
• When should speed limits be adjusted and by what increment?
• How often can the speed limits be changed?

Various examples of speed control algorithms and the corresponding approaches to implementation issues are described below.

The VSL system along the New Jersey Turnpike was installed in the early 1960s and is still being used today. Although the VSL system is currently operated manually, it was automatic in the past. The VSL were automatically calculated and posted according to average travel speeds collected by copper loop detectors located in the pavement (United States Department of Transportation, 2002). In order to avoid creating a second area of congestion, the VSL signs upstream of the traffic issue were posted as 10 mi/h faster than the speed of the downstream traffic. For example, if the downstream traffic was traveling at 25 mi/h, the upstream VSL would be 35 mi/h. Supervisors only manually intervened when setting speeds for construction work zones or if travel lanes were shut down along the roadway. The algorithm for speed reduction was simple: the speed limit was reduced in 5 mi/h increments with a minimum speed limit setting of 30 mi/h. The VSL system not only displayed the reduced speed limit, but it also displayed a REDUCE SPEED AHEAD message on VMS as well as the rationale behind the speed reduction. When appropriate, the distance from the warning sign to the congestion, crash, construction, etc. was also displayed (United States Department of Transportation, 2002).

The current VSL system employs approximately 250 signs (both VSL and VMS) along the entire New Jersey Turnpike corridor and is manually operated. The maximum regulatory speed limit along the New Jersey Turnpike is 65 mi/h, and there are other areas with a regulatory speed limit of 55 mi/h. Most of the time when there is a downstream issue, supervisors manually reduce the speed to 45 mi/h (except during poor weather conditions). During poor visibility conditions, the speed is determined based on how many mile markers are visible from a stationary location along the corridor. If three mile markers are visible, then the speed limit is posted as 35 mi/h. If two mile markers are visible, then the speed limit is set as 30 mi/h and operators consider closing the roadway. Currently, the posted VSL apply to all lanes, though VSL may vary across lanes in the future.

The speed limits along I-4 in Florida are determined with loop detectors and side-fire radar, which detect volume, speed, and occupancy. Weather conditions are visible through CCTV, although the VSL system along I-4 was primarily built for speed harmonization due to large dynamic waves frequently observed along the roadway rather than to observe weather. The Florida DOT reported that the loop detectors provide extremely reliable data. The side-fire radar systems have improved over the years, but they are still not fully reliable today.

The displayed speed limits along I-4 in Florida are regulatory and are based on the 85th percentile speed in 5 mi/h increments. When an event occurs that requires a speed alteration, the VSL system informs the traffic management staff and then recommends an appropriate speed. The speed selection algorithm along I-4 accounts for the design speed of the roadway, which depends on roadway curvature, superelevation, sight distance, etc. The staff may then accept or alter the recommendation. Once a suitable speed has been accepted, the speed limit is posted according to the following rules (FDOT: Traffic Engineering and Operations Office, 2010):

1. The posted speed limit is reduced and the yellow warning light begins flashing.
2. The new traffic flow is monitored and it is ensured that the new speed limit is appropriate.
3. If necessary, the speed is reduced in 5 mi/h increments while never dropping the speed 20 mi/h or more under the normal roadway speed limit.
4. Once the event has cleared, the normally posted speed limit is displayed and the flashing yellow light is turned off.

Because of ongoing construction on the I-4 corridor in Florida, the Florida DOT completely turns off the VSL system to accommodate work zones in the vicinity. This obviates the need to move signs, maintain electrical and communications to signing systems, and ensure that adequate data collection is taking place, particularly in areas where the freeway surveillance systems are disrupted by the ground works associated with grading and pavement reconstruction.

US 27 in Florida also uses loop detectors to determine speed limits. The Florida DOT later added side-fire radar systems just to detect current speeds along US 27 and to check compliance rates. US 27 does not experience much construction; therefore, it is not necessary for the VSL to accommodate work zones.

Georgia uses various sensors placed 1/3 mile apart on I-285. Sensors transmit data every 20 seconds, including the total volume and average speed. Weather stations do not contribute to the VSL system, and the Georgia DOT does not use probe data—primarily because there is a lag in that data. VSL speeds are determined based on speeds downstream of the sensors, so the cause of a slowdown in traffic is not a determinant. The VSL system is fully automated, but there is the option to change the speed manually as well. The VSL is manually changed for situations like work zones. In Georgia, construction typically occurs at night when there is less traffic volume. However, motorists can drive much faster with less traffic, so the VSL is used to lower speeds in active work zones.

The Minnesota DOT set speeds using an algorithm developed by the University of Minnesota – Duluth. When activated, the VSL would display a speed 5 mi/h lower than the posted speed (which is 55 or 60 mi/h) with a minimum speed of 30 mi/h. The same speed did not have to be set for the entire corridor. Instead, when congestion was detected, as many as three sets of lane control signals could be activated prior to the congestion. Because the lane use control signals for the active traffic management system (ATMS) equipment also functioned as the display modules for the VSL, the activation of VSL in advance of the congestion was desirable for the purposes of step-down speed harmonization. With lane control structures located every ½ mile, the VSL could be activated as much as 1.5 miles upstream of the congestion. This allowed traffic management staff to reduce speeds gradually as the traffic approached congestion.

The Nevada VSL system is part of a larger wind-warning system. Wind-speed data is tied to two RWIS stations; one on the north end and one on the south end of the valley. For the VSL to be activated, one of the RWIS stations must record a wind gust of 30 mi/h or more. Once activated, the speed limit is lowered to 45 mi/h. It is not raised back to its operating speed (either 50 mi/h or 55 mi/h depending on the section) until neither RWIS station registers a wind gust > 30 mi/h for 30 minutes. The system operates automatically, but there is some human interface from the traffic management center for confirmation.

The displayed speed limit along OR 217 is determined by in-road, radar-based, downstream sensors from Wavetronix that measure 85th percentile speed at a 1 minute interval. The displayed speed is calculated as the lower of the following values: 1) 85th percentile speed, or 2) Speed of downstream traffic + 5-10 mi/h. If the calculated speed is less than 30 mi/h, then the system displays SLOW (Mitchell, 2016). The Oregon DOT has found these sensors extremely reliable. There are at least seven to eight different segments throughout the entire VSL corridor, and each segment is evaluated separately and assigned an appropriate speed limit. The speed setting algorithm ensures that the changes in speed between different segments are no more than 10 mi/h. Although the current VSL system is advisory, the VSL algorithm can easily be converted from an advisory to a regulatory VSL system. Oregon statutes regarding basic speed establish strict criteria for the installation of non-advisory speed limits. In addition, case law has established precedent for drivers overturning citations for violating posted maximum speeds, which, in Oregon, are signed with SPEED LIMIT signs instead of the SPEED signs found on rural primary highways.

The OR 217 speed algorithm accounts for both weather conditions and congestion levels. The final displayed speed limit depends on which piece reports the most needed condition change (weather vs. congestion). There is a friction factor sensor at each speed sensor location which considers roadway condition during calibration. When the friction factor reaches a certain level, the displayed speed limit is calculated based on the current weather conditions instead of congestion levels. A summary of the weather responsive algorithm is shown in Figure 2 (Mitchell, 2016).The necessary speed alterations based on weather and/or traffic are automatically calculated and implemented within the VSL algorithm itself. The algorithm does not account for roadway curvature since OR 217 is a freeway-level facility. In addition, the Oregon DOT does not have much experience in accommodating work zones in the vicinity of VSL since the last major construction in the area was the installation of the VSL system itself.

Condition Code	Visibility	Grip Factor	Surface Condition Classification	Condition Speed	Weather Message	Actual Sign Message
A	<Visibility Threshold	>= Grip Factor Threshold	Moist, Wet	Maximum Speed - 10 MPH	"LOW VISIBILITY"
B	<Visibility Threshold	< Grip Factor Threshold	Moist, Wet	Minimum Speed	Slippery when wet sign + "USE CAUTION"
C	>=Visibility Threshold	>= Grip Factor Threshold	Moist, Wet	Maximum Speed	None
D	>=Visibility Threshold	< Grip Factor Threshold	Moist, Wet	Maximum Speed - 20 MPH	Slipper when wet sign + "USE CAUTION"
E	<Visibility Threshold	>= Grip Factor Threshold	Frosty, Snowy, Icy, Slushy	Maximum Speed - 10 MPH	"LOW VISIBILITY"
F	<Visibility Threshold	< Grip Factor Threshold	Frosty, Snowy, Icy, Slushy	Minimum Speed	ICE sign + "USE CAUTION"
G	>=Visibility Threshold	>= Grip Factor Threshold	Frosty, Snowy, Icy, Slushy	Maximum Speed	None

The Tennessee VSL system on I-75 is weather-responsive with speeds changed based on visibility during fog conditions. It functions in a hybrid fashion with speed changes occurring both automatically and manually. Speed limits are determined by a conditional visibility algorithm due to weather event(s) related to fog, traffic speed, and stopping distances. The same speed is set throughout the corridor using the following parameters:

• Speed = 70 mi/h when visibility is < 10 miles and ≥ 1,320 ft.
• Speed = 50 mi/h when visibility is < 1,320 ft. and ≥ 480 ft.
• Speed = 35 mi/h when visibility is < 480 ft. and ≥ 240 ft.

Environmental sensors are used with the I-75 system and are reported by the Tennessee DOT to be very reliable.

The VSL system along I-66 in Virginia has dynamic (instead of fixed) speed segments. Dynamic segments allow the speed limit to apply to different lengths of the roadway depending on existing needs. A smoothing speed algorithm is used to appropriately alter vehicle speed within each dynamic section to maintain suitable traffic flow. The smoothing algorithm determines the slowest speeds along the corridor, and then it transitions the oncoming traffic into that slower speed.

Note that the Virginia DOT is still improving the current I-66 speed setting algorithm. Weather, roadway curvature, sight distance, and pavement type/condition are not included in the speed limit calculations for I-66, although some of those variables may be included in future iterations. I-66 currently utilizes Wavetronix speed sensors, which have been extremely reliable in providing relevant traffic conditions.

Since the I-66 VSL system in Virginia has only been consistently active for approximately six months, the Virginia DOT is still evaluating the effectiveness of the VSL system for reducing speeds. The Virginia DOT stated that maybe the most important determinant of the effectiveness is if the algorithm is successfully transitioning drivers into both higher and lower speed zones.

In addition to the VSL system along I-66, the Virginia DOT is currently designing a VSL system on I-77, which will primarily be used for visibility purposes. The VSL will be determined based on the available visibility with the goal of reducing speed variance. Similar to I-66, I-77 will also have dynamic speed segments, but the length of these segments will change depending on visibility levels. The algorithm will determine the areas with the worst visibility and then set the appropriate speed limits around those areas. Sight distance is included in the speed setting algorithm since it is a visibility-based system. Wet conditions, roadway curvature, and pavement type/condition are not included in the speed limit calculations for I-77, although some of those variables may be included in future iterations. The planned I-77 VSL corridor will use Wavetronix sensors to capture data.

Two of Washington's VSL systems have similar methods of operation: I-90 (near Snoqualmie Pass) and US 2. Both systems are regulatory, operate in rural areas, and display speed based on an operator look-up table, which accounts for current pavement conditions, visibility, weather (i.e. rain, snow), and incidents, as shown in Table 5. Currently, the operator uses the table to determine the appropriate speed and then manually displays it. Roadway curvature, sight distance, and pavement type/condition are not considered in the speed setting algorithm. The displayed speed limits are not necessarily the same throughout the entire corridor, and there is lane discrimination (e.g., high occupancy vehicle (HOV) lanes might have a different speed limit than the general purpose lanes). The Washington State DOT reported that both VSL systems have been effective at reducing speeds, and speed variation is small. In addition, the Washington State DOT stated that their current sensors are very reliable, and they have extensive experience in calculating travel times based on speed converted from occupancy measurements.

Table 5. Washington State Department of Transportation speed limit reference.

Traction Requirements	Speed Limit	Pavement Conditions	Visibility	Weather	Blocking Incidents
None	65	Dry or Bare/Wet.	Good: Clear > 0.5 Miles.	Fair To Moderate Rain.	Incident On Shoulder.
Traction Advisory	55	Light Snow, Slush, or Ice In Places.	Moderate: Fog < 0.2 Miles.	Hard Rain.	Incident On Shoulder.
Tractor Trailer Requirement/Vehicle Over 10,000 GVW Chains Required	45	Comp. Snow/Ice, Deep Slush, Shallow Water.	Poor: Blowing Snow < 0.1 Miles	Heavy Rain Or Snowfall.	Lanes Blocked Traffic Moving.
Chains Required All Vehicles Except All Wheel Drive	35	Severe Freezing Rain, Deep Snow, Slush Or Standing Water.	Poor: Blowing Snow < 0.1 Miles.	Heavy Rain Or Snowfall.	Lanes Blocked Traffic Stopped Ahead.
Emergencies or Extreme Conditions Only	25	Use this speed for severe conditions as requested by crews on the scene. Confirm with supervisor, when available. Poorest possible road conditions and human life endangered. Conditions should be well documented. Return to higher speed limit as soon as possible.	Use this speed for severe conditions as requested by crews on the scene. Confirm with supervisor, when available. Poorest possible road conditions and human life endangered. Conditions should be well documented. Return to higher speed limit as soon as possible.	Use this speed for severe conditions as requested by crews on the scene. Confirm with supervisor, when available. Poorest possible road conditions and human life endangered. Conditions should be well documented. Return to higher speed limit as soon as possible.	Use this speed for severe conditions as requested by crews on the scene. Confirm with supervisor, when available. Poorest possible road conditions and human life endangered. Conditions should be well documented. Return to higher speed limit as soon as possible.

The displayed speed limits along I-5, I-90 (Bellevue to Seattle), and SR 520 in Washington are computed using the same method. All three systems are regulatory and located in urban areas. The displayed speeds are determined and adjusted every minute by monitoring downstream conditions: 1) the 85th percentile speed is calculated, 2) multiple speed values are compared in the corridor, 3) smoothing/transitional calculations are performed, and 4) the displayed speeds are updated as needed. Since the displayed speeds are calculated by using measured downstream conditions, there is no need to include wet conditions, roadway curvature, sight distance, nor pavement type/ condition in the speed calculations. The displayed speed limits are not necessarily the same throughout the entire corridor, and there is lane discrimination (e.g. HOV lanes might have a different speed limit than the general purpose lanes). All three of these VSL systems have been effective at reducing speeds.

Manual versus Automatic Operations

Many VSL systems operate in a hybrid fashion using a combination of automated and manual speed changes. There are fewer instances of a system being fully manual or entirely automated, but there are examples of each.

Speed limits on I-495 in Delaware are manually determined by the chief traffic engineer of Delaware DOT, the traffic management center manager of the DOT, or by request of the Delaware State Police, according to weather and road conditions. Using expert opinion and on-the-ground input limits unexpected speed variation due to faulty sensors or poorly calibrated control algorithms.

Presently, the VSL system along the New Jersey Turnpike is manual; however, the system used to be automatic. The New Jersey Turnpike Authority is currently working with other entities, including IBM and Rutgers University, in order to restore the automatic capabilities of the VSL system. The NJ Turnpike Authority noted that automatic VSL systems allow more throughput; therefore, transitioning back to an automatic system is ideal. Alternatively, the Nevada DOT system is fully automated and speeds are changed using wind-speed data from RWIS stations. A threshold wind gust of 30 mi/h automatically activates the VSL system so that the changeable speed limit signs show a reduced speed limit.

When an agency uses a hybrid approach, they typically rely primarily on an algorithm to automatically change the speed limit and supplement with a human interface. This may involve looking at data or video feeds to confirm the VSL is appropriately set for current conditions or overriding the automated speed limit for an extenuating variable. For example, the Georgia DOT system on I-285 automatically adjusts the VSL using speed data transmitted from sensors. However, agency personnel can override the system to manually change the speed limit during nighttime construction to reduce speeds in work zone areas. In addition, the VSL system in Florida along I-4 recommends a certain speed limit based on field sensor output, and the operator must then approve or alter the proposed speed. On I-5 and I-90 in Washington, operators monitoring the system can override automatically adjusted VSL if necessary, though this is not desirable for typical operations.

Advisory versus Regulatory Operations (Enforcement)

The success of VSL systems is highly dependent on compliance, and therefore it is essential that regulatory systems are consistently enforced. However, in real-world deployments, particularly those in the United States, many systems are still advisory or cannot be enforced as intended.

In some cases, State laws prevent VSL systems from being enforced. In Minnesota, the VSL systems on I-35W and I-94 were advisory because regulatory systems would have required a legal change. Even so, stakeholders in Minnesota shared the same views as many other agencies: VSL systems require enforcement to gain driver compliance. If the VSL system not enforced, it is suggested that speeds need to match drivers' expectations of what is sensible.

OR 217, located in the Portland, Oregon area, is an advisory system due to limited shoulder space along the roadway and also due to State law. In order to implement a regulatory VSL system in Oregon, a long legal process would be necessary to change State law to accommodate VSL along interstates. Currently, State troopers and local police in Oregon "enforce" VSL by using a "basic rule" whereby law enforcement officers judge whether drivers are traveling safely and prudently rather than examining vehicle speed. Oregon DOT is planning to utilize VSL systems in other areas once the State law has been altered to allow VSL installation on interstates, including a 30 mile, weather-based VSL system.

In many other cases, VSL systems were intended to be regulatory, but actual enforcement was limited. One major obstacle is the lack of direct access to speed limit information by law enforcement. The former Missouri VSL system was located on I-270 in St. Louis. It commenced as regulatory, but law enforcement was reluctant to issue citations because they were unsure of the current speed limit. Consequently, the system was changed to advisory, but compliance became an issue. The system was therefore ultimately deactivated. Missouri has no other VSL systems as of May 2016.

The Georgia Highway Patrol as well as law enforcement from 14 local jurisdictions can enforce speeds on I-285. Prior to activating the VSL system, the Georgia DOT met with the Highway Patrol to explain the system. As many others have experienced, the reaction from law enforcement was not positive with concerns centralized on the officers not knowing the current speed that should be enforced. To help address this issue, the Georgia DOT provided the Highway Patrol with a direct data feed so they can see the signs at all times. Additionally, the Highway Patrol is using a different citation tactic to work around needing to know the exact speed. Instead of focusing on speed as the offense, law enforcement issues citations for reckless driving or driving too fast for conditions. To further support law enforcement, the Georgia DOT archives all of their VSL data and can provide information to the Highway Patrol when needed to verify the set speed at a specific time.

The VSL in Nevada is also regulatory. Law enforcement response has not been positive, primarily because of the hardware and software problems that have caused issues with the VSL signs. For example, a 45 mi/h speed limit may be displayed in one direction, but the signs display a 55 mi/h speed limit in the opposite direction. This has caused law enforcement to lose confidence in VSL, and Nevada DOT is considering temporarily turning off the system to replace the hardware. Law enforcement is not directly notified when the VSL is activated, but they are aware by default because they are notified when the larger wind-warning system is closing routes; so the speed limit reduction is implied.

Finally, it can be difficult to enforce speed limits in conditions where it is unsafe for law enforcement to exceed the posted speed limit. For example, the VSL system on I-77 in Virginia is regulatory and enforceable, but speeds are most often decreased due to heavy fog. Heavy fog is not only a safety issue for drivers, but it is also a safety problem for law enforcement officials. Therefore, enforcement along I-77 is a complex issue that transportation officials in Virginia are still working through.

OUTCOMES

Performance Measurement

Depending on functional requirements and system goals, VSL systems have been evaluated with various performance measures or measures of effectiveness (MOEs) as follows:

• Traffic efficiency: average speed and travel time at a certain time interval (e.g., 1 minute or 5 minutes), travel time reliability, traffic throughput, driver journey times, traffic flow stability, number of significant shockwaves.
• Safety: general crash rates (categories by crash severity: fatal and injury, property damage only, and crash types: rear-end, sideswipe, and others), crash rates during certain seasons (e.g., winter crash rate if a VSL is deployed for winter severe weather conditions).
• Other: driver subjective ratings, compliance rates, emissions measured by environmental sensors.

Agencies either apply these general performance measures directly or adapt them on the basis of special requirements and goals of the deployment site. The Virginia DOT defined MOEs for the planned VSL system on I-77 to guide the evaluations of system effectiveness. The primary MOEs focus on reducing the quantity of various crash types (e.g. fatal, injury, property, weather-related, work zone-related, etc.) along the corridor. The crash reduction goals would be met if the number of collisions over a five-year period following VSL operation is less than the number of collisions in the 5 years just before system implementation. Crash severity reduction goals were also developed related to injury and property crashes. The crash severity goals would be accomplished if the severity of injury and property crashes decreased every year following system implementation. In addition, the speed compliance goal would be achieved if the rate of compliance improves for a period following the system introduction when compared to a period just before system implementation (URS, 2012).

Compliance

States have observed varying levels of driver compliance with VSL systems. Compliance rates depend on multiple factors (e.g., regulatory vs. advisory systems, enforcement strategies, public education/outreach, etc.). Also, some speed homogenization projects, such as on I-94 in Minnesota, reported high compliance rates, perhaps because drivers are aware of the risks of high speeds in bad weather or work zones. Some deployments (e.g., A99 in Munich, Germany, and many others in Europe) adopt automated enforcement, which is effective in improving compliance. The feasibility (i.e., adoption issues and public support) of automated enforcement in the United States should be studied in the future. In addition, European sites typically report higher compliance rate and larger benefits. Future research should consider if cultural differences between U.S. and European drivers affect system effectiveness. Various State experiences related to VSL compliance rates are discussed below.

The manual VSL system along the NJ Turnpike is regulatory; therefore, the speed limit is enforced by the State Police to ensure that drivers are abiding by the posted speed limits. In addition, any time there is a severe collision along the VSL route, the police will issue any necessary citations based on the VSL that was posted at the time of the incident (United States Department of Transportation, 2002).

An outreach program was implemented (e.g. brochures, radio announcements, websites, etc.) for the VSL System along I-4 in Florida to educate the public about the purposes of the VSL system; however, without proper enforcement, minimal compliance is still observed. In contrast, Florida is obtaining high compliance rates along US 27 – a two-lane, divided, rural, high-speed roadway. Here, the VSL are regulatory and are being enforced along this roadway by the Florida Highway Patrol, which is believed to be causing the higher compliance rates.

The VSL system was advisory in Minnesota because a regulatory system would have required a legal change. Minnesota shares similar views as many other agencies – VSL requires enforcement to gain driver compliance. If the VSL is not enforced, the speed needs to match drivers' expectations of what is sensible.

The VSL system on OR 217 is currently advisory due to limited space for enforcement and State law. Even though current compliance along OR 217 is not perfect, the VSL system is still considered successful. Oregon has observed a substantial reduction in speed differentials, improved harmonization, increased roadway capacity, and a reduction in crashes along OR 217. However, Oregon would expect to see higher compliance rates with a regulatory system versus their current advisory system.

As mentioned previously, the former Missouri VSL system on I-270 in St. Louis was ultimately deactivated due to compliance-related issues. As of May 2016, Missouri has no other VSL systems.

System Benefits

With the variety of objectives and implementation approaches across VSL systems, benefits vary from deployment to deployment. Table 6 shows results for a number of representative VSL projects from the United States and other parts of the world. Not every deployment in Table 6 has been evaluated in the literature. European sites have been included in the results to better illustrate how system benefits can vary from site to site due to various influencing factors.

Speed homogenization projects usually used simple algorithms in response to real-time traffic, road, and other conditions (e.g., weather, work zone, incidents, visibility, etc.). Many of these studies reported improvement in traffic safety via before-and-after analyses (with some exceptions such as the I-270/I-255 corridor in Missouri, likely because of low compliance rates). Many of the multi-objective projects reported that VSL had positive effects on mobility, safety, and the environment. There are some discrepancies, although it is difficult to generalize the reasons for these discrepancies due to many uncontrolled factors among different sites, such as different compliance rates, heterogeneous driver behaviors, and various road geometries.

Table 6. Variable speed limit system practices and field results.

Location, Time	Variable Speed Limit (VSL) Summary	Evaluation Results
Speed Homogenization Projects
Germany, 1990	Advisory VSL. Only three speed limit options: 100, 80, or 60 km/h.	20-30% reduction in crash rates.
E18 in southern, Finland, 1990	A central control unit analyzed the data and selected one of three speed limits, 120 km/h, 100 km/h, or 80 km/h, to display, based on driving/road conditions; the system is advisory.	95% of drivers reported positive ratings of its effectiveness. Compliance rates were as high as 76%. Significant safety improvements attributable to the weather VSL implementation: accidents during the winter dropped by 13% and during the summer by 2%.
Attiki Odos Toll Motorway, Greece, 2004	Speed signs are to notify drivers when the advisory speed limit inside the tunnel is different from the other parts of the motorway. The system provides advice to motorists approaching the tunnel regarding the safe speed limit inside the tunnel.	Significant reduction in injury accidents by 10%.
I-494, Minnesota, 2006	Reduce the speed of the vehicles approaching the work zone.	25-35% reduction in maximum 1-minute average speed and a 7% increase in throughput between 6 and 7 a.m., although no throughput increase between 7 and 8 a.m. Even though the speed limit was advisory, motorist compliance was significant.
I-270/I-255 Corridor, Missouri, 2010	The maximum and minimum speed limits on the corridor are 60 mi/h and 40 mi/h, in 5 mi/h increments. Uses a 5 minute update interval (less in case of incidents). The system is advisory.	No mobility gains (in terms of throughput improvement or congestion reduction) were observed. The evaluation did show a significant reduction in number and severity of crashes by 8%. Speed limit compliance remained surprisingly low, even though the signs were mandatory.
Multi-Objective Projects
Rural A2, Amsterdam, the Netherlands, 1992	Inducing more homogenous speed and flow as well as more uniform lane usage. Reducing the posted speed limit from 120 km/h (≈ 75 mi/h) during normal conditions to a range between 70-90 km/h (≈ 43 to 55 mi/h) depending on congestion conditions. The system is advisory.	Produced more homogenous traffic flow. No clear increase of system capacity was observed (6 months after deployment).
M25, United Kingdom, 1995	Ambitious goal: managing congestion by smoothing traffic flow, reducing stop-and-go waves, controlling traffic speed, improving the road user's driving experience, and managing fuel consumption. The system is advisory. Simple algorithm firstly in 1995: speed limits selection based on preset volume thresholds. Later complex algorithm: consider both volume and speed.	Simple algorithm: complaints about the speed limit reduction when there was no sign of congestion. Low compliance rate. Complex algorithm: throughput increased by 1.5%, emissions reduced between 2% and 8%, more uniform flow and headway. Journey times showed little overall change, journey time reliability increased significantly.
A99, Munich, Germany, 2002	Objective: Improve bottleneck flow and ensure safety. The system is regulatory. Algorithms based on the calibrated Fundamental Diagram relationship of speed, density and flow.	Harmonized traffic flow. Dampened shock waves. Increased traffic flow and improved safety.
E4, E22, Sweden, 2003	Both advisory and regulatory. Goal: increase throughput, reduce shockwave, improve safety.	5 to 15 km/h (≈ 3 to 9 mi/h) reduction in speeds across the study sites, high rates of speed compliance (in particular in severe weather conditions), fewer disturbances in traffic flow, and less severe shockwaves. Reduce travel time by 5%. Most effective when they combined with additional speed enforcement and better information.
MD 100, Maryland, 2009	Smooth the transition between free flow to congested state. Algorithm consider driver response.	Increase average speed and throughput, shorter travel time.
I-5, I-90, Washington, 2010	Include a few preset speed thresholds. When thresholds reached, adjust VSL in 5 mi/h increment, with a 35 mi/h lower bound. Operator can overwrite automatic VSL manually.	Reduced average speed, reduced flow, travel time reliability increased.
A7/E15 south of Lyon, France, 2011	Objective: traffic throughput and safety improvement. Triggered by pre-set traffic flow thresholds (3000 vehicles per hour) with maximum speed limit of 110 km/h (≈ 68 mi/h).	Increased average speed by 4-10%, reduced the number of bottlenecks by 50%, reduced average travel time by 30 seconds, no change in lane capacity, reduced incidents by 17%. Low compliance rate.
I-35W, Twin Cities, Minnesota, 2010	VSL displayed 1.5 miles upstream gradually reducing the speed of incoming traffic. Using constant deceleration rate to decide VSL at the end of queues. Updated every 30 seconds.	Reduced travel time, increased traffic volume, less deceleration rate.
I-4, Florida, 2014	Objective: to improve traffic flow; to reduce rear-end and lane change crash risks. FDOT conducted an engineering and traffic investigation that identified reasonable and safe speeds under different weather and traffic conditions; e.g., some section in congested period has VSL at 20-30 mi/h—lowering upstream speed limits by 5 mi/h and raising downstream speed limits by 5 mi/h.	Not available at time of this synthesis.
I-66, Virginia, 2016	A component of active traffic management, to improve safety and operations on I-66 through better management of existing roadway capacity. The ATM includes advisory variable speed limits, queue warning systems, lane use control signs, and hard shoulder running.	No specific VSL effects were analyzed. Active traffic management has limited operational and safety impacts during the weekday peak periods and some impacts during the midday and off-peak weekday periods (2% to 6% improvement).
New Jersey Turnpike1	Deployment for both congestion and road weather management; operated manually; regulatory for all travelers.	The speed control strategy effectively decreases traffic speeds in adverse conditions. Speed management and traveler information dissemination have improved safety by reducing the frequency and severity of weather-related crashes (improvement quantity is not available).

Life-Cycle Costs

Limited information is available on the cost of VSL systems. Through interviews with operators, it was roughly estimated that the cost of deployment a VSL system along a route varies from less than $10 million to almost $40 million. This cost is highly dependent on the existence of current intelligent transportation system facilities, such as traffic detectors, VMS, and gantries.

It was even difficult for some agencies to estimate the cost of the VSL system(s) in their States. For example, many pieces of hardware, devices, and processes had already been implemented prior to VSL deployment in Washington; therefore, the actual cost of their VSL system, sensors, and maintenance is unclear. In addition, since the Georgia and Nevada systems are fairly new (both are less than 2 years old), neither agency has comprehensive data yet on lifecycle costs. Also, the Minnesota DOT is considering using one sign for all lanes (instead of lane-by-lane signs) to reduce their maintenance and operations costs.

The approximate cost to install the VSL systems on I-35W and I-94 in Minnesota was $16 million and $10 million, respectively. These costs do not include the sensors since they already existed, but they do include the lane control signals and structures.

The total cost of the I-66 VSL system in Virginia was $39 million. However, this cost estimate was unique to I-66 due to additional costs related to communication, cameras, infrastructure, gantry construction, etc. The gantries themselves cost approximately $24 million. The total cost of the planned I-77 VSL system in Virginia is $9.6 million (Earnest, 2015). This figure includes a fair amount of additional upgrades (e.g. power, etc.).