Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
Public Roads
Featuring developments in Federal highway policies, programs, and research and technology.
This magazine is an archived publication and may contain dated technical, contact, and link information.
Federal Highway Administration > Publications > Public Roads > Vol. 74 · No. 4 > Using GPR to Unearth Sensor Malfunctions

January/February 2011
Vol. 74 · No. 4

Publication Number: FHWA-HRT-11-002

Using GPR to Unearth Sensor Malfunctions

by James A. Arnold, David R. P. Gibson, Milton K. "Pete" Mills, Michael Scott, and Jack Youtcheff

Ground-penetrating radar assesses loop detectors to determine necessary repairs so that all vehicles, including motorcycles, will be able to trigger traffic signals.

This motorcycle is crossing a functioning traffic sensor loop, which notifies the traffic signal controller of the vehicle's presence and triggers a signal change. FHWA researchers are applying ground-penetrating radar as a means to detect malfunctioning sensors.
This motorcycle is crossing a functioning traffic sensor loop, which notifies the traffic signal controller of the vehicle's presence and triggers a signal change. FHWA researchers are applying ground-penetrating radar as a means to detect malfunctioning sensors.

It's probably happened to nearly every motorist at one time or another. You stop for a red light, perhaps at night, when few or no other vehicles are on the road. You wait, and wait, and wait some more, but the signal does not change to green. The problem? It could be a malfunctioning loop wire sensor.

Inductive loop wire sensors embedded in the roadway surface are devices that indicate the presence or passage of vehicles and provide information that supports traffic management applications such as signal control, freeway mainline and ramp control, incident detection, and gathering of vehicle volume and classification data to meet State and Federal reporting requirements. Gray sealant that fills cut lines on roadways indicates the location of the loop wire part of the sensors in the pavement.

Malfunctioning in-roadway loop sensors can prevent traffic signals from sensing the presence of vehicles, a problem that can be particularly frustrating and even dangerous for smaller vehicles, such as motorcycles and bicycles. Malfunction rates for loop detectors can run as high as 14 percent, according to a Transportation Research Board paper, "Malfunction Detection and Data Repair for Induction-Loop Sensors Using I-880 Data Base," by H. J. Payne and S. Thompson. Anecdotal evidence puts malfunction rates as high as 25 percent.

The extent to which malfunctioning is attributable to the loop sensors versus the communications systems that link sensors to the central traffic management system is unknown. But informal discussions with traffic researchers identify installation errors and utility trenchers cutting the wires as the main sources of hardware malfunction.

The Indiana Department of Transportation (INDOT), for example, was able to reduce its malfunction rate by requiring, first, that installations be done as preformed loops during pavement resurfacing. Second, INDOT recommends that the splices attaching the loop wires and the lead-in wires from the controller cabinets be made inside large junction boxes (also called pull boxes) at the side of the road where they are protected from moisture and are readily accessible for repairs.

To address sensor malfunction, the Federal Highway Administration (FHWA) in 2006 initiated a Small Business Innovation Research (SBIR) project, 06FH2 "Step-Frequency Ground Penetrating Radar for Location and Evaluation of In-Roadway Sensors" to develop a nondestructive evaluation (NDE) method employing ground-penetrating radar (GPR) for detecting and assessing in-roadway inductive loop sensors.

"The project staff also addressed issues such as the evaluation of sensor sensitivity, the state of the art for NDE technology prior to the project, how the research improved NDE technology, and why evaluation of pavement condition is important both for sensor evaluation and for pavement design and maintenance," says Joe Peters, director of FHWA's Office of Operations Research and Development (R&D).

How the Project Started

To assess whether the physical structure of a loop system is at fault when a malfunction occurs, a detailed examination of the sensor is necessary. Previously, destructive sawcutting and removal of a portion of the loop were required to conduct that physical assessment.

In 2004, FHWA staff members from the offices of Infrastructure R&D, Safety R&D, and Operations R&D participated in an Exploratory Advanced Research team review of GPR. When examining an NDE GPR intensity mapping of a roadway, they noticed long lines on the graph that turned out to be a water pipe and conduits. This observation initiated a discussion about just how small a feature could be detected with NDE GPR and its lower detection limits.

The NDE staff at FHWA soon determined that the then-state-of-the-art GPR and acoustics systems could not measure features as small as loop wires, much less cracks in a wire or other potential problems such as moisture in pavement, sawcuts, or loop sealant. This finding led to discussion of whether researchers could make improvements in the hardware and software so that GPR and acoustics systems could detect loop wires and loop wire defects. As a result of this discussion, FHWA decided to pursue research in this area.

How Traffic Inductive Loop Sensors Work

Traffic detector engineers need to know not only whether and where a loop is present but also whether it is working and how well it is working.

Traffic sensor loops are coils of wire installed in slots 2 to 4 inches (5 to 10 centimeters) deep beneath the pavement surface. The loops are connected inside roadside junction boxes by splices to shielded lead-in cables that in turn connect to detector electronics on a loop amplifier card located in traffic controller cabinets. Junction boxes are usually approximately 1 cubic foot (0.028 square meter) in size and placed underground with a removable cover flush with the ground surface.

Most inductive loop detector systems operate in a frequency range of 20 kilohertz (kHz) to 100 kHz. All inductive loop detector systems have an oscillator in the loop amplifier card. An oscillator is a device for generating oscillating electric currents or voltages by nonmechanical means. The oscillator current in the loop wires generates a magnetic field around the loop. The passage of the conductive components of a vehicle through the loop's magnetic field generates a current in the conductive vehicle components. The conductive components are those capable of conducting electricity. In turn, the currents induced in the vehicle's conductive components also generate a magnetic field, which reduces the loop's magnetic field and inductance. Inductance is the property of an electrical circuit measuring the induced electric voltage compared to the rate of change of the electric current in the circuit. The decreased loop inductance causes an increase in the loop's oscillator frequency. The magnitude of the frequency shift indicates the loop's sensitivity or ability to detect vehicles.

Some loop detectors fail to function because they have breaks in the loop wires such that they no longer conduct currents. Others do not operate correctly due to reduced sensitivity, which can be caused by a problem with the loop installation, loop splice connection, lead-in cable, or pavement conditions.

"Modern digital loop detector electronics have adequate detection sensitivity, so when there is a problem, it is attributable to one of these other factors," says Dan Middleton, program manager, of the Texas Transportation Institute.

Shown here is FHWA's prototype sensitivity probe mounted on a GPR antenna array on the front of a vehicle. The equipment consists of a calibrated wire wrapped with multiple turns around a square frame mounted above the pavement surface in a plane parallel to the roadway probe loop and with the outer loop wires of both loops on parallel planes.
Shown here is FHWA's prototype sensitivity probe mounted on a GPR antenna array on the front of a vehicle. The equipment consists of a calibrated wire wrapped with multiple turns around a square frame mounted above the pavement surface in a plane parallel to the roadway probe loop and with the outer loop wires of both loops on parallel planes.

Measuring Loop Sensitivity

Because traffic engineers need to know whether inductive loop sensors are present and how well they are working, FHWA researchers proposed to measure the detection sensitivity using equipment mounted on a vehicle passing over the buried loops. The researchers placed a prototype sensitivity probe on a GPR antenna array on the front of the vehicle. A cable connected the probe with a frequency measurement device inside the vehicle. Combining, filtering, and amplifying the voltage from the probe enabled the researchers to measure the frequency shift and thus determine the presence and sensitivity of a loop. The researchers spaced the probe far enough ahead of the vehicle so that they could measure the initial resonant frequency from the roadway loop prior to the frequency shift caused by the sensor's detection of the vehicle.

As noted earlier, the measured maximum frequency change is proportional to the loop's detection sensitivity. The researchers calibrated the measurement sensitivity of the vehicle probe with a standard detector loop system of known sensitivity. As the measurement vehicle traveled down the roadway, the calibration enabled researchers in the vehicle to correlate that data with NDE data from the GPR system and thereby tell whether a correctly working loop was present.

Graph. The vertical axis is labeled "Probe Signal in Volts" with divisions starting at minus 2 and going up to minus 1, zero, 1, and 2, marked off on both the left and the right of the graph. The horizontal axis is labeled "Approximate Distance in Feet" with divisions starting at 10 feet (3 meters) and marked off in increments of 5 feet (1.5 meters) up to a distance of 45 feet (14 meters) on both the bottom and top of the graph. The probe signal begins at a little above and below 0 volts at the 15-feet (4.6-meter) marker, increases gradually to 2 and minus 2 volts at 25 feet (8 meters), which is labeled "Front Wire of Inductive Loop," continues at full strength until about 33 feet (10 meters), which is labeled "Rear Wire of Inductive Loop," where it declines immediately to about 1 and minus 1 volts and then down to just above and below zero at about 44 feet (13 Meters). The signal shows an abrupt break from full strength to almost zero at both the front and rear of the wire.
As shown in this graph, the probe's reflected signal strength increases as the vehicle, traveling at 10 miles per hour (16 kilometers per hour), approaches and travels over the loop sensor. The signal strength then declines as the vehicle moves away from the sensor. The probe signal is sampled 100,000 times per second (100 kHz) for 10 seconds, producing 1 million data points.

The measured signal strength of a working loop increases as the probe loop sensor approaches the sensor loop, goes over the front of the loop, travels across the body of the sensor, and then declines as the probe travels away from the loop. The sharp breaks as the probe travels across the leading and trailing edges of the sensor provided additional confirmation of the loop's geolocation.

The Technology As It Existed

The highway community has used GPR for measurement and evalua-tion of road conditions since the mid-1970s, as reported by R. M. Morey in the Transportation Research Board report Ground Penetrating Radar for Evaluating Subsurface Conditions for Transportation Facilities. Transportation engineers also employ this technology to measure pavement thickness using calibration cores. The GPR approach often is less expensive than traditional methods that require a greater number of core samples.

Diagram. Shown here is a schematic cross section of a pavement structure at left and an associated GPR output at right. In the pavement cross section, the top layer is labeled "Asphalt Layers" and two layers below that are labeled "Base Layers." The Asphalt Layers are labeled Ea and Ta, and the Base Layers are labeled Eb and Tb. A key on the left equates E to the real part of the dielectric constant and T to the travel time of radar wave within the layer. A box above the pavement is labeled "GPR Antenna." One arrow goes from the GPR Antenna down to the pavement and then penetrates to each successive layer with a new arrowhead at the beginning of each new layer. Arrows return upward from each of the downwards arrowheads to indicate that a portion of the radar signals is reflected backward from that pavement layer back up to the GPR Antenna. The mapping of the GPR signal reflections demonstrate the correlation between the signal peaks and the distances between the GPR Antenna and the pavement surface and the interfaces between the various pavement layers. The signal strength is greatest between the GPR Antenna and the pavement surface, and weaker yet noticeable signal strength peaks occur at the interfaces of the different pavement layers. An arrow beside the GPR output points downward and is labeled "time (nanoseconds)."
(Left) This schematic drawing of a GPR antenna shows signals being reflected by each pavement layer. (Right) This time trace shows the signals being returned by the reflection of the radar by each layer as the GPR antenna moves down the roadway

"Many transportation professionals recognize the efficiency of the ground-penetrating radar method," says Mort Oskard, retired operations research engineer with FHWA's Advanced Research Team.

The radar signals employed in GPR are electromagnetic waves transmitted from an antenna source. These waves are continuous in step frequency GPR (SF GPR) systems, while they are impulses in impulse GPR (I GPR) systems. For both SF GPR and I GPR, radar reflections occur at the boundaries between the dielectric materials (insulators that do not have free charges inside them) used in civil engineering applications, such as pavement layer interfaces. The materials on each side of a pavement interface often have different dielectric properties. Therefore, these materials have different wave propagation and loss characteristics that cause wave reflection and refraction to occur at material boundaries, allowing the GPR to detect them. Radar waves also can detect and image metal inclusions or boundaries, including reinforcing steel within concrete, because they produce strong GPR wave reflections and do not effectively penetrate metals or other conductors.

The height offset from the pavement surface of a GPR antenna used in pavement analysis can range from 10 inches (25 centimeters) up to 3 feet (0.9 meter) or more and thus are called air-coupled systems because the radar couples with the surface through the air. This contrasts with ground-coupled systems, which are mounted only 2 inches (5 centimeters) or less above the pavement, so the signal does not travel for a significant amount of time in the air and thus interacts directly with the surface. The SF GPR system and corresponding antennas used in the FHWA research can operate in an air-coupled or ground-coupled configuration.

Diagram. An outer square box is labeled "Pavement." Inside that box is a smaller rectangular one labeled "Loop Sealant." Inside that smaller box are three circles labeled "Electrical Wire Encased in Insulation."
Shown here is a plan view representation of a loop wire embedded in a sawcut in the road pavement and sealant covering up the sawcut.v

Technicians may use each type of GPR system to record the time, amplitude, and phase of radar wave reflections, and computer software can be used to process two- or three-dimensional data. SF GPR data used in the FHWA study was collected in three dimensions. Traffic detector engineers process the data using a variety of algorithms, including classical synthetic aperture methods, which can generate a mathematically focused image if dielectric material properties are known. Super-resolution methods such as MUltiple SIgnal Classification (MUSIC) algorithms have the potential to provide even more refined images.

GPR using classical synthetic aperture radar (SAR) detects features 0.25 inch (6 millimeters) or smaller during measurements of pavement thickness or other applications. However, even high-resolution GPR systems were only able to resolve the separation between two closely spaced subsurface features (such as the two ends of a break in a loop wire) if the separation was at least 1.6 inches (4 centimeters) wide. In 2001, staff members with FHWA's Office of Infrastructure R&D stated in a Web report, the "HERMES II Bridge Inspector Project," that their most advanced GPR system "did not provide the necessary range resolution to definitively image typical delamination cracks, but field testing has shown that reinforcing steel and bridge deck details are typically rendered in [output] images." More recent technology allowed some delamination cracks to be imaged by a prototype GPR system, but results remained inconsistent, and regulatory requirements subsequently reduced performance. The regulations that reduce I GPR performance involved Federal Communications Commission rules in 2002 that reduced the power output and frequency content available to GPR users.

Technology Improvements That Were Needed

For engineers to use GPR to image loop wire features and potential defects or deterioration, significant improvements were needed in the resolution achievable by an NDE system and the corresponding visualization software. Related improvements in resolution were necessary to characterize the sealant surrounding the loop wire in the sawcut and determine its condition and potential problems due to moisture. Measurement of moisture content is critical to determine whether the loop wire will conduct the inductive signals to the loop detector card in the signal controller cabinet.

In other words, both defects and material deterioration associated with sensor, sealant, and pavement failures must be characterized with a much higher degree of precision than is required for conventional GPR used in pavement and subgrade applications. Cracks and wire features can be as small as 0.125 inch to 0.0625 inch (3 millimeters to 1.6 millimeters), requiring a significantly improved GPR detection and resolution capability. A corresponding step up in the frequency of the GPR radar signals used to image these features, and an improvement in the imaging algorithms was applied to produce images of the loop wire sawcut and embedded loop wire.

Technology Development And Testing

To summarize, the researchers developed and tested two complementary technologies: a passive probe sensor and a SF GPR-based method for data collection and analysis imaging. The passive probe successfully determines whether the loop is functioning and, if functioning, shows its location through a plot of its signal strength. The SF GPR method then images the loop wire in the pavement to evaluate potential cracks, defects, or deterioration in the wire or adjoining pavement.

GPR Detection of Traffic Loop Sensors and Results

Based on the need for a rapid, efficient technology for NDE inspection and evaluation of loop wires, FHWA defined four objectives for the GPR detection of traffic sensor wires, and achieved the following diagnostic results:

Computer-generated data image. This is a computer-generated image of SF GPR return intensities. The width position in feet is shown on the vertical axis and the scan position in feet on the horizontal axis.
Computer-generated data image. This is a computer-generated image of SF GPR return intensities. The width position in feet is shown on the vertical axis and the scan position in feet on the horizontal axis.
Shown here are migrated computer-generated field images of surface layer (top) and loop wire depth layer (bottom) signal returns from the SF GPR signal used to image pavement characteristics. The top photo shows the radar return signal information from the surface layer of the pavement, and the bottom photo shows the radar return signal information from the depth at which the loop wire lies.

Objective: Locate a loop wire sensor with a new SF GPR technique.

Result: The capability to locate loop wire sensors with a new SF GPR technique was demonstrated at slow vehicle speeds (<5 miles per hour, mi/h) (<8 kilometers per hour, km/h).

Objective: Perform test to determine whether the loop is functional, using both the passive approach (sufficient if the detector is connected to a sensor) and the active approach (necessary if the detector is not connected to a sensor).

Result: A passive probe measurement device developed during the study identified functional in-pavement loop wire signals and produced a flat signal response when no functional in-pavement loop wire sensors were present. A second probe measurement device with an active design was not used because SF GPR imaging provides the capability to detect disconnect--ed inductive loop wires.

Objective: Scan the details of a sensor using GPR and classify the record based on the results of the previous step, to be used as ground truth.

Result: The GPR successfully imaged functional in-pavement loop wire sensors identified by the passive probe measurement and confirmed them to be free from indications of defects or deterioration. A laboratory experiment successfully detected gaps in wires.

Objective: Perform detailed analysis of GPR data to assess the condition of a sensor, with the goal of detecting the causes of malfunction.

Result: The researchers analyzed the GPR condition data for functional sensors, indicating a continuous loop wire with no breaks. Additional field testing will further validate the SF GPR imaging algorithms, especially for nonfunctional inductive loop wires. With improved hardware capabilities, refinements of the MUSIC algorithms will be possible in the future to further increase the resolution, accuracy, and speed of acquisition.

The loop wire imaging method developed and implemented in the FHWA study currently meets basic detection needs for many loop wire evaluation applications, but the researchers expect that further advancements in technology and analysis will make the images sharper and more versatile.

The researchers achieved most of the basic objectives, including successful location and sensitivity measurement with the loop wire probe sensor, and imaging of loop wires and nearby pavement features such as loop wire sawcuts. The researchers detected and imaged loop wires, both in the laboratory and in the field with three turns of 14-gauge wire, corresponding to standard installation practices.

Primary goals that they did not achieve due to budget and time constraints included detection of defect and deterioration features in field loop wires and the lack of an active probe sensor to complement the passive probe sensor. Also, due to the wide emission spectrum of the SF GPR, the research team was required to use reduced power for the GPR relative to optimum levels and had to notch some emissions at specific frequencies to comply with emerging SF GPR rules. These changes reduced the resolution available from the advanced algorithms developed and will require additional research to overcome.

The researchers tested GPR over simulated pavement using treated canola oil, as shown here. By adding materials to the canola oil, it is possible to match the dielectric constant of the oil to that of asphalt, allowing lab testing of the radar responses of different objects in the simulated asphalt without having to lay real asphalt pavement over them.
The researchers tested GPR over simulated pavement using treated canola oil, as shown here. By adding materials to the canola oil, it is possible to match the dielectric constant of the oil to that of asphalt, allowing lab testing of the radar responses of different objects in the simulated asphalt without having to lay real asphalt pavement over them.
Shown here is laboratory test apparatus used to conduct parametric performance studies. By using asphalt roofing shingles composed of materials very similar to asphalt pavement, it is possible to match the dielectric constant to that of asphalt. This allows lab testing of the radar responses of different objects in the simulated asphalt of varying depths by changing the number of layers of shingles. Again, this can be done without having to lay real asphalt pavement over them.
Shown here is laboratory test apparatus used to conduct parametric performance studies. By using asphalt roofing shingles composed of materials very similar to asphalt pavement, it is possible to match the dielectric constant to that of asphalt. This allows lab testing of the radar responses of different objects in the simulated asphalt of varying depths by changing the number of layers of shingles. Again, this can be done without having to lay real asphalt pavement over them.

Conclusion

Some important goals remain before the new technology can be commercialized. The passive loop probe sensor measurement hardware performs well but will benefit from additional field hardening to make it practical for commercialization. SF GPR imaging methods are functional but need to be tested further on defective or deteriorating field loop wires. Improvements to the hardware and complementary software will bring the final elements of a commercially viable, higher resolution product to market.

Other needs are the ability to detect defects immediately after loop wire installation as part of a quality control procedure and to detect deterioration as quickly as possible after it has occurred to minimize the negative impacts of sensor downtime. These improvements would provide a way to prioritize repairs and maintain the efficiency of the transportation system.

"This research developed technologies that meet many loop wire evaluation needs, and it also developed capabilities and tools that can solve other highway infrastructure nondestructive evaluation problems, such as crack detection and crack density mapping of pavements as well as delaminations in pavement lifts," says FHWA's Peters. "Research and development is proceeding in some of these areas, such as the evaluation of this technology on asphalt and concrete pavements."

Photo. Shown here is a van with a GPR system attached to the front bumper. The vehicle is approaching a diamond-shaped sawcut in the pavement that is filled in with a gray-colored sealant, indicating the location of an embedded loop sensor.
Photo. This close-up shows the GPR system (mounted on the van's bumper) as it approaches a square-shaped sawcut in the pavement. The sawcut, which indicates the location of an embedded loop sensor, has been filled with gray sealant.
Researchers are using the Advanced Pavement Evaluation system (mounted on the front of the van) to scan loop wires embedded in the pavement. The loop wires are diamond-shaped (left) and square-shaped (right), as indicated by the adhesive-filled sawcut marks in the pavement surface.

James A. Arnold is a research electronics engineer on the Enabling Technologies Team in FHWA's Office of Operations R&D. He received a bachelor's degree in electrical engineering from the University of Delaware and a master's degree in electrical engineering from Florida Institute of Technology. His expertise includes civilian- and defense-related telecommunications, radionavigation, and spectrum management.

David R.P. Gibson, P.E., is a highway research engineer on the Enabling Technologies Team in FHWA's Office of Operations R&D. He has a bachelor's degree in civil engineering and a master's degree in transportation from Virginia Polytechnic Institute and State University. His expertise includes traffic sensor technology, traffic control hardware, modeling, and traffic engineering education.

Milton K. "Pete" Mills is an electrical engineer, now retired, from FHWA's Office of Safety R&D. He holds a bachelor's in electrical engineering from North Carolina State University and a master's from The Catholic University of America. At FHWA, he managed development and evaluation of systems for sensing vehicles from infrastructure.

Michael Scott, Ph.D., is a research engineer specializing in GPR for Starodub, Inc. Scott received a B.S. in mechanical engineering from Texas A&M University and an M.S. in engineering science and mechanics and a Ph.D. in civil engineering from Virginia Polytechnic Institute and State University. During the past 3 years, he has been the principal investigator on a SBIR project on SF GPR.

Jack Youtcheff, Ph.D., is the team leader for the Pavement Materials Team in FHWA's Office of Infrastructure R&D. He received a bachelor's degree in chemistry and a doctorate in fuel science from Pennsylvania State University. His expertise includes materials characterization, asphalt technology, and asphalt chemistry.

For more information, contact James Arnold at 202-493-3265 or james.a.arnold@dot.gov, David Gibson at 202-493-3271 or david.gibson@dot.gov, Milton Mills at 202-244-1136 or pete.mills@erols.com, Michael Scott at 202-493-3124 or mscott.ctr@dot.gov, or Jack Youtcheff at 202-493-3090 or jack.youtcheff@dot.gov.

The sponsoring agency of this project was FHWA's Turner-Fairbank Highway Research Center. This research was conducted through FHWA's SBIR Program. The U.S. Department of Transportation's SBIR Program is administered by the Research and Innovative Technology Administration (RITA)/John A. Volpe National Transportation Systems Center. The authors would like to acknowledge the efforts of Leisa Moniz, Linda Duck, and Darren Shaffer of the Volpe Center. Their help on the initiation and administration of this complex research was key to its success.

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration