Radio Frequency Identification Applications in Pavements

CHAPTER 7: EARLY DETECTION OF REFLECTION CRACKS

INTRODUCTION

HMA overlays are among the most common rehabilitation methods for a deteriorated pavement. A properly designed overlay increases the structural capacity of the pavement, inhibits water from penetrating into the subgrade, and restores friction resistance. Rehabilitation using HMA overlays is also advantageous because of the short construction time required, which minimizes travel delays for road users.

One of the most common distresses in HMA overlays is the development of new cracks in the overlay over the existing cracks or joints in the underlying deteriorated pavement. These are termed reflection cracks. Although reflection cracks can occur in HMA overlays over HMA pavements, they are most commonly experienced in HMA overlays over PCC pavements because of the regular transverse joints in the underlying layer. This latter type of reflection cracking is the focus here. Such cracks start at the bottom of the HMA layer at the joint locations between slabs and then propagate to the surface.

The two most critical movements of the concrete slabs are shown in figure 128. Horizontal movements are caused by changes in temperature; these cause Mode I or opening mode cracking. Vertical movements are caused by wheel loading; these cause Mode II or shear mode cracking. A good overview of Mode I and Mode II fracture mechanics can be found in Broek.⁽²⁶⁾

Figure 128. Drawing. Horizontal and vertical movement of concrete slabs causing Mode I and Mode II cracking.

Reflection cracks begin at the bottom of the HMA overlay and gradually grow toward the surface. Once the fully propagated reflection cracks are observed at the surface, the damage is done. Water can penetrate through the reflection cracks in the asphalt layer to the subgrade, decreasing its structural capacity. During low temperature conditions, the entrapped water in the cracks expands as a result of freezing, causing further crack growth. As these cracks get wider, smoothness decreases and damage to vehicles increases.

Early detection of reflection cracks before they reach the surface is especially important for warranty projects. Cui et al. provide an overview of the implementation of warranties on highway projects.⁽²⁷⁾ In 1990, FHWA initiated Special Experiment Project Number 14 to encourage State transportation departments to implement contracting techniques that would lower overall project costs. By 2004, more than 30 States had used warranties in some of their projects. The use of warranties helps reduce lifecycle cost, improve quality, and encourage contractor innovation.

The warranty period for HMA overlays has typically been 2 to 4 years. Reflection cracking may start during this period but may not become visible at the surface until later. The use of crack sensors can give the State transportation departments the ability to detect reflection cracking early, enabling them to make a more informed decision on holding the contractor responsible for either repairing the damage if still under warranty or allowing the department more time to set aside sufficient funds to cover damages.

Some research has been conducted on the use of RFID tags to detect cracks. However, most has not gone beyond the proposal stage. Morita and Noguchi envisioned the use of RFID tags to detect structural cracks in combination with electrically conductive paint or an electrically conductive sheet affixed to the structure and connected to the RFID tag using copper wire.⁽²⁸⁾ When a crack initiates, the conductive paint breaks, resulting in an increase in resistance. The crack width could theoretically be predicted based on the reader-tag communication because the electrical resistance changes with varying crack width. However, Morita and Noguchi did not pursue this application any further than the proposal stage. Rather than using an RFID tag and reader, they directly measured the resistance of the electrically conductive paint strip as the crack propagated through the structure.

Wood and Neikirk proposed a passive sensor to provide information about the condition of welded connections in steel structures after earthquakes.⁽²⁹⁾ Although they acknowledged the possibility of using RFID technology for this application, they did not explore this idea and instead pursued evaluation of electronic article surveillance stickers. Neither Morita and Noguchi nor Wood and Neikirk went further than hypothesizing that RFID technology could work in detecting fatigue cracks in concrete or steel structures.

Growth of an initiated crack due to cyclic loading is classically expressed using Paris' Law.⁽³⁰⁾ As shown in figure 129, Paris' Law relates the rate of crack growth to repeated loading cycles:

Figure 129. Equation. Paris's Law for crack growth.

Where:

C = crack length
N = number of loading cycles
A, n = fracture properties of HMA
ΔK = change in stress intensity factor during cyclic loading

The stress intensity factor is a measure of the driving force for crack propagation. It depends on the geometry of the structure, the loading conditions and magnitudes, and the crack length. The stress intensity factor is very difficult to measure. In practice, it is often replaced by a proxy such as the CMOD.

Lee et al. measured crack opening and crack growth in conjunction with tests to determine the optimum rubblized depth for preventing reflection cracking.⁽³¹⁾ The tests were conducted in Mode I bending loading on 12- by 8- by 2-inch (300- by 200- by 50-mm) asphalt slabs. Two PCC slabs were separated by 0.4 inches (10 mm) to simulate an open joint. The concrete slabs were, in turn, supported on an engineered synthetic rubber pad 1.2 inches (30 mm) thick to simulate the foundation conditions. A crack gauge was used to measure the horizontal deformation at mid-depth of the asphalt slab. The results of the horizontal deformation and vertical propagation of the crack were measured as a function of loading cycles. The crack was observed to have propagated about 60 percent through the 2-inch (50-mm) overlay at about 200,000 cycles. This corresponded to a horizontal deformation of about 0.018 inches (0.45 mm). This information provides some basis for estimating the range of displacement that must be considered during development of the crack sensor.

This chapter describes the conceptual development of an RFID-based wireless sensor to detect reflection cracks before they reach the surface of the pavement and become visible. One of the great advantages in trying to detect reflection cracks in pavements is that the potential crack locations are well defined (i.e., over the joints in the underlying slabs) as opposed to conventional fatigue cracks, which can develop almost anywhere. Because of the potentially very low cost of RFID crack detection systems, they could conceivably be placed at all potential reflection crack locations along an overlay rehabilitation project (i.e., at all joints in the underlying rigid pavement slabs).

The concept for the RFID-based wireless reflection crack sensor is to configure the RFID tag so that when a crack has grown to some critical length, the antenna circuit is broken, thus causing the tag to stop transmitting. Two possibilities where explored: a conductive paint that would crack at a certain strain, and an overlapped copper antenna "microswitch" that would open at a certain CMOD. Key factors in the sensor development are read range and survivability. In the field, the sensor must be readable through the thickness of the asphalt overlay and at least 1 to 2 ft (0.3 to 0.6 m) of air. The sensor must be able to survive the high temperatures and compaction stresses of paving operations.

As documented in this chapter, the RFID-based wireless crack detection system has been developed well beyond the basic conceptual stage. Prototypes were developed and successfully evaluated via laboratory testing and field trials. Unfortunately, project schedule and resources preluded comprehensive field evaluation over sufficiently long time periods to validate the ability of the system to detect early-onset reflection cracks under in-service pavement conditions.

CRACK DETECTION SENSOR DESIGN

The concept for early detection of reflection cracks is to adapt an RFID tag to "fail" in the sense that it can no longer be read once a significant crack initiates. To achieve this functionality, a short section of the antenna link is replaced with a conductive but frangible link. The modified RFID tag is placed above a joint in an existing PCC pavement layer before paving the HMA overlay. When the frangible antenna link fails at a prescribed strain level, the RFID tag stops working, indicating the initiation of a reflection crack. RFID tags could be placed in pairs at the bottom of the HMA overlay, one crossing the joint in the underlying PCC slab and one parallel and slightly away from the joint. Failure of the tag crossing the joint while the companion tag remains readable would be a clear indicator of early reflection crack formation.

The RFID tags used for this project are the Avery Dennison™ AD-223 shown in figure 130. These are UHF Gen 2 tags with the Monza® 3 chip, the newest chip developed from Impinj®. The chip is cut from the tag, leaving about 0.075 inches (2 mm) of the original antenna as shown in figure 131. The modified antenna is then connected to the 0.075-inch (2-mm) stub. The reader used for all of the tests in this chapter is the Mercury® 5 ThingMagic reader.

The efficiency of an antenna is governed by its electrical reactance. Reactance is a measure of the effect of capacitance and inductance on a time-varying current. A half-wave dipole has a reactance of zero. A shorter than half-wave antenna has a capacitive reactance, and a longer one has an inductive reactance. In this chapter, all the read range tests have been conducted for a half-wave dipole and shorter.

When shortening the tag from the half-wave dipole, capacitive reactance is introduced, which causes deterioration of the tag's read range performance. This is because part of the signal is "wasted" as stored electromagnetic energy in the reactive capacitance.

Note: Overall tag length is approximately 6 inches (150 mm).
Figure 130. Photo. RFID tag AD-223.

Figure 131. Photo. RFID chip with 0.075 inches (2 mm) of antenna on each side.

In practice, the length of the RFID dipole antenna can be shortened by introducing a frangible link that breaks at a target strain or displacement level. Two types of frangible antenna links were investigated. The first one is based on electrically conductive paint designed to break at a critical strain level. The second uses slightly overlapped metal strips to form a microswitch in the antenna that opens-i.e., breaks the antenna connection-at a critical CMOD. In both cases, the modified RFID tag is required to work/give a readable signal from a given distance before a crack has initiated and give no readable signal from the same distance after a crack has initiated but not yet appeared on the surface of the pavement.

Conductive Paint Antennae

Read Range

A conductive paint was initially thought to be the best choice for the frangible link, in large part because of the ease and flexibility of its application. The read ranges for three possible antenna configurations were evaluated: a loop antenna, a dipole antenna, and a C-shaped antenna.

As a first attempt, Aquadag® E™, a conductive carbon paint from M.E. Taylor Engineering, Inc., was used to paint a 0.2-inch (5-mm)-wide loop antenna connected to the RFID chip from the Avery Dennison™ tags. The loop antennae were painted on a flexible plastic substrate made from conventional transparency stock. Small dabs of a thicker silver conductive paint were used to create better connections between the Aquadag® E™ paint and the chip. Three tags were made with loop antennae of one wavelength, two wavelengths, and three wavelengths as shown in figure 132 through figure 134.

Figure 132. Photo. One-wavelength loop antenna.

Figure 133. Photo. Two-wavelength loop antenna.

Figure 134. Photo.Three-wavelength loop antenna.

The loop antennae were 0.2 inches (5 mm) wide and 12.9, 25.8, and 38.7 inches (328, 656, and 984 mm) in total length based on the wavelength calculation shown in figure 135 :

Figure 135. Equation. Wavelength calculation.

Where:

λ = wavelength
c = speed of light = 1.18x10¹⁰ inches/s
v = frequency = 915 MHz for UHF tags.

The maximum read distance was evaluated for all tags, with results shown in table 27. All the readings were taken over 20-s intervals on the T (transmit) side of the antenna.

Table 27 . Read range for the carbon paint loop antenna tags.

The maximum read range for the loop antennae through air was 1.6 ft (0.5 m), which is marginal in the field when the tag is placed at the bottom of the HMA overlay. In an effort to extend the read range, the loop antennae were replaced with dipole antennae. Typically, dipole antennae are created by two quarter-wavelength conductors extending from a central connection; this was done using the conductive paint. As with the loop antennae, silver paint was used to create better connections between the Aquadag® E™ paint and the chip. Figure 136 illustrates one of the fabricated dipole antennae and RFID chip. The maximum read distance for the dipole tag was approximately 4 ft (1.2 m).

Figure 136. Photo. Dipole carbon paint antenna.

The read performance of the half-wavelength dipole tag was evaluated as one of the antenna legs was shortened to simulate breakage by a reflection crack. A series of 0.2-inch (5-mm) cuts were made on one side of the dipole antenna, and the effect on maximum read range was evaluated. The results are shown in table 28. It can be seen that the maximum read distance starts dropping significantly after the antenna is shortened to about half of its original length. If the antenna is cut at about 0.4 inches (10 mm) away from the chip, the tag cannot be read at a distance beyond about 1.3 ft (0.4 m). This would imply that if the antenna were kept at a distance of 1.6 ft (0.5 m) away from the embedded tag, the reader would be able to detect that the antenna had broken as a result of reflection cracking. A second dipole tag fabricated and tested in the same manner showed similar results.

Close examination of the read ranges for the modified tag after cutting one side of the antenna (table 28) shows that the read range did not drop as precipitously as expected as the antenna length was shortened. Even when there was only about 0.8 inches (20 mm) left on one side the tag, it could still be read from about 2.0 to 2.6 ft (0.6 to 0.8 m). The larger the drop in read range after the antenna is broken, the easier it will be to detect whether a crack has initiated. For this reason, additional tests were conducted in which one side of the dipole antenna was already snipped at 0.8 inches (20 mm) as shown in figure 137. Read range was then measured as the other side of the antenna was progressively shortened. The antenna for these tests was made of silver conductive paint from Engineered Conductive Materials, LLC. The reason the tests were not continued with the carbon paint is because when one side of the antenna is only about 0.8 inches (20 mm) in the half-wavelength carbon dipole, the maximum read distance is approximately 2.3 ft (0.7 m). This particular silver paint was used because it was thought that it might give better read ranges.

Table 28 . Read range for the carbon dipole tag with one side shortened.

Figure 137. Photo. Silver dipole antenna with one side snipped at 0.8 inches (20 mm).

The results from these tests, summarized in table 29, show that it would be best to have both sides of the antenna break at a distance of 0.8 inches (20 mm) or less from the chip. Before failure, this tag could be detectable at a read range of about 4.2 ft (1.3 m); after failure, the read range is about an order of magnitude smaller. Another similar test with the same silver paint was conducted for another possible configuration, a C-shaped antenna instead of a straight dipole, as shown in figure 138. The maximum read range for this tag with no cuts to the antenna was about 8.2 ft (2.5 m). When one side of the antenna was trimmed to within 0.24 inches (6 mm) of the chip, the read range for the tag dropped to only about 1.5 ft (0.45 m). When both sides of the antenna were trimmed to within 0.24 inches (6 mm) of the chip, the tag could not be read at any distance. The reason for the C-shaped configuration is to cause the antenna to break simultaneously on both sides, which would give a much lower read range once the crack had initiated than if it broke only on one side. However, the C-shaped antenna also keeps the size of the tag comparatively small, raising the concern that it cannot cover the entire area where a reflective crack might occur. This does not make it a good candidate for the crack sensor application.

Table 29 . Read range for the silver dipole antenna with one side fixed at 0.8 inches length while the other side is progressively shortened.

Figure 138. Photo. C-shaped silver dipole antenna.

The high conductivity of the silver paint allows evaluation of antenna effects for the "best case"-i.e., longest read range-conditions. However, as will be shown in a later section, the carbon paint was the only coating that broke anywhere near the target strain level and as a consequence, it is the coating that must ultimately be used to modify the RFID tag. For this reason, more modified tags having carbon paint dipole antennae were constructed and tested. As expected, the read ranges of the tags with the carbon paint were not as long as for the silver dipole antennae because of the lower conductivity of the carbon compared with the silver. The maximum read range of the tags with the carbon dipole antennae in these new tests was only about 1.6 ft (0.5 m) compared with about 3.9 ft (1.2 m) for the silver. Although in earlier tests (table 28), the carbon paint dipole also read up to 3.9 ft (1.2 m), because of variations from tag to tag it would be more cautious to base the decisions on the worst results. To capitalize on the properties of both materials, tags with hybrid dipole antennae combining both silver and carbon paint were fabricated. The 0.8 inches (20 mm) of the antenna closest to the RFID chip was painted with carbon and the rest of the dipole length was painted with silver paint, as shown in figure 139. It was expected that this tag would have a maximum read range close to that of the pure silver dipole while at the same time have a frangible link at the critical location close to the chip that would fail at the target strain level. Read range test results for this hybrid silver and carbon dipole antenna tag are shown in table 30. The maximum read range before trimming the antenna length was 5.9 ft (1.8 m), comparable to the pure silver dipole antenna. The read range of the tag dropped by approximately 40 percent to 3.3 ft (1 m) or less as the remaining dipole leg was shortened to 0.8 inches (20 mm).

Figure 139. Photo. Silver and carbon dipole antenna.

Table 30 . Read range for the silver and carbon dipole antenna as one side is shortened.

The conclusion from these read range tests is that the dipole configuration is best for the crack sensor. Also, because of its higher initial read range before breakage and low read distance after the carbon paint is broken, the hybrid silver and carbon dipole antenna provides the best configuration for the reflection crack sensor.

Fracture Strains

Initially, the critical strain at which the conductive paint was targeted to break was estimated to be about 1,000 µε based on elastic strain distributions at the bottom of an asphalt layer. Consequently, tests were conducted to find a conductive paint that broke at about that strain level. Two carbon conductive paints, two conductive silver paints, and one conductive silver epoxy were tested in the laboratory to determine the strain at which they would break. Details of these paints are given in table 31. They were strained in a four-point bending test using an Instron® Model 1331 compression machine. A strip of each of the paints was painted on a plastic film. The painted film was then glued to the bottom of the beam. The adhesive used for this was SG496 (Omega® Engineering Inc.), a methyl-based cyanoacrylate 1-part glue commonly used for strain gages. A beam was fabricated from 7075-T6 aluminum, a material that remains elastic at strains up to 4,600 µε. The calculated beam dimensions to reach the target strain of 1,000 µε under reasonable applied load levels are shown in figure 140. An ohmmeter was used to measure the resistance in the painted strip of conductive paint to check for breakage. The complete test setup is shown in figure 141. Although the required critical strain was 1,000 µε, the tests were run to the elastic limit of 4,600 µε. Unfortunately, none of the cured conductive paints broke at this strain level.

Table 31 . Description of the conductive paints tested to determine failure strains.

1 inch = 25.4 mm
1 ft = 0.305 m
Figure 140. Drawing. Beam dimensions for the four-point bending test.

Figure 141. Photo. Thin film test setup.

As a next step, another set of tests was developed to localize the strains similar to what occurs across the mouth of a fracture. Two polymethyl methacrylate (PMMA, trade name Plexiglas®) sheets were attached along the bottom of the beam; the sheets just touched each other, simulating the joints in the PCC slab in a pavement. This test specimen setup is shown in figure 142. Thin strips of the conductive paints were then painted across the simulated joint. The 18-percent silver conductive paint did not break. The carbon Aquadag® E™ conductive paint broke at about 300  . Because the Aquadag® E™ paint actually broke, albeit at strain levels below the target, it was selected for the tag antenna studies described in the previous section.

Figure 142. Photo. Painted Aquadag® E™ on the two abutting PMMA sheets.

Based on the read range tests from the previous sections, it was determined that the antenna needed to break close to the chip for the read distance to shorten significantly. In the PMMA plate test setup, the antenna would fail directly over the "pre-cracked" plates. However, this implies that in the field, one would need to paint directly on the top of PCC slabs or the underside of the overlay, which is not practical. A carrier for the paint was needed to force the maximum strains to occur next to the chip. Therefore, a new test configuration was developed. A 4- by 9-inch (100- by 230-mm) polycarbonate polymer (trade name Lexan) sheet was fabricated with a 3.5-inch (90-mm)-diameter hole in its center to act as a stress/strain raiser. The 3.5-inch (90-mm)-diameter hole in the 4-inch (100-mm)-wide Lexan sheet causes the elastic stresses and strains to increase by a factor of 10 in the ligaments at the edges of the hole. The plate length was chosen as 9 inches (230 mm) so that the Lexan sheet would fit between the middle rollers where the bending strains are constant in the four-point beam loading system. A continuous 0.2-inch (5-mm) strip of Aquadag® E™ was painted along one edge of the plate (including over the ligament on one side of the hole), and a dipole tag with a 0.2-inch (5-mm)-wide antenna strip of Aquadag® E™ was painted along the other, as shown in figure 143. The dipole tag was painted so that the chip was slightly offset from the highest strain location in the ligament at the edge of the hole in order to break the antenna close to the chip. Two ohmmeters were used to monitor the conductivity in the painted strips to determine breakage. The antenna of the dipole tag broke about 0.2 inches (5 mm) away from the thinnest part of the ligament at a strain of about 12,000   (after adjustment for strain magnification effects). The antenna broke very close to the chip as expected, after which the tag could not be read at long range. The continuous strip of paint on the other side of the plate did not break during the test.

Figure 143. Photo. Painted dipole tag on one side and painted Aquadag® E™ only on other side of the polycarbonate sheet.

Based on insights from an additional literature search on measurement of CMOD, a reasonable estimate is that a reflection crack will have initiated and grown partially but not completely through an overlay at a CMOD on the order of 0.04 inches (1 mm). A CMOD of 0.04 inches (1 mm) over a gauge length of 4 inches (100 mm) corresponds to a strain of 10,000 µε. These strain levels cannot be achieved with the bending beam apparatus except when using the polycarbonate sheet with the hole for concentrating and magnifying the stress and strain levels. However, this apparatus was problematic in practice because of difficulties in getting a secure attachment to the bottom of the aluminum beam. For these reasons, direct tension tests were conducted to determine the failure strains of the paints. Direct tension dogbone specimens were fabricated from polycarbonate with the dimensions shown in figure 144. Two tests could be conducted simultaneously on the same specimen by coating the opposite sides. The test setup is shown in figure 145 and figure 146. The tests were run in displacement control mode at a rate of 0.005 inches (0.127 mm) per min. Adhesion problems between the paint and the polycarbonate were observed in some of the previous bending tests; the paint in some cases would just fall off the polycarbonate sheet. To remedy this problem, the Aquadag® E™ carbon conductive paint was painted on a strip of watercolor art paper that had been glued to the polycarbonate specimen using strain gauge adhesive. The strips of paint were 0.2 inches (5 mm) wide. Figure 147 shows a typical strip of conductive paint on the polycarbonate specimen with wires for connecting the ohmmeters.

1 inch = 25.4 mm
Figure 144. Drawing. Dogbone specimen dimensions.

Figure 145. Photo. Direct tension test setup.

Figure 146. Photo. Direct tension test specimen attachment.

Figure 147. Photo. Conductive paint on the dogbone specimen.

The test was continued to a strain of 26,000 µε, well above the strain at which reflection cracks should initiate in an HMA overlay. The carbon paint did not break. There was some concern that the watercolor paper might be creeping and that therefore the carbon paint was not experiencing the true applied strains. To account for this possibility, the test was repeated with two strips of carbon paint applied directly on the polycarbonate specimen after its surface had been roughed with extra-fine sandpaper to enhance adhesion. The painted strips were 0.2 inches (5 mm) wide, similar to the first test. This test was continued to a strain of 25,000 µε, and the carbon paint still did not break, although some small cracks were visible.

One last tension test was conducted on the Aquadag® E™ carbon paint and the 18-percent silver conductive paint. The width of the painted strips was narrowed to 0.1 inch (2 mm) for this test on the assumption that a crack would propagate more easily over the narrower width. The test was run to 30,000 µε without any breakage/conductivity fault in either the carbon or the silver paint.

The overall conclusion from all these tests is that while the carbon and the silver conductive paints can be used to make suitable antennae for the RFID chip, none was reliably brittle enough to serve as a frangible link. Therefore, the approach was changed from a strain-based brittle material concept to a simple displacement-based mechanical microswitch scheme for interrupting the antenna circuit, as explained in the following section.

Antennae With Mechanical "Microswitch"

The final crack sensor design was based on modifying the antenna of an existing RFID tag using electrically conductive metal strips that overlap just enough so that the antenna mechanically opens or disconnects when a crack occurs. Essentially, the RFID crack sensor will behave like a switch, so when a crack occurs, the antenna circuit is broken and the tag turns off and gives no signal

It was decided that the crack sensor should be developed so that it detects cracks that have propagated through approximately 60 percent of the overlay. This corresponds to a 0.018-inch (0.45-mm) overlap in the antenna microswitch.⁽²⁶⁾ One of the main reasons for this decision was to keep the crack sensor manufacturing as simple as possible. As the target overlap gets small, the fabrication becomes more difficult.

A 36 gauge (0.005-inch (0.125-mm)-thick) copper sheet was cut into strips about 0.1 to 0.15 inches (2 to 3 mm) wide for creating the dipole antenna of the RFID tag. One side of the dipole was connected to the 0.1-inch (2-mm) stub of the original antenna using the two-part electrically conductive silver epoxy from M.E. Taylor Engineering, Inc., to assure a good connection. The other side of the dipole was fabricated to have a 0.018-inch (0.45-mm) overlap of the thin copper sheet over the stub of the original antenna

To determine the best configuration of the dipole, read range studies were conducted for different antenna lengths as was done earlier with the conductive paint antenna. These tests were initially conducted in the Pavement Materials Laboratory at UMD. Because of inconsistencies in the indoor readings thought attributable to RF reflections from metal interior objects, the read range studies were moved outdoors to parking lot E behind the Engineering Laboratory Building at UMD to minimize any interference (figure 148). It was expected that the results would be less variable and more reliable, and in addition, the outdoor conditions better represent the actual field scenario.

Figure 148. Photo. Read range tests in parking lot E.

The first set of tests was done for a symmetrical antenna, starting with a half-wavelength dipole with two sides of 3.2 inches (82 mm) each and then continuing by shortening each side of the dipole. For each length, readings were taken with one side alternately being connected and disconnected. The dipole tag was sandwiched between two thin polycarbonate pieces to keep it in place and to simulate the expected encapsulation of the sensor in the field. To create a good connection at the mechanical overlap, a wooden toothpick was sandwiched between the polycarbonate pieces on top of the overlap (figure 149). The readings were taken in four directions, north, south, east, and west. The tag was optimally placed in front of the transmit side of the antenna.

Figure 149. Photo. Symmetric dipole used in the read range study.

The results for the symmetrical dipole read range are shown in figure 150. The maximum read distances are very similar in all four directions. Except for a slight anomaly at 6.7 inches (170 mm), the maximum read distance consistently decreased as the total dipole length shortened. Some possible suitable configurations chosen from these results were the following:

• 4 inches (100 mm) dipole length;
  Connected read at 7.6 ft (2.3 m);
  Disconnected read at 1.6 ft (0.5 m).
  
  
• 3.5 inches (90 mm) dipole length;
  Connected read at 5.2 ft (1.6 m);
  Disconnected read at 1.3 ft (0.4 m).
  
  
• 3.1 inches (80 mm) dipole length;
  Connected read at 4.4 ft (1.35 m);
  Disconnected read at 1 ft (0.3 m).
  
  

Next, a read range study was conducted with one side of the dipole fixed at 3.2 inches (82 mm) and the side other starting at 2 inches (50 mm) and then progressively shortened. The longer 3.2-inch (82-mm) antenna leg contained the overlapped microswitch. The same testing method as for the symmetric antenna readings was followed. Readings were taken with the long side connected and disconnected. The dipole tag was put between two thin polycarbonate pieces and a toothpick was used as before to ensure a good connection at the overlapped microswitch (figure 151). The readings were taken in four directions, north, south, east, and west. Two replicate tags were tested to evaluate reproducibility. Again, the tags were placed in front of the transmit side of the antenna for all readings.

1 inch = 2.54 cm
Figure 150. Graph. Read range for the symmetric dipole in each direction connected and disconnected.

Figure 151. Photo. Asymmetric dipole used in the read range study.

The results of these tests are shown in figure 152 through figure 155. The read ranges for the four different directions were very similar, as were the readings from the two replicate tags. Based on these results, it was decided that the best dipole length for this application was 3.2 inches (82 mm) on one side and 0.6 inches (15 mm) on the other. This configuration, when connected, could be read at more than 7 ft (2.1 m) but only at less than 1 ft (0.3 m) when disconnected. This difference in read ranges is large enough for the antenna to be read from a moving vehicle when connected and not read from that same distance when disconnected. The 7-ft (2.1-m) read range when connected is actually greater than required in the field; however, this adds a safety margin to help ensure that the tag can be read even after possible signal attenuation through the HMA overlay.

1 inch = 2.54 cm
Figure 152. Graph. Read range for tags disconnected and connected in the north direction.

1 inch = 2.54 cm
Figure 153. Graph. Read range for tags disconnected and connected in the south direction.

1 inch = 2.54 cm
Figure 154. Graph. Read range for tags disconnected and connected in the east direction.

1 inch = 2.54 cm
Figure 155. Graph. Read range for tags disconnected and connected in the west direction.

PROTOTYPE DESIGN

Fabrication

Because the sensor must be placed on the surface of the existing pavement prior to the overlay, for practical field applications the RFID tag needs to be encapsulated to protect it from high temperatures and stresses during paving operations. However, the encapsulation must be able to transmit displacements to the antenna so that eventually the overlapped microswitch can open.

Polycarbonate (Lexan™) was used for the encapsulation. Its melting temperature is 513 °F (267 °C) and its glass transition temperature is 302 °F (150 °C), which are sufficient for paving conditions. The glue used for attaching the RFID tag to the polycarbonate and for assembling the polycarbonate parts was the SG496 strain gauge adhesive used previously in the laboratory studies.

The different parts used for encapsulating the tag are shown in figure 156, and the actual encapsulated tag is shown in figure 157. A 12-inch (300-mm)-long by 0.4-inch (10-mm)-wide piece of 0.095-inch (2.4-mm) thick polycarbonate was used as the top mounting plate for the RFID tag. This part transmits the crack opening movements in the overlay to the antenna. The 12-inch (300-mm) length is sufficient to span the area where the reflection crack might initiate. A longitudinal groove deep enough to fit the RFID tag is cut into the mounting plate and the dipole ends are glued in the groove. Two small 0.6- by 2.0-inch (15- by 50-mm) cross pieces of 0.18-inch (4.5-mm)-thick polycarbonate are glued to the end of the top mounting plate. The grooved polycarbonate bottom protection piece covers the dipole, and the mounting plate and is about 1.2 inches (30 mm) shorter than the top mounting plate. The bottom plate is glued only in one place to the top piece; the bottom plate provides a protection cover but does not interfere with the movements in the top plate and the sensor. Only the two cross pieces of the assembled sensor are to be affixed to the underside of the asphalt overlay. Once a crack initiates, these cross pieces begin to spread, putting the sensor into tension and decreasing the overlap in the microswitch until it opens and the RFID tag stops giving a readable signal.

Figure 156. Photo. Parts for encapsulating the RFID tag.

Figure 157. Photo. H-sensor for reflection crack detection.

Field Survivability and Read Range

Field tests were conducted to determine the survivability of these tags when subjected to high temperatures and compaction stresses during paving. Another objective of the field tests was to finalize the length of the antenna by evaluating read ranges of various lengths of tags through an asphalt overlay. This would give a better understanding how the tags perform through HMA as opposed to just air.

Eighteen RFID tags were prepared for the field tests. Two of them were encapsulated in the H-sensor configuration (figure 157). The other 16, however, were simply sandwiched between two 0.4-inch (10-mm)-wide polycarbonate plates, one of which was grooved so that the chip would not be damaged. These tags were used to evaluate the optimum antenna length. To determine which antenna length would be the best for the crack sensor design, the maximum read distances for the connected case and disconnected case must be evaluated. The 16 tags were therefore paired in groups of two in which one tag was in the connected condition and the other disconnected. For simplicity, in the connected case, the copper antenna on either side of the chip was attached using conductive silver epoxy instead of overlapping one side. Six different lengths, as listed in table 32 , were evaluated. A 2-inch (50-mm) HMA milling and overlay project on Campus Drive at UMD was selected for the field trials. The encapsulated tags were placed longitudinally at 2-ft (0.6-m) intervals, as shown in figure 158. Quick set epoxy putty was used to make sure the tags were not displaced during the paving operations (figure 159 through figure 161).

Two days after the paving was completed, a read range study was conducted (figure 162). The number of successful reads over a 20-s interval was recorded. Although some tags were run over by equipment during the paving operations, only 1 tag of 18 failed to give any readings. The results for the maximum read distances for the tags are given in table 32. The 5-ft (1.5-m) height entries do not necessarily represent the maximum distance at which the tags could be read; 5 ft was just the highest distance above the pavement that could be achieved in the field. Establishing the actual maximum read distance beyond 5 ft was not necessary because a bumper-mounted reader antenna would not be at a higher distance than that above the pavement. Based on the results, the symmetric dipole with a 1.6-inch (40-mm) antenna on each side of the chip was determined to be the best configuration. When connected, it read at least 5 ft (1.5 m), and when disconnected, it read at only about 0.6 ft (0.2 m). If the reader antenna were mounted anywhere from 1.6 to 5 ft (0.5 to 1.5 m) above the pavement, the tag would give a signal when there was no crack and would not read once there was a crack in the pavement.

Figure 158. Photo. Encapsulated tags placed in the roadway prior to paving.

Figure 159. Photo. Close-up view of one of the encapsulated tags.

Figure 160. Photo. Two H-sensors in the pavement prior to paving.

Figure 161. Photo. Paving in progress.

Figure 162. Photo. Read range test setup.

Table 32 . Field test results.

Laboratory Evaluation

A laboratory evaluation was undertaken to demonstrate that the crack sensor design can detect cracks in a 2-inch (50-mm) thick asphalt overlay. The end cross pieces of the encapsulated RFID tags were glued to the bottom of asphalt beams. The idea was that when the beam experienced a crack during three-point bending, the crack sensor microswitch would disconnect when the crack reached about 50 percent or more of the height of the beam.

The asphalt specimens were prepared at the standard fatigue tests size of 15 inches (380 mm) length, 2.5 inches (63 mm) width, and 2.0 inches (50 mm) height. The sides of the asphalt beams where painted white to make it easier to observe cracking. The ends of the copper antenna were glued to the grooved polycarbonate plate as shown in figure 163. Copper wires were glued to each side of the RFID tag antenna using conductive silver epoxy and then connected to an ohmmeter to measure resistance and determine at what point the conductivity was lost. The polycarbonate plate to which the RFID tag was glued was kept at a length equal to the effective length of the beam during bending. For the laboratory tests, the protective top cover for the tag was unnecessary and not used. However, to make sure the antenna overlap remained in place with a good connection, small polycarbonate pieces were glued across the groove over the microswitch.

Figure 163. Photo. RFID tag glued to the asphalt beam.

The specimens were then loaded for three-point bending using an Instron® compression machine. The tests were run in displacement control mode at a rate of 0.2 inches (5 mm) per minute. For the first three tests, the overlap of the copper antenna to the existing antenna was kept at about 0.02 inches (0.5 mm) as discussed previously. However, in all three instances the microswitch opened prior to a crack initiating. This could have been the result of several factors, including the slow loading rate used during the three-point bending, the properties of the beam, and other details of the test setup. For this reason, the third test was continued until the crack propagated 50 percent through the beam thickness even though the antenna was disconnected. Based on this test, it was determined that an antenna overlap of about 0.06 inches (1.5 mm) was required.

The next step was to run a test that would more closely represent the use of the crack sensor in the field. Instead of measuring resistance, the reader and antenna were positioned close to the bending machine. Although even in air the 1.6- by1.6-inch (40 by 40 mm) RFID tag could not be read from further away than 5 ft (1.5 m), in the laboratory during the test it was read at a distance more than 6 ft (1.8 m). This was most probably because of wave reflections coming from the metal machines surrounding the antenna. The antenna was placed 6 ft (1.8 m) away from the RFID tag. The connection was lost at a vertical displacement of the asphalt beam of about 0.8 inches (20 mm). At this point, the crack had propagated about 50 percent through the 2-inch (50-mm) beam thickness at a CMOD of between 0.1 to 0.15 inches (2 to 3 mm) (figure 164).

Figure 164. Photo. Crack propagation when the RFID tag stopped reading.

CONCLUSIONS AND RECOMMENDATIONS FOR REFLECTION CRACK DETECTION

Conclusions

HMA overlays are the most common rehabilitation methods for deteriorated pavements, and the most common distresses in these rehabilitated pavements are reflection cracks. The development of a crack sensor for detecting early onset of reflection cracks would be beneficial to State transportation departments because they could hold the contractor responsible for repairs in warranty construction projects, and it would help them plan better for future rehabilitation activities.

The reflection crack sensor concept is to implement a switch in the antenna circuit of the RFID tag: when there is no crack the RFID tag gives a signal that can be read at a distance, but after a crack forms, the signal becomes too weak to be read. The first design for the crack detection sensor employed conductive paint that would rupture at strain levels corresponding to a certain CMOD. The conductive paint dipoles had good read ranges, reaching about 6 ft (1.8 m). However, the drawback with the conductive paint was that none of the evaluated paints broke at the desired strain. The Aquadag® E™ carbon paint and the 18-percent silver paint survived strains up to 25,000 µε without breaking.

The second sensor design was based on the use of copper antennae on both sides of the RFID tag, with one side having a mechanical overlap microswitch with the existing antenna cut from the original AD-223 RFID tags. A crack causes the antenna to be pulled in tension, eventually causing disconnection at the overlap microswitch and thus an open circuit in the antenna.

The RFID tag was encapsulated between polycarbonate protective covers to protect it from the high stresses and temperatures of paving operations. The ends of the copper modified antenna were glued to the polycarbonate mounting plate, which is then affixed to the asphalt overlay. The plates both protect the tag and transmit displacements to the tag.

A survivability check of the encapsulated tags was conducted by embedding them in a 2-inch (50-mm)-thick overlay. Seventeen of 18 tags tested survived the paving operations; the tags could be read at distances of more than 5 ft in the optimal antenna configuration. Subsequent laboratory beam testing confirmed the concept of the overlapping copper antenna microswitch for the crack detection sensor. The sensor successfully detected the crack once it had grown to about 50 percent of the height of the beam.

As documented in this chapter, the RFID-based wireless crack detection system has been developed well beyond the basic conceptual stage. Prototypes were developed and successfully evaluated via laboratory testing and field trials. Unfortunately, project schedule and resources preluded comprehensive field evaluation over sufficiently long time periods to validate the ability of the system to detect early-onset reflection cracks under in service pavement conditions

Recommendations

The tests to determine the amount of the overlap were conducted in a three-point bending configuration with a static load applied in the middle of the beam. This results in vertical deflections well beyond those expected in the field. Future tests should be conducted using a Texas Overlay Tester so that the beam fails in a way more similar to how the pavement would fail. Through these tests, the CMOD as a function of vertical crack propagation can be determined more accurately, and the overlap can be adjusted as needed. The flexibility of the design presented in this chapter allows one to use the sensor to detect small cracks or strains (less overlap) or detect larger cracks or strains (more overlap)

Laboratory testing using the Texas Overlay Tester should be followed by field trials on in-service pavements. The duration of these field trials would need to be on the order of 5 to 10 years to allow sufficient time for reflection cracks to initiate and propagate.