Use of ultrasonic tomography for the assessment of cri(cid:20)cal aspects of doweled joints in concrete pavements

The cri2cal (design) parameters to be controlled for concrete pavements construc2on, in addi2on to the concrete strength, are slab`s thicknesses and doweled bars posi2on at contrac2on joints, as they aﬀect hugely structural performance, and, if not properly built may cause early ruptures. Construc2on compliance of such parameters is nowadays assessed through nondestruc2ve techniques (NDT), such as ultrasonic low-frequency waves. Alignments of inserted doweled bars at contrac2on joints are analyzed herein for actual slabs as well as how thickness aﬀects such posi2on. Tests were carried out over an airport hangar apron a<er it has been built. Twenty-ﬁve concrete slabs were assessed using ultrasonic tomography allowing to verify disconformi2es of thicknesses and bars ver2cal and horizontal misalignments, with fair accuracy on measuring devia2on angles of bars.


INTRODUCTION
The inspection of concrete pavement thickness is usually performed through sample`s coring, allowing to analyze the concrete slab thickness based on a limited number of samples (due to the extraction costs and time-consuming). The same can be stated to past methods for checking dowel bar's location, length, and amount in transversal joints, requiring destructive inspections when under suspicious situation. In addition, since they are destructive methods, their productivity is low. Also, demands a subsequent repair at any positions wherever the samples were extracted.
Moreover, one should note that the limited number of samples obtained at each slab when coring the pavement might impair the assessment of constructive failures such as lack of uniformity in the slabs thickness. It occurs, mostly, because of the standards permissiveness, in some countries, regarding the employment of techniques and equipment uncapable to ensure a suitable 'latness for the before built base layer, usually a roller compacted concrete (RCC) or cement-treated crushed stones (CTCS). Balbo (2014), for instance, presented (through sample's coring) a signi'icant variation in the slab thickness of a concrete pavement placed with a slipform paver; the thickness ranged from 175 mm to 314 mm, whereas the design speci'ication was 230 mm. The author states that such variations are associated to the base material (RCC) spreading with a motor grader, leading to base layer 'inishing 'laws and super'icial unevenness.
Also, according to Balbo (2014), it is necessary to 'ix ways to mitigate the concrete slab thickness variation, seeking to reduce the slabs degradation risk, since the fatigue phenomenon is quite sensible to thickness variations. Effective load transfer between joints is of main concern to airport infrastructure and it depends not only on bar alignment as well as on the concrete temperature and the underneath asphalt or compacted concrete base to bear and transfer the aircraft gear loads (Federal Aviation Administration, 2021). Paramount concern is also given to slabs full doweled (both longitudinal and transversal joints) when movement locking could be even more critical compared to slabs with conventional longitudinal tie bars.
Concerning the inspection of dowel bars position, Gancarz et al. (2015) describes that, in the past, albeit the dowel bars misalignment (Figure 1) was considered matter of concern, the transportation agencies did not conduct inspections to check the dowel bars orientation after concrete placement. According to Tayabji (1986) and Rao et al. (2009), inspections of dowel bars location after construction were not conducted due to the lack of practical protocols. The main concern about the dowel bars proper positioning is related to the risk of locking the joint. Although the incidence of one dowel bar sharply misaligned is not reason of concern, if several dowel bars in the same joint present signi'icant misalignments, the risk of joint locking is increased (STUGES et al., 2014).
According to the Federal Highway Administration (FHWA, 2018) the locking of an individual joint would not affect the pavement performance. However, the locking of successive joints increases the risk of breaking the slab due to movement restriction during the concrete expansion and contraction ascribed to seasonal and daily thermal and moisture oscillations, resulting, consequently, in the malfunction of the system, which may lead to transverse cracks parallel to the joints.
For these reasons, seeking to reduce the incidence of problems related to joints malfunction, most of the road agencies usually establish tolerances on the precision of dowel bars positioning and alignment. However, according to the American Concrete Pavement Association (ACPA, 2006), despite these tolerances can be considered strict, these patterns are not usually related to the pavement performance in 'ield.
In Brazil (DNIT, 2013) deviations of the dowel bar end of ±1% based on its length are allowed as long as at least two thirds of the dowel bars, at the same joint, do not present deviations higher than ±0.7%. Such limits arbitrarily adopted are too conservative when compared to the limits admitted in countries with strong tradition of concrete pavement construction, like United States and Germany, as depicted in Table 1. Tolerances for dowel bars misalignments in Brazil are like the British ones at the early 1970s (PARMENTER, 1973). According to the author, the dowel bars misalignment tolerances before the concrete pouring were: 1% of rotational misalignment (horizontal or vertical) of one third of the bars at the joint; the other two thirds of bars should be within a tolerance of 0.5%; and no dowel bar should differ in alignment from the adjacent one in more than 1.0%. Parmenter (1973) veri'ied the position of dowel bars in 'ield through removal of concrete still fresh to reevaluate the bars alignment and coring samples of hardened concrete with 100 mm of diameter at both ends of the bars. In this study, the author pointed out the tolerances afore mentioned were overly rigid and misalignments of 3% and 4% would be acceptable. This already demonstrated the need for reviewing the acceptance criteria based on 'ield performance data.
In the present study, the MIRA ultrasonic tomography device was used to assess a newly built concrete pavement in an airport apron, mainly aiming to evaluate the equipment applicability to pinpoint the dowel bars, besides measuring the concrete slabs thickness to improve the understanding about the reasons for bars misalignments.

MIRA ULTRASONIC TOMOGRAPHY DEVICE
Among the current techniques for non-destructive tests (NDT), ultrasonic tomography stands out for enabling to check concrete slabs thickness, inspect the position of embedded elements like longitudinal reinforcement and dowel bars, as well as the incidence of distresses such as cracks and delamination.
On ultrasonic tomography devices have transducers generating sound waves which are sent and re'lected at the interface of materials with different acoustic impedances. The details about the signal propagation through a material are derived from the analysis of the re'lected signal and can be found elsewhere (HOEGH et al., 2011;LYBAERT, 2015).
Because of the high difference on the air and concrete acoustic impedance, mostly of the energy is re'lected and almost none is transmitted. Accordingly, an incident ultrasonic pulse is re'lected by the concrete heterogeneous structure, resulting in an overlapping of many re'lections of the received signal, commonly called as structural noise. Such a noise trends to conceal the object of interest, suppressing its detection (SCHICKERT et al., 2003).
The limitations caused by the concrete heterogeneities might be mitigated by using the dry point contact (DPC) ultrasonic tomography, which applies a matrix of paired transducers that send and receive the ultrasound signal (LYBAERT, 2015).
According to Hoegh et al. (2011), the use of several pairs of transducers in each scan allows the redundancy of information required to assess heterogeneous media, like Portland cement concrete (PCC). Furthermore, the DPC transducers can transmit elastic waves of low frequency (25 to 85 kHz) capable of penetrating greater depths.
The MIRA equipment ( Figure 2) applies the DPC ultrasonic tomography technology for nondestructive assessment of distress or objects embedded in the concrete, like steel (AGUIRRE et al., 2013). The equipment is composed of 48 transducers for emission and reception, linearly arranged in 12 channels of 4 transducers; each channel with 4 transducers sends out sequentially shear wave pulses while the other 11 channels receive the re'lections of internal interfaces. This arrangement allows 66 separate measurements of transmission-reception transducers per scan (LYBAERT, 2015).  To calibrate the signal speed of propagation in the material, the device measures the signal propagation time between the transducers and applies the Synthetic Aperture Focusing Technique (SAFT) to analyze and reconstruct (image formation) based on the shear waves re'lections (HOEGH et al., 2011).
According to Hoegh et al. (2012), each MIRA's scan provides a SAFT B-scan with the vertical axis indicating the depth of any re'lection (caused by the differences in the acoustic impedance) and the horizontal axis indicating the localization along the device aperture (250 mm). Changes in the acoustic impedance cause a high intensity re'lection (red color) associated to the distress position at the B-scan (for instance, interface between layers, damages in the concrete or steel bars), as depicted in the Figure 3.

Equipment opera$ng modes
The data acquisition with MIRA can be conducted using two operating modes: the "explore" mode which allows performing tests in arbitrary positions through the B-scans visualizations on the structure; and the "scan" mode applied to create a data folder to store the results of the surface complete scan (mapping) (GERMANN INSTRUMENTS, 2016).
Whilst the explore mode enables localized data acquisition (for instance, at the middle of the concrete slab) in few seconds, the scan mode allows the tridimensional (3-D) reconstruction of the evaluated structure using the software IDealviewer which is an equipment supplement.
However, according to the equipment manual of operation (Germann Instruments, 2016), in the scan mode, the measurement should be taken, at the most, every 250 mm horizontally and every 100 mm vertically, in relation to the device displacement during the data acquisition, allowing to map the studied region for its subsequent reconstruction (as an image). It is worth noting that the shorter the spacing for data acquisition in the MIRA's scan mode, the greater the number of redundances for the 3-D image construction. However, shorter spacings imply a greater number of scan points and, consequently, making the survey time-consuming and increasing the equipment operation cost as well.
Regarding the MIRA equipment, 'ield tests performed on the explore mode enable taking several arbitrary measurements (like dozens) in few minutes. Furthermore, the scan mode allows the pavement surface scanning to check the slab thickness uniformity, as well as the reinforcement and dowel bars positions.

Accuracy of MIRA tomographic device for concrete pavement applica$ons
In recent years non standardized 'ield tests were conducted to explore MIRA's inspection potential and accuracy on determining the thickness of pavement layers, dowel bars positioning TRANSPORTES | ISSN: 2237-1346 6 and covering of steel bars in continuously reinforced concrete pavements. Such device has become very popular for road construction control (including tunneling, bridges etc.). In 2010, the United States Army Corps of Engineers (USACE) Engineer Research and Development Center (ERDC) was demanded by the United States Air Force (USAF) for the evaluation of several non-destructive methods for pavement thickness inspection, seeking to determine their capacity of estimating it accurately. Twelve different equipment for nondestructive evaluation were employed to assess forty experimental sections of hot mix asphalt (HMA) and Portland Cement Concrete (PCC) pavements in Vicksburg (USACE Experimental District, Mississippi). For calibration purposes, one core of the pavement surface layer was extracted by section (EDWARDS and MASON, 2011).
At the time, MIRA was the equipment with the best overall performance among those tested both for HMA and PCC, being the only equipment capable of measuring 100% of the PCC test sections with a mean error below 11 mm. The mean error for HMA layers was signi'icantly higher (17,53 mm); however, just 43% of HMA sections were tested. Moreover, the equipment provided real-time measurements, stripping away the coring samples for equipment calibration, allowing its use in a completely non-destructive manner. Figure 4 compares the PCC slabs thicknesses estimations with MIRA against the cored samples. It has been observed a high correlation (R² = 0,9932) and accuracy (slope = 0,9958) between the thickness measurements (ultrasonic versus cores).  Concerning the concrete highness over the steel reinforcement, results showed high correlation (R² = 0,991) and accuracy (slope = 0,972) for MIRA's measurements, with an absolute mean error of 3,18 mm. Similarly, the comparison between the thickness results collected with MIRA versus the results measured from cores presented good correspondence (R² = 0,967) and accuracy (slope = 0.971). Vancura et al. (2013) employed the MIRA to check the thickness of a jointed plain concrete pavement (JPCP) and compared the results with extracted cores. Two pavements newly built were inspected in a South District of Minnesota, with design thickness of 200 mm and 225 mm. For both the sections, twenty-three cores were extracted and 1,476 thickness measurements were performed.
Absolute mean error of the thickness measured with MIRA was 2,67 mm. In addition, the test results with MIRA have shown the data collected from cores did not capture the extreme peaks and valleys of the pavement thickness variation (at the slabs bottom) emphasizing that coring each 305 m (method adopted in the study) does not characterize the thickness variation properly.
Lybaert (2015) reports the use of MIRA in roads located at Ghent and Bekkevoort (Belgium) to check the dowel bars positioning in transverse joints of concrete slabs due to suspicions of construction problems impairing bars alignment. The results revealed the dowel bars misalignment along several joints. As an outcome, the author emphasizes that both the pavements were fully reconstructed, showing some road agencies in developed countries do not put up with some gross construction mistakes.
Van der Wielen et al. (2017) conducted tests on a newly built JPCP with 250 m length, 5 m width and concrete slab thickness of 200 mm laid on an asphalt base. The tests embraced thickness inspection and dowel bars position veri'ication in the joints with ground penetrating radar (GPR) and ultrasonic tomography equipment (MIRA). Seeking to con'irm the pavement thickness estimated through both the equipment, 142 leveling measurements were performed at referred points with total station (Leica iCON Robot 50), which, according to the authors, presents 1 mm of accuracy in estimating the elevations. The surface elevation at each reference point was measured before and after the concrete placement.
Regarding the pavement thickness measurements, the authors report that MIRA presented the best thickness estimation with a mean error of 3,52 mm (1,6% of relative error) and standard deviation of 2,35 mm. The best results with GPR were obtained with a 900 MHz antenna and calibrated speed, resulting in mean error of 4,05 mm (1,9% of relative error) and standard deviation of 3,19 mm.
Analyzing the dowel bar positioning, Van der Wielen et al. (2017) concluded that GPR works very well to pinpoint the dowel bars along the pavement section and, although the dowel bars might be successfully detected by MIRA, the method is less ef'icient than GPR due to the inability of continuous measurements.

MATERIALS AND METHODS
Located in the city of São Roque, São Paulo, Brazil, the private executive airport was designed years ago to receive national and international 'lights alleviating the demand of executive jets using Congonhas (São Paulo downtown airport) and Campo de Marte Airport. The referred airport has a main runway extension of 2,470 m and serves almost 120 private executive jets as parking and maintenance hub nowadays. The airport hangar I apron was built using squared JCPC slabs of 4.8 m x 4.8 m and thickness of 180 mm, concrete with fct,f = 4.5 MPa and dowel bars in any joint since airport pavements are submitted to a non-directional traf'ic, differing from bus corridors or roads channelized traf'ic. The slabs were placed on a CTCS base and well-graded crushed stone subbase, both 150 mm thick. Subgrade soil was a silt-clay with CBR = 8%.
The construction method used lateral forms comprising areas of 24 m by 24 m for concrete laying. One should note that the dowel bars were installed using a conventional steel basket for 'ixing the bars; once assembled such devices, with strong regards for bars parallelism and levelling, besides correct distance among them, were installed in the designed contraction joints positions, 'ixed over the cemented bases by using nail pistol with fasteners. No checking of vertical alignment of the dowel bar positions is actually done as a speci'ication requirement since the standard for that (a road standard is used in Brazil, even for airport pavements) is lacking about such aspect.
Opposite requirements are mandatory in US since the Federal Aviation Administration lays down in the standard speci'ication (FAA, 2018) that the dowel bars should be set up in the joints following the appropriate horizontal and vertical alignment. Accordingly, the dowel bars should be horizontally spaced within a range tolerance of 19 mm, whilst the vertical translations should not be greater than 12 mm. Additionally, the method applied to install the dowel bars must ensure that their horizontal and vertical misalignment is not higher than 2%.
About the method for spreading the base layer material, the FAA (2018) de'ines that cementtreated material should be spread with a mechanical spreader capable of receiving, spreading and shaping the mixture uniformly, avoiding segregation. The equipment should be supplied with a strike-off plate and gates capable of adjusting to the layer thickness and width. Equivalent regulations are present in Brazilian standards for compaction of road base materials, cemented or not.
Tests with MIRA were conducted in November 2019 a couple of days before its opening for landings and take-offs. The tests used MIRA's explore mode to analyze the slab thickness; for dowel bars positioning assessment the scan mode was employed. Twenty-'ive slabs were assessed with measurements taken at the slabs center for thickness checking. Regarding the dowel bars evaluation, data from two joints (one transverse and other longitudinal) of one slab were collected on the scan mode.
The data collection on the scan mode requires to mark an orthogonal grid line on the pavement surface to de'ine the points where the equipment should be positioned allowing to achieve the desired information. Thus, after ensuring the joints are clean, the guidelines were marked on the pavement surface, as depicted in the Figure 5. Visibly it is a time-consuming procedure forbidding general check for all the pavement joints, requiring random approach in actual technological control procedure.
As the MIRA's effective aperture of the transducer's array is 250 mm, the spacing between lines should not surpass 250 mm. Additionally, the user manual recommends the grid lines should be spaced at a distance shorter than 250 mm at x-direction and shorter than 100 mm in the y-direction, related to the device movement on the pavement surface during the data collection. Thereby, it was taken a spacing of 200 mm in the x-direction and 100 mm in the y-direction, as depicted in the In both the joints, eight lines in the x-direction and twenty-four in the y-direction were marked on the pavement surface. Thus, each joint surveyed resulted in 192 points of data acquisition with MIRA.
The equipment was set up in the scan mode for data acquisition at the minimum depth, which is 500 mm. After that, the software Idealviewer was used to reconstruct the mapped area which allows the 3-D visualization of scanned joints as well as the visualization of the tridimensional planes (B-scan, C-scan and D-scan) of the whole joint at the measurement's areas ( Figure 6).

RESULTS AND DISCUSSIONS 4.1. Verifica$on of the slabs thickness using the explore mode
Twenty-'ive slabs were selected to perform the survey on the explore mode, collecting information on the thickness at the center of the concrete slabs. Figure 7 depicts  The thickness obtained with MIRA's explore mode survey of slabs ranged from 151 mm to 222 mm; the average thickness was 172 mm with standard error of 2,95 mm and low coef'icient of variation (8,6%). Results from descriptive statistics for a con'idence level of 95% for the database are presented in the Table 2. The kurtosis positive result presented in Table 2 indicates a tailed distribution (leptokurtic), when compared to the normal distribution. Additionally, as the mean calculated is lower than the mode, the distribution is classi'ied as asymmetric to the left. The asymmetry coef'icient was calculated by the second Pearson's coef'icient, and the asymmetry grade is considered moderate, as its module is in the range between 0.15 and 1.0.
The signi'icant veri'ied variations in the concrete slabs thickness in 'ield compared to the design thickness result from the type of material and technique employed to spread the base, which does not ensure the 'latness required for the base surface layer, leading to the concrete slab non-uniformity thickness. As well know by paving engineers experts only hot mix asphalt mixtures, as well as trowel 'inished concretes (no compacted ones) ensure 'latness for the base surface.

Verifica$on of the slab thickness and dowel bars alignment from the scan mode
Although the 3-D reconstruction (Figure 8) affords relevant information about the concrete slab thickness and uniformity, besides the image discontinuities which may characterize both the joint and a distress in the pavement structure, in such a visualization mode, it is impossible precisely interpret the positioning and the dowel bars alignment conditions directly by the software Idealviewer. Conversely, the software enables visualize the plane B-scan of the whole studied section at eight positions in the y-direction equally spaced from 100 mm, with four positions over each slab (split by the sawed contraction joint). Such a kind of visualization allows to check the slab thickness uniformity over the studied section, beyond enabling to check the dowel bars placement.
Thereby, to check the dowel bars alignment, the B-scans selected were "y = 200 mm" and "y = 500 mm". As these positions are equidistant from joint, it was possible to visualize the dowel bars, even when small translations might have occurred.
Considering the dowel bars length, it normally ranges from 460 to 610 mm in airport pavements, according to FAA (2021), longitudinal translations higher than 50 mm can be veri'ied from the other B-scans. However, if it is necessary to investigate minors' longitudinal translations, spacing shorter than 100 mm should be considered for the device movement, which increases the results precision. In contrast, it demands a greater number of scan points on the pavement surface for the same survey area.
Dowel bars depth measurements (z) in the x-direction, in relation to the system origin (x, z) depicted in the B-scan images, were determined considering the re'lection center which indicated the dowel bars presence.
Thereby, from dowel bars depth measurements on the B-scans "y = 200 mm" and "y = 500 mm", the possible vertical tilt related to the project alignment was calculated employing basic trigonometry, admitting straight dowel bars. Similarly, the horizontal skew was calculated from determining the horizontal distance from the coordinate system (x, z).
The slab thickness was estimated drawing a line between the transverse section edges visualized on the images, and afterward the edge measurements were placed. Thus, the dowel bars positioning, as well as the slab thickness at the evaluated joints are depicted in the Figure 9 and One should note on the longitudinal section B-scans (Figure 10a and Figure 10b) the slab thickness close to the left edge is about 130 mm, signi'icantly (28%) lower than the 180 mm design thickness. This 'inding, aside from underlining the incidence of failures along pavement constructions control process, could bring consequences to the pavement performance, because, although it does not happen in a widespread way, slab thickness thinner than the designed one is a heavy concern for pavement performance.
The image backwall has shown important thicknesses variations, both transversal and longitudinal. Since concrete surface 'latness is ensured buy constructive process, it is clear from the images that the cemented base surface is unlevelled even compaction densities achieved have met the specs. This is one aspect connected to the unevenness resulting after compaction of materials with no lubricant, whatever by type or amount, not capable of favoring the envelopment of aggregate grains.
Thickness negative variation are more concerning since both static as dynamic resistance of concrete slabs is hugely sensible to thickness, reducing drastically its fatigue life; or even jeopardizing a static rupture. This is explained by the fact that actual slab stresses, when thickness de'iciency exists, can quite surpass the critical design stress, what induces non expected mechanical responses to load and curling. Moreover, one can see in Figures 10a and 10b that the dowel bars height from the slab bottom remains practically the same along the whole extension, while the concrete cover decline as it is closer to the slab left edge, suggesting that the lack of uniformity in the concrete slab is probably related again to the base layer irregularity. It is also observed that, considering the joints sawing depth, which is generally 1/3 of the slab thickness (in this case, 60 mm), more than 50% of the dowel bars at the longitudinal joint are at an inadequate depth and, possibly, were partially or totally cut during the joint sawing operation.
It is worthwhile noting that the analyzed slab did not present lateral forms, just sawn joints; and, although it is impossible to af'irm (since it is not known), likely, one of the joints is in the direction of the compaction roller, while the other one would be at the compaction perpendicular direction. Thus, there may be a sign that the compaction direction is constraint of a more sensible direction; in airport large areas the compaction process can differ a lot from highway construction.
Although the compaction direction is not matter of concern in road and bus corridors pavements due to constraint in the number of lanes, in case of airports, harbors and industrial areas, this aspect must be considered because of the slabs' signi'icant dimensions (width and length).
Also, due the base layer unevenness, the thickness checking from a limited number of points (for instance, one or two points per concrete slab), even if carried out through non-destructive technique, tends not to be representative because of the inability of evidencing uniformity failures in the slab thickness.
Then, despite the results obtained with MIRA at the center of the inspected slabs, using the device explore mode, indicate a reasonable precision for slab thickness compared to the design, the sections obtained from the scan mode show the measures at the concrete slab center may not represent properly the real slab thickness.
In addition, it is possible to observe the lack of signal intensity where there should be a dowel bar in both the images (Figure 10a and Figure 10b). This might be associated to the dowel bar translation far away the joint (design position).
The vertical tilt and horizontal skew amplitudes related to the dowel bars design alignment conditions are presented in Tables 3 and 4. The negative signal indicated the dowel bars rotation on the opposite direction from that assumed as positive.
As can be seen from Table 3, at the transverse joint, 68,75% of the dowel bars do not present vertical or horizontal alignment within the tolerances established by the FAA (2018) standard. Analyzing the results for the longitudinal joint (Table 4), 43,75% of the dowel bars do not satisfy the limits set up for vertical tilt, while 75% of the dowel bars do not ful'ill the horizontal skew tolerance.
Although the dowel bars misalignment may occur because of different reasons, the lack of suitable stiffness of metal baskets for bars support and how they are attached on the base layer are usually considered the critical factors when the bars are placed using such a technique. However, very stiff baskets could undermine stresses distributions and levels close to contraction joints, what must be analyzed carefully for a design decision.
Additionally, the lack of base layer surface 'latness will affect the appropriate placement of the metal baskets, as discussed, impairing bars positioning due to twists and bends during the 'ixation, leading to dowel bars vertical tilt and horizontal skew from the original design position.

CONCLUSIONS
Several 'ield non-destructive tests using the MIRA ultrasonic low frequency waves device were performed at an apron in an executive airport site, over a newly built JPCP.
The tests were carried out considering two operation modes available in the device: the explore mode that enables to perform the tests at arbitrary positions (here, centered at the concrete slabs inspected); and the scan mode, used to create a folder to store the database generated from a complete surface scanning (joints region mapping) to later 3-D reconstruction.
A total of twenty-'ive concrete slabs were inspected using the explore mode to check their thickness and one concrete slab was selected to map two of its joints using the scan mode to check the slab thickness along the joint as well as the dowel bars placement compliance.
Regarding results from explore mode, it is revealed an important variability in concrete slab thickness, ranging between 151 mm and 222 m, with average of 172 mm and coef'icient of variation of 8,6%. These results are closely related to the employment of inappropriate techniques and equipment which do not ensure the base layer surface evenness, resulting in non-uniformity of the concrete slabs thickness.
Then, albeit using the equipment on the explore mode allow to reach a good productivity in determining the concrete slabs thickness, because of the base layer irregularity the slab thickness veri'ication from a limited number of points (as performed in this study) may not be representative; it is required a signi'icant number of tests at the same slab for a suitable checking of concrete slab uniformity and thickness or through the slab sections scanning (transverse and longitudinal).
The tests carried out on the scan mode allowed to survey the pavement surface verifying slab thickness uniformity, as well as the dowel bars placement. The results show the slab thickness non-uniformity surrounding both the transverse and longitudinal joints. From longitudinal section B-scans it was possible to observe the thickness at the edge of the investigated slab was 28% slightly thinner than the design. Additionally, more than 50% of the dowel bars at the assessed longitudinal joint had some wander related to slab depth. However, such aspects should be taken as built-related scatter mainly due to unevenness of the base surface (lack of 'latness).
Most of the dowel bars at both the joints assesses presented misalignment higher than the FAA tolerances. However, it is relevant to highlight these tolerances are not usually based on performance data. Furthermore, the excessive lack of 'latness over the base layer surface can affect the bars alignment during the 'ixation of the metal baskets, due the risk of bending and twisting of such devices. Therefore, controlling the vertical geometric levelling of the base layer contributes to mitigate variations in the slab thickness uniformity and to ensure the bars alignment when dowel baskets are used as well.
Therefore, as main conclusion, since base types are a design decision, it is strongly recommended the use of special baskets permitting vertical adjustments of the bars levels ensuring al of then (in one joint) shall be at the same covering distance from the 'inished concrete surface.
Ultimately, the use of non-destructive evaluation methods which enable partially scan the pavement surface like is the case of MIRA, trends to contribute to analysis that allows to check, among other parameters, the thickness uniformity along the pavement sections, besides the positioning of reinforcement and dowel bars.
From the investigated sections B-scans it was possible to check the dowel bars vertical and horizontal misalignments, underlining the potential of using MIRA for checking the dowel bars placement compliance for in service pavements.