Use of Magnetic Tomography Technology to Evaluate Dowel Placement
Laboratory Evaluation (continued)
A comprehensive series of tests were conducted to verify the accuracy of MIT Scan-2 results. Several factors are known to affect the accuracy of MIT Scan-2 results, including the bar depth, magnitude of position error (side shift), and amount of misalignment. The effects of these factors on MIT Scan-2 results were evaluated systematically to verify the accuracy of MIT Scan-2 and to determine the operating range of MIT Scan-2. Within the operating range, the device is expected to provide the specified level of accuracy. The measurement errors were determined by comparing MIT Scan-2 results to manual measurements.
The range of values of the test parameters for the testing program was selected in consideration of both the MIT Scan-2 specifications and the application requirements. According to manufacturer specifications, the device provides the specified accuracy under the following conditions:
- Dowel bar depth 150 + 40 mm (4.3 to 7.5 in.)
- Horizontal and vertical misalignment less than +40 mm (1.6 in.)
- Lateral shift (side shift) less than +80 mm (+3.2 in.)
When the above conditions are satisfied, the manufacturer-specified accuracy is within 4 mm (0.16 in.) on rotation and bar depth.
The ranges of values of the key parameters specified for MIT Scan-2 are adequate to cover most situations for highway applications, except for the following:
- The upper limit of 190 mm (7.5 in.) for depth may not be sufficient for testing on projects where the dowel bars are placed too deep. While most specifications require the bars to be placed within 25 mm (1 in.) of mid-depth, it may be permissible to allow a deeper placement, as long as adequate concrete cover is provided. If a minimum cover of 76 mm (3 in.) were accepted, the dowel bars can be placed as deep as 235 mm (9.25 in.) in 330-mm (13-in.) pavements.
- The horizontal misalignment range of +40 mm (+1.6 in.) is not sufficient for testing skewed joints. Skewed joints need to be tested with the rail set parallel to the joint saw cut, which leaves the dowel bars at an angle with respect to the rail. The results thus obtained will show a uniform apparent horizontal misalignment corresponding to the skew angle in addition to the actual horizontal misalignment. The apparent horizontal misalignment corresponding to the typical joint skew angle (1 in 6) is 76 mm (3 in.) for 450-mm (18-in.) dowel bars. MIT Scan-2 must be able to accommodate this additional amount of apparent horizontal misalignment to enable testing on skewed joints.
During the course of this study, MIT GmbH made software enhancements to allow testing on skewed joints. To verify that the results for skewed joints are accurate to the same degree as the results for square joints, the evaluation range for horizontal misalignment was extended to +120 mm (4.7 in.).
The limits on the range of depths, however, are physical limitations (relating to the signal to noise ratio), which cannot be overcome without a hardware modification. The sensitivity of measurement errors to depth is shown in Figure 12 (supplied by MIT GmbH). The error is less than 4 mm (0.2 in.) for depths up to about 200 mm (7.9 in.). At depths above 200 mm (7.9 in.), the measurement error increases rapidly. This is only a limitation for the devices that were optimized for highway applications, where the typical bar depths range from about 100 to 165 mm (4 to 6.5 in.). For testing thicker pavements, the signal strengths can be increased to shift the zone of accurate results upward. In fact, the signal strength of the devices shipped to the United States was reduced by half from the original design to accommodate testing that included dowel baskets. Excessively high signal strength causes the response signal to saturate the sensors, making it impossible to obtain any information about the dowel position.
The accuracy of MIT Scan-2 and effects of various factors affecting the results were evaluated primarily based on tests on a single dowel bar. The only exception is the limited testing conducted to verify the minimum allowable bar spacing. MIT specifies a minimum bar spacing of 250 mm (9.8 in.). Limited testing conducted using multiple dowel bars has shown that a bar placed within 200 mm (7.9 in.) of the testing sample does increase the measurement error, but a bar placed 255 mm (10 in.) or farther away has no effect on the test results. The key parameter is the d/z (bar spacing over depth) ratio. The error increases with increasing depth. At shallower depths, a closer bar spacing can be accommodated. In typical highway applications, the 250-mm (9.8-in.) limit on bar spacing does not pose a problem.
The nominal ranges of the test parameters evaluated in the laboratory testing are as follows:
- Depth is 100 to 200 mm (4 to 8 in.)
- Side shift is -100 to +100 mm (-4 to +8 in.)
- Horizontal misalignment is -120 to +120 mm (-4.7 to +4.7 in.)
- Vertical misalignment is -40 to +40 mm (-1.6 to +1.6 in.)
The evaluation results are shown in Figures 13 through 22. The following can be observed from the test results:
- Within the limits of the ranges of test parameters (depth, side shift, horizontal misalignment, and vertical misalignment), the errors on horizontal and vertical misalignment results are within the MITspecified limits of +4 mm (+0.16 in.). The only exception is that the error on vertical misalignment for the bar placed at 200-mm (7.9-in.) depth exceeds the specified limit. The errors that exceed -4 mm (-0.16 in.) in Figures 14, 16, and 18 are for the bar depth of 200 mm (7.9 in.).
- The measurement error on both horizontal and vertical misalignment is independent of all other parameters (i.e., side shift, depth, horizontal misalignment, and vertical misalignment), except that the error on vertical misalignment is higher for depth greater than 200 mm (7.9 in.). Figures 13 through 20 do not show any trend in the measurement error in either horizontal or vertical misalignment as a function of side shift, depth, horizontal misalignment, or vertical misalignment.
- Figure 21 shows a bias in the depth results. As discussed earlier, the greater error on depth results may be due to the differences in metal composition between the calibration bar and the test sample. The average error in depth measurement is -4.3 mm (-0.17 in.). The standard deviation of the depth error is 1.4 mm (0.06 in.). The depth results may be shifted due to variations in metal composition.
- Figure 22 shows a bias in the side shift results. The average error in side shift measurement is -4.4 mm (-0.17 in.). The standard deviation of the side shift error is 4.1 mm (0.16 in.). Side shift is difficult to measure precisely by hand without the aid of instrumentation such as those built into the MIT test track (Figure 6). The results of the tests conducted at the MIT test track do not show any bias in side shift measurements, and the results are within the specified range.
The laboratory testing results confirm that MIT Scan-2 provides accuracy that is both reasonable and useful for horizontal and vertical misalignments within the following limits:
- Depth 100 to 190 mm (3.9 to 7.5 in.)
- Side shift +100 mm (+4 in.)
- Horizontal misalignment +40 (+1.6 in.) plus a uniform rotation of +80 mm (+3.1 in.)
- Vertical misalignment +40 mm (+1.6 in.)
Note that the operating range for horizontal misalignment is a range of misalignment (+40 mm [+1.6 in.]) plus a uniform rotation. Although the laboratory evaluation covered the full range of horizontal misalignment (+120 mm [+4.7 in]), the tests were conducted using a single dowel bar. In an actual joint, extreme horizontal misalignments can cause the ends of neighboring bars to come very close to each other, which in turn can cause additional error due to overlapping response signal. On skewed joints, the magnitude of apparent horizontal misalignment is large, but a major portion of the registered horizontal misalignment is due to a uniform rotation caused by joint skew, which does not cause a problem for neighboring dowel bars. For the actual misalignment, the range originally specified for horizontal misalignment (+40 mm [+1.6 in.]) is more than adequate to cover all practical cases. If the actual horizontal misalignment exceeds +40 mm (+1.6 in.), the exact quantitative results are not important; it is sufficient to know that the horizontal misalignment is large.
(This section continued on following page.)