Metrological evaluation of a flaw detector used in non-destructive testing by ultrasound

Nondestructive tests are widely used in various sectors of industry, and its main techniques are based on the use of ultrasound. The flaw detector is a device used in ultrasonic testing. In this article, it is studied the influence of the flaw detection settings on the measurement result. It was observed that when the speed of sound and the thickness of the standard block are known, the equipment gives the best results, with a maximum error of 08%. The results show the importance of having a standard block calibrated in order to minimize errors.


Introduction
Nondestructive testing (NDT) are widely used in industry, particularly in the evaluation of mechanical parts. Among the NDT techniques, the most widely used are based on the use of ultrasound, for example, for measuring the thickness or location of flaws and discontinuities [1]. Many of these measurements are performed using specific equipment, as a flaw detector. This equipment is used to detect defects in materials, can be used in external or internal environments, and plays an important role in the reliability of nondestructive tests carried out [2].
The purpose of this paper is to obtain a metrological evaluation of measurements carried out by a flaw detector in order to show the importance of knowing the different features present in an ultrasound test.

Case study
In order to evaluate the results given by the flaw detector and to show the importance of understanding the fundamentals involved in the test, different measurements were performed. Three transducers with different nominal frequencies were used, and tests were conducted using different equipment settings. Tests were performed using a standard block type 1, with speed of sound and dimensions previously calibrated. The development and implementation of this research were performed at Inmetro's Laboratory of Ultrasound (Labus), which has the entire infrastructure to carry out the proposed work.

Flaw detector
The flaw detector used in this work was the model portable analyzer EPOCH 600 (Olympus NDT, USA) -see figure 1. That allows changing various settings of the tests [2]. . Therefore, the speed of sound in material under test is an input on the equipment, which is used for converting the time of flight measured by the equipment, in length.

Test body
A standard block type 1 was used for the tests [3]. The block has a thickness of 25.036 mm with expanded uncertainty of 0.003 mm (k = 2.0 p = 0.95) and a height of 100.002 mm and expanded uncertainty of 0.003 mm (k = 2.0, p = 0.95). Those dimensions were calibrated in the Dimensional Metrology Laboratory of Inmetro (Lamed). The block has an ultrasonic propagation speed of 5910.6 ms -1 with expanded uncertainty of 6.3 ms -1 (k = 2.36, p = 0.95), as calibrated at Labus.

Experimental procedure
Tests were performed in different ways, all using transducers with different nominal frequencies and changing the configuration for calculating the time of flight considering the first peak or the maximum peak. The transducers used in the tests were the following: V306 2.25 MHz, V309 5 MHz, and A320s 7.5 MHz, all with a diameter of 0.5" (Panametrics NDT, USA). For the first, second and third tests, the calibrated speed of sound of the block (5.910 ms -1 ) was used as input on the Epoch 600.
In the first, the block height measurements were carried out using as reference the time of flight between the start signal and the first reflection (first echo).
In a second moment, the height of the block was measured, with reference to the time of flight between the first reflection (first echo) and second reflection (second echo).
In the third test the equipment was set before the measurements. The adjustment was performed by measuring the block height and fixing it by means of the "zero setting" in the signal received by the equipment, in order to presented the same value calibrated by Lamed. Then, the third reflection of the block thickness was measured, equivalent to a size of approximately 75 mm (three times the block thickness). We chose to measure the third reflection, because according to the ABNT NBR 15824 -Nondestructive testing -Ultrasound -Thickness measurement, the measuring instrument shall be deemed fit to measure thickness in a range of ± 25% of the thickness of the block used in the default setting. In other words, the size of 75 mm is the limit for the measurement set in a block of 100 mm [4].
Finally, we used a feature of the Epoch 600, called "auto-calibration". By means of this tool, it is possible to estimate the value of the propagation velocity for the material tested by a known dimension value. Thus, the transducer was positioned on top of the block and it was informed to the equipment their height (100 mm). By doing that, the speed of sound of the block was estimated. Thereafter, it was measured the third reflection block thickness using the speed of sound estimated by the equipment.

Measurement uncertainty
In order to assess the reliability of measurement results, showing a quantitative indication of the quality of results, the measurement uncertainty was estimated. For a general relation ( ), the general formula for the model uncertainty is obtained by the equation 1: Where u c is the combined standard uncertainty associated with the outcome of the measurement (or calculation) h, and u j the standard uncertainty, evaluated as Type A or Type B, associated with each variable parameter x j used to express the value of h. [5].
Therefore, the initial step is to identify the sources of uncertainty. The following sources of uncertainty were identified: repetition (Type A), the speed of sound in block 1 (Type B) and the equipment resolution (Type B).
The uncertainty of repetition comes from ten measurements; the uncertainty of the speed of sound in block 1 comes from of the calibration certificate emitted by Labus (c =5910.3 ms -1 ; U = 6.3 ms -1 ; k = 2.36); and, finally, the resolution of the equipment (0.01 mm). Table 1 presents the mean values and their respective uncertainties, for the test conducted using the time of flight between the excitation signal and the first echo as reference. Table 2 depicts the results for the test conducted using time of flight between the first and second echo as reference. In table 3, one can see the results for the test using the "zero setting", while table 4 shows the results achieved using the auto-calibration procedure.

Results and discussion
Assessing table 1, it is noted that as the nominal frequency of the transducer increases, the values of the results tend to decrease. As the wavelength decreases when the frequency increases a higher resolution can be achieved. Moreover, transducers with higher nominal frequency have thinner coupling layer, than the total time of flight is shorter and, consequently, the thickness values are closer to the reference value. Besides, the values measured using as references the first peak are smaller to those measured in relation to the maximum peak. This is due to the maximum peak always appears after the first peak (bigger total time of flight). Here, it is important to mention that the previous analysis is valid only for results presented on table 1. In the other result, the differences in time of flight are mathematically annulled (table 2) or corrected using the NDT equipment settings (tables 3 and 4). The results achieved using the excitation signal and first echo as reference showed an error greater than the other tests, with the maximum error of 3.09% (table 1). On the other hand, determining the time of flight between the first and second echo, the maximum error falls to 1.12%. A possible explanation for this difference is that the time of flight between the excitation signal and the first echo contains the time of flight taken in the coupling layer of the transducer. Hence, that time of flight is bigger than the one measured between the first and second echo, where the time of flight in the coupling layer is annulled. Knowing the speed of sound in the material and its thickness, it is possible to perform the zero adjustment. This setting gave the best result, with a maximum error of 0.08% (table 3). Here it is important to highlight that all errors were smaller than the estimated measurement uncertainty (table 3). Finally, knowing only the thickness, it is possible to use the "auto calibration". This setting showed a maximum error of 1.03% (table 4).

Conclusion
In general, the flaw detectors allow having control of various setting parameter. However, despite the advantage of having greater control over the testing carried out, it is clear that the operator must hold theoretical knowledge and experience about the performed test. Just one configuration incompatible with the test can generate inconsistent results compromising the inspection. Besides, the transducer frequency, as well as the reference used for measuring the time of flight, are factors that may influence on the final outcome. Considering the different possible settings in the NDT equipment, the operator training has major importance in order to mitigate errors by choosing the appropriate parameters settings. Moreover, it is evident the importance of having a standard block with the speed of sound, as well as their dimensions calibrated, in order to minimize errors.