Figure 1 - Desired Sound Field Characteristics of the Array
The space available in smaller tubes prohibits individual conventional transducers being mounted to accomplish this so the array is assembled as a single unit. A piece of acoustically absorptive backing is first machined as a cylinder with eight flat areas where the piezoelectric elements are mounted. Fig.1 shows the sound field configuration of the array. The array elements are angled toward the axis of the tube to achieve the oblique incidence for shear wave inspection as discussed earlier and as shown in the figure. It can also be seen that although the sound fields do not overlap in the polar plane, critical flaws are typically large enough that they will be detected by at least one array element.
The notches were made using a 0.8 mm (0.032 inch) blade. This does not accurately simulate the closed cracks likely to be encountered in actual heat exchanger tubing but it does represent an easily manufactured standard for initial testing. The presence of two corners at the bottom of the notch rather than one in a true crack may account for some slight inflections we noticed in the acoustic profiles but we don't believe the effect is significant in this evaluation.
Nine circumferential notch samples were fabricated with depths ranging from 0.12 mm (.005 inch) to 1.14 mm (0.045 inch) in 0.12 mm (.005 inch) increments. The probe was moved axially through the sample until the maximum signal response was encountered. This maximum signal level was then plotted as a function of notch depth in Fig.3. Notches with depths greater than 0.8 mm (0.032 inch) yield the same peak response. This is probably due to the corner geometry being larger than the lateral sound field dimensions. At smaller notch depths, however, the peak response decreases proportionally. This decay probably becomes very nonlinear as the notch depth approaches zero but the decay appears to be nearly linear over the range plotted. Further testing is necessary to characterize the nature of this response.
|Figure 2 - Notch Echo Amplitude as a Function of Circumferential Notch Depth||Figure 3 - Notch Echo Amplitude as a Function of Partial Notch Depth|
The partial notch geometry is shown in Fig.4. Again nine samples were fabricated with depths incremented by 0.12 mm (0.005 inch). In these samples, as the notch depth increases, the polar dimension of the notch also increases and this more closely represents crack geometry encountered in tube walls. Fig.4 also shows the maximum axial response as a function of notch depth. Again we see that there exists a notch size above which no further increase in response echo occurs. It can also be seen that notches only .12 mm (.005 inch) deep and having a circumferential length of 2.5 mm (.1 inch) yield signal levels only 4 db below the response of a near through-wall notch. This is an encouraging indication of the sensitivity of this inspection method.
Figure 4 - Crossover of Echo Levels
Figure 5 - Crossover Level as a Function of Notch Depth
At some point in this rotation, the responses from the two elements will be of equal amplitude. If the alarm threshold in the flaw detector is adjusted to this level or lower, then a notch of this size or larger cannot be missed by passing between two adjacent sound fields. We define this signal level as the "crossover level" for the particular notch size--the level at which the response signals from two adjacent array elements crossover each other. Naturally this "crossover level" will be a function of flaw detector setup and calibration. For our evaluation, we adjusted the instrumentation so that a 1.1 mm deep notch response yielded a signal 80% of the total screen height. We then measured the "crossover level" in terms of screen height as a function of notch depth. Those results are presented in Fig.6 and indicate that an alarm threshold set at 40% screen height will guarantee that all flaws of this geometry and having a depth greater than 0.6 mm (0.024 inch) will be detected.
The angle notch geometry is shown in Fig.7. Nine samples were fabricated all similar to the 0.64 mm deep partial notch except that the angle of the notch relative to the tube axis was varied from 0 degrees to 16 degrees in 2 degree increments. As expected, the peak signal response decreases with increasing notch angle as shown in the figure. Crossover level as a function of notch angle is shown in Fig.8 and indicates that a 20% alarm threshold will guarantee flaws of this geometry oriented less than 5.5 degrees from the tube axis will be detected.
Figure 6 - Notch Echo Amplitude as a Function of Notch Angle
Figure 7 - Crossover Level as a Function of Notch Angle
Figure 8 - Longitudinal Profile with Several Half-Node Reflections
Fig.9 shows echo-dynamic profiles for partial notches of various depths.
If we assume that the 0.62 mm (0.025 inch) deep partial notch is representative of a critical flaw and if the flaw detector alarm threshold is set at -12 db on the RELATIVE AMPLITUDE axis, then the notch will be detected for a l.5 mm (0.06 inch) probe displacement. Knowing this, the maximum probe speed can be estimated. Although the alarm circuitry can be latched on a single threshold violation, a more conservative approach limits probe speed such that the notch would be detected on no fewer than five consecutive pulse cycles. Using a pulse repetition frequency of 1250 Hz for each channel, the maximum probe speed is computed to be greater than 0.45 m/sec. (18 in/sec.). Even using the 0.12 mm deep partial notch, the maximum probe speed is 0.25 m/sec. (10 in/sec.). Naturally these probe speeds should be evaluated on real flaws since the echo-dynamic profile will probably vary.
The Paper was presented on the UTonline Application Workshop in May '97
|| UTonline ||