·Table of Contents
·Workshop - Landmine Detection Equipment
Possibilities of Material Classification by Means of Ultrasound
Damir Markucic,Faculty of Mechanical Engineering & Naval Architecture, University of Zagreb, Croatia
Contact
·Table of Contents ·Workshop - Landmine Detection Equipment | Possibilities of Material Classification by Means of UltrasoundDamir Markucic,Faculty of Mechanical Engineering & Naval Architecture, University of Zagreb, Croatia Contact |
Beside research and development of new sophisticated methods and techniques for landmine detection, the preliminary experiments were performed with the intention to analyze possibilities to improve classical detection methods by means of ultrasound.
With this scope, the ultrasonic response from contact (entry) surface of objects is analyzed. The contact and ultrasonic immersion technique were applied. The possibilities of classification are discussed due to impact of range of surface roughness values on variability of results.
Keywords: ultrasound, surface roughness, repeatability, classification.
Ultrasonics and acoustics as non-destructive testing methods are regularly used with two main scopes: flaw detection and material characterization. Special area of application of these physical principles is landmine detection. Regarding the nature of soil structure with many different macro constituents, the landmine detection by means of elastic waves is quite different than flaw detection in homogeneous materials. The problematic is more wisely combination of detection and characterization - therefore called here classification, because between many responses the proper one should be distinguished, recognized and evaluated.
Air and soil in comparison with homogeneous metals, are not so "suitable" media for elastic wave propagation. To avoid this, elastic waves should be emitted close to or during physical contact with the buried object. So, there are two possible set-ups to transfer pulses from transducer to buried object: non-contact and contact. Both set-ups comprise many impact factors, advantages and disadvantages, but to get any response from the object the crucial factor is entry surface. Starting from the character of the surface of buried objects, the surface topography and roughness are notable and common for both set-ups.
Finally, we always aim at successful performance of detection procedures. While the reliability of pyrotechnicians (personnel) is out of the extent of this paper the capability of detection method is in the focus. It is also presumed that inherent characteristics of the equipment e.g. sensitivity, resolution, reliability, signal to noise ratio etc., are suited for the purpose and are satisfactory.
The impact of surface roughness of buried objects on detection capability is considered. Such kind of detectability could be analyzed and validated through seria of repeated performances to determine how surface roughness impacts the consistency of results of detection.
During detection, the acquired signal differentiation and evaluation is done by utilization of threshold or alarm, but behind signal there is a measurement. The general term for variability between repeated measurements is precision [1]. Two extremes of precision are repeatability and reproducibility. The first describes the minimum and the second the maximum of variability of results.
Under repeatability conditions the elements of testing (or measuring) system are considered constant, so they do not contribute to the variability. Only inherent variability of the method and/or testing system is present. Under the controlled intermediate conditions, like variation of surface roughness, one could gain the impact on the variability of results. This approach enables quantified judgement whether the difference between the two (or more) measured responses is due to the inherent variation in the measurement, or if it is due to the different "quality" of the responders.
To consider the impact of surface of antipersonnel landmines (APL) the actual surface roughness for several types of landmines was measured. The example values are given in Table 1.
| type of APL | Ra, mm | Rz, mm | Rmax, mm | |
| PAM-1 (thinwall) | 2,80 2,24 | 17,37 13,37 | 22,40 20,32 | |
| PAM-1 (thickwall) | 0,75 1,16 | 3,84 4,84 | 4,76 6,72 | |
| PAM-2 | 0,97 0,88 | 5,23 4,28 | 6,48 5,04 | |
| PAM-3 | 0,57 0,35 | 3,64 2,50 | 6,80 3,66 | |
| VS-50 | 0,45 0,54 | 3,76 4,01 | 7,28 7,76 | |
| Table 1: Surface roughness for several types of antipersonnel landmines | ||||
The standards, technical specifications and recommended practices refer that surface roughness (Ra) for non-destructive ultrasonic testing should be smaller than 6,3 mm [2, 3, 4, 5]. It is assumed that surface roughness greater than 12,5 mm have significant impact on testing results. Following this requirement and the values from Table 1 according to the surface roughness of landmine parts the specimens of the same material and dimensions were prepared.
In the first phase of research the amplitude responses from the specimens were measured at 2 and 4 MHz nominal frequencies. The measurements were performed with contact and immersion techniques.
Amplitude responses in contact measurements were observed through coefficient of decibel loss. Data are given in Table 2.
| Ra, mm | 0,28 | 0,54 | 1,59 | 4,61 | |
| decibel loss | |||||
| 2MHz | 
| 0.2676 | 0.2695 | 0.2558 | 0.2236 |
| s | 0.0077 | 0.0101 | 0.0090 | 0.0128 | |
| 4 MHz |
| 0.1790 | 0.1711 | 0.1778 | 0.1128 |
| sr | 0.0154 | 0.0149 | 0.0077 | 0.0062 | |
| Table 2: Mean values and standard deviations of contact measurement results for 2 and 4 MHz. | |||||
Before repeatability and reproducibility (R&R) analysis the statistical tests for
Grubbs' test showed that measurements on specimen with Ra = 4,61 mm are outliers. This is also visible on graph shown in Figure 1.
Fig 1: Results of contact measurements for specimens at 2 MHz.
|
R&R analysis based on the criteria of critical differences showed that all measurements at 2 MHz are not repeatable but they are reproducible, while the measurements of three specimens (without eliminated outlier) at 4 MHz are repeatable. The values of R&R analysis are shown in Table 3.
| @ 2 MHz (for 4 specimens) | @ 4 MHz (for 3 specimens) | @ 2 MHz (for 3 specimens) | |
| sr | 0.0101 | 0.0072 | 0.0090 |
| sR | 0.0958 | 0.0072 | 0.1117 |
| sL | 0.0953 | 0.0008 | 0.1113 |
| 0.0459 | 0.0060 | 0.0138 |
| CrD | 0.0163 | 0.0116 | 0.0145 |
| CRD | 0.3778 | 0.0120 | 0.4410 |
| Table 3: Critical difference values of R&R within specimens for 2 and 4 MHz. | |||
where is:
sr - standard deviation of repeatability
sR - standard deviation of reproducibility
sL - standard deviation between measurements
CrD - critical difference of repeatability
CRD - critical difference of reproducibility
The repeated R&R analysis for the first three specimens at 2 MHz also showed that they are repeatable. Consequently it is concluded that measurements of fourth specimen significantly differs from others what is also visible on graph shown in Figure 2.
Fig 2: Results of contact measurements for specimens at 4 MHz. |
Furthermore, it has to be emphasized that greater values of contact surface roughness e.g. in the range of Ra = 4,61 mm, considerably impacts the repeatability conditions of amplitude responses in ultrasonic contact technique.
Analogue as for the results of contact measurements, the R&R analysis of immersion measurement results was performed. While in the immersion technique the beam pattern depends on the distance, the measurements were done at two characteristic distances of transducer from specimen: at near zone (N) and at half value of near zone (N/2).
The peak voltages of pulses reflected from entry surface were measured. Data are given in Table 3 and graphically presented in Figure 3.
| Ra, mm | 1,19 | 3,44 | 7,33 | 9,80 | |
| Up, mV | |||||
| 2 MHz |
| 384,3 | 374,0 | 254,2 | 217,4 |
| @ N/2 | s | 1,53 | 0,00 | 0,80 | 6,45 |
| 2 MHz |
| 410,0 | 406,0 | 391,0 | 250,8 |
| @ N | s | 2,11 | 1,00 | 0,00 | 0,80 |
| 4 MHz |
| 508,7 | 488,7 | 450,3 | 189,5 |
| @ N/2 | sr | 4,16 | 7,37 | 8,39 | 7,57 |
| 4 MHz |
| 496,7 | 464,7 | 400,2 | 225,4 |
| @ N | sr | 7,64 | 12,69 | 8,30 | 6,15 |
| Table 3: Mean values and standard deviations of immersion measurement results for 2 and 4 MHz. | |||||
Fig 3: Amplitude responses from specimens in immersion at 2 and 4 MHz. |
It is obvious that amplitude responses become smaller at greater values of surface roughness. This is not so notable for roughness within Ra = 4 mm, what is similar to the conclusions as for the contact measurement technique. That is, the measured variability within this range of surface roughness lies within the repeatability conditions. Furthermore, greater variabilities are caused by other variations or by higher values of surface roughness.
Surface roughness of artificial buried object like antipersonnel landmines have negligible impact on the variabilities of results of ultrasonic detection. It is also shown that this conclusion is valid for contact and non-contact set-ups.
Conclusion stands for values of surface roughness smaller than Ra = 4 mm and herein utilized pulse frequencies. It is expected that responses obtained from other macro constituents that could be found in the soil will be significantly different. These macro constituents could be either artificial articles or natural objects, but with greater values of surface roughness.
This inference is a step toward further research and development of classification and differentiation of buried antipersonnel landmines in realistic conditions by means of elastic waves.
Beside APLD topics, having in mind reliability of common ultrasonic testing, the standard requirements for roughness of contact surface should also be reviewed and adjusted to ensure as good as possible repeatability and reproducibility of testing results.
| © AIPnD , created by NDT.net | |Home| |Top| |