·Home ·Table of Contents ·Aeronautics and Aerospace

Development and Application of a Software for Studying the Interaction of Sound Pulse with Defects and Checking the Performance of Signal Analysis Tools VK Ravindran and BC Bhaumik Quality Division Propellants and Chemicals,Vikram Sarabhai Space Centre Thiruvananthapuram-695022 Kerala State India Contact

ABSTRACT:

Calibration, performance checking and developing the understanding of the physics of interaction of ultrasound is an integral part of routine Ultrasonic NDE. Conventionally this is done by fabrication of Reference / Calibration blocks and making use of standardised procedures. A software system has been developed by the authors for improving and overcoming certain limitations of this practice. It is used for learning the interaction of sound pulse with different geometric conditions of defect. Knowledge gained through this learning process is utilised to improve interpretation of inspection results.

INTRODUCTION:

In the modern engineering practices, NDE is an essential tool for inspecting materials and products for assessing service life, acceptability, risk assessment etc. Since NDE is an indirect measurement method, in the event of locating a defect, details of the defects are to be extracted through analysis of the signatures obtained through inspection. Ultrasonic inspection is one of the major NDE methods used for this purpose. In Pulse echo Ultrasonic inspection, pulses of ultrasound are generated by suitable transducers and sent to the material or product under inspection. This sound pulse interacts with defects inside the material and gets scattered or reflected back into the transducer. It is well known that the nature of the received pulse depends on the nature of reflector. The nature of the resultant reflected pulse can be identified by its shape, amplitude, frequency contents etc. Changes in these parameters could occur due to the superimposition of one reflected pulse over the other from different geometrical locations of the same reflector. These pulses, received by the transducer, are used by NDE professionals for interpreting the defects. Conventionally, amplitude of this signal pulse is used for getting approximate idea of the size of the defect. The other features (as listed above) of the resultant signal pulse are also being used to gain an insight into the nature of the defect. However for this purpose, these features are to be extracted from the raw sound pulse obtained through inspection and presented in interpretable form.

In order make these interpretations accurate and effective, a thorough understanding of the sound pulse generated by the particular transducer used for inspection and its interaction with defects under various conditions is required. Also, for effective usage of computer based signal analysis, tools involved are essentially to be checked for their performance. These performance checks are done by fabricating various known defects to suit the objective. However, in most situations it becomes difficult or impossible to realise all these physical conditions through fabrication activities. Computer simulation can be utilised to fill some of these gaps.

SOFTWARE DEVELOPMENT FOR SHOP-FLOOR APPLICATION - A NEW APPROACH:

Due to the advent of low cost high speed Personal Computers, the application of computers are wide spread. During the last decade applications of computer technology in Ultrasonic NDE have been explored. However till now no serious approach has been made towards standardisation of any of these works. Even the well-known aspects of the spectral analysis have not been utilised as general practices in industry around the world . Basic reason for this is the lack of standardised software systems for conventional NDE application. Authors are attempting to set a new trend in software development for shop-floor applications to enhance the interpretation ability and system performance check . The presently developed software can be used for understanding the interaction of sound pulse produced by different transducers in different materials in much easier way. The essence of the software is explained in the following sections through practical example to show advantage and applicability.

COMPUTER GENERATION OF ARTIFICIAL SIGNALS FOR CHECKING THE PERFORMANCE OF SIGNAL ANALYSIS SOFTWARES:

In pulse echo ultrasonic systems, ultrasonic sound energy is generated as pulses. As mentioned earlier these sound pulses have specific spectral characteristics depending on the transducer and equipment. Hence spectral analysis tools are integral part of all ultrasonic signal analysis software systems. Since NDE is a measurement science (mostly relative in nature) performance is to be checked in the applicable range of features. The software developed by the authors generates digital artificial signals with optimum standard conditions suitable for checking performances of various displays and features of Analysis software used in NDE.The software makes use of simple mathematical relations shown in equation 1 for generating digital signals with different known frequencies and sampling rate. Combinations of these individual frequencies will be mixed (see equation 2 for example) to create known standard signal conditions to check various performances of the signal analysis software.

A[ i ] = Ao Cos ( 2 P i f p / fs + f) ------------------ (1)

A[ i ]  : Cosine wave in digital form stored in an array.

Ao      : Amplitude of Cosine wave.

i          : Sample number (0, 1, 2 ............... N). N : Maximum number of samples.

f p       : Frequency of Cosine wave.

fs         : Sampling Rate of the digitizer.

f         :Phase shift in radians.

Ar[ i ] = A1[ i ] + A2[ i ] ---------------------------(2).

Ar[ i ]    : Resultant Cosine Wave in digital form stored in an array.

A1[ i ]   : First Cosine Wave in digital form stored in an array.

A2[ i ]   : Second Cosine Wave in digital form stored in an array.

PERFORMANCE CHECKING OF THE SIGNAL ANALYSIS SOFTWARES:

Example (1)
It can be seen from the Table 1 that the highest positive and negative amplitudes measured by the Analysis software used match with those of the standard signal generated by the presently developed simulation software. The small deviation seen in the measured value is not a measurement error. It is due to sampling of the continuous wave to form a digital signal. It is seen during the course of experimentation with different sampling rates that for low sampling rates of the order of below five times the Cosine wave frequency this deviation is more. If the value of the measured peak frequency is compared for serial numbers 1 & 2 in the Table 1, it can be seen that accuracy improves with the size of the window. The column number 5 reveals that the frequency scale is linear up to 16 MHz

Sl.no	Features of the Signals Generated using Equation - 1						Corresponding Features Extracted by Signal Analysis Software under performance evaluation.				Window size selected for Analysis (samples)
	Amplitude		1 *	2*	3*	4*	Amplitude		5*	6*
	+ve	-ve					+ve	-Ve
1	100	100	1	20	0	500	99.9	99.9	1.01	49.9	100-300
2	100	100	1	20	0	500	99.9	99.9	1.00	49.8	1-500
3	100	100	0.5	20	0	500	99.9	99.9	0.50	49.9	100-300
4	100	100	8	200	0	300	99.9	99.9	8.09	49.9	100-300
5	100	100	16	200	0	300	99.9	99.9	16.1	49.9	100-300
Table 2: Results of Performance Check by Analysing Signals Simulated by Software

(*1) Frequency of Cosine Wave in MHz.(*2)Sampling rate of digitiser in MHz.(*3)Phase shift.(*4) Length of cosine wave in number of samples.(*5)Peak frequency of the spectrum in MHz.(*6)Amplitude of peak frequency.

Example (2)
Signal Analysis software systems generally contain various displays in the frequency mode. Resolution of frequency scale at various sampling rates, expansion of frequency scale etc. may vary. Using the known simulated standard signals an assessment of the performance is made in the fig. 6 as an example.

Fig 6

Example (3)
It can be seen from comparison of column 1 & 5 of Table 2 that the analysis software shows error in frequency reading when sampling rate increases. Likewise it has been seen that certain analysis software systems fail in some other parameter under certain conditions.

Sl.no	Features of the Signals Generated using Equation - 1						Corresponding Features Extracted by Signal Analysis Software under performance evaluation.				Window size selected for Analysis (samples)
	Amplitude		1 *	2*	3*	4*	Amplitude		5*	6*
	+ve	-ve					+ve	-Ve
1	100	100	2	200	0	500	99.9	99.9	3.03	49.9	100-300
2	100	100	2	100	0	500	99.9	99.9	2.60	49.9	100-300
3	100	100	2	20	0	500	99.9	99.9	2.20	49.9	100-300
4	100	100	12	200	0	300	99.9	99.9	13.2	49.9	100-300
5	100	100	16	200	0	300	99.9	99.9	17.1	49.9	100-300
Table 2: Results of Performance Check by Analysing Signals Simulated by Software

(*1) Frequency of Cosine Wave in MHz.. (if any)(*2)Sampling rate of digitiser in MHz..(*3)Phase shift.(*4)Length of cosine wave in number of samples.(*5)Peak frequency of the spectrum in MHz. 6. Amplitude of peak frequency.

APPLICATION OF COMPUTER SIMULATION

In pulse echo ultrasonic system the sound pulses are generated and sent through the material. When this transmitted pulse encounters a reflector(discontinuity/ defect) certain part of the pulse energy is reflected back to the receiver. If another reflector exists in the vicinity, the reflection from that also reaches the receiver with a time gap. These two reflected pulses form a resultant pulse according to the principle of superposition.(The principle of superposition was applied in various fields of Physical sciences. In Ultrasonic simulation activity this principle is seen utilised for different purposes. Though not much relevant to this work one such paper is Ref.1). A single large reflector can be considered as an assembly of small reflectors of specific size(see fig.5b ). The resultant pulse for this large reflector can be simulated using the reflected pulse from a small reflector at same material distance(see fig.5a & equation 3). The reflected pulse from the small reflector (fig. 5a) is to be collected using the same transducer, equipment , material and material thickness.

Fig 5

Pr(i) = P(i) + P(i+d1) + P(i+d2) + P(i+d3) + P(i+d4) -------------(3)

P(i) = digital array of pulse reflected from small reflector(see fig 5a). i = 0 to pulse width in steps of sampling rate. d1,d2,d3,d4 = delay in arrival time of reflected pulses from small constituent reflectors(see fig.5). P(i+d1) , P(i+d2), P(i+d3), P(i+d4) = digital array of delayed reflected pulses. Pr(i) = resultant signal pulse array.

In this case(fig.5b) d1=d4. In fig.5c two reflectors are shown corresponding to reflectors 1 & 2 of fig 5.b. In general case for simulation activities the shape of the small reflector need not be rectangular. Experimental validation of fig.5c is done in fig 8d of example (4). Experimental validation of a case similar to that in fig.5b is shown in fig.4 of example(6).

Fig 4
Fig 8

Once the digital resultant signal array is obtained the analogue display is created using suitable display algorithm. Software makes use of this methodology to simulate different defect conditions. Some useful applications are detailed below.

Example (4)
Side Drilled Holes(SDH) are made on Graphite Blocks(see fig. 8). These SDHs are made to study the interaction of sound pulses reflected from 2 SDHs closely separated in this material. The resultant A-Scan (RF) signals obtained by inspection using 1MHz probe(B1SN) are given in fig. 1. The collected signals revealed that when the defects are closer of the order of pulse width or less the pulses interfere according to the principle of superposition. For the defects as close as of the order of 1mm or less(c & d in fig. 8) the signal conditions widely differ( c & d in fig. 1). The exact distances between the fabricated defects are not measurable in this case because of facility constraints and the SDHs are 20 mm deep inside the material.

Fig 1
Fig 2

Similar conditions were simulated very easily with different separations of the defects using the developed software system. The resultant signals are given in the figs. 2 b,c,d. These defect conditions are created from signal in fig.1a. by moving the defects 1 and 2 closer using the developed software. From the resultant signals so obtained and their correlation with that obtained from the SDHs in graphite samples fabricated, it is clear that computer simulation can be utilised to gather more information for interpreter.

Defects in fig. 8a are well separated(8mm in the direction of the sound path) and hence no interference(see fig.1a) . Defects in fig.8b are separated by a distance of the order of 3 mm(in the direction of the sound path) which is of the order of pulse width and hence the two pulses started to interfere(see figs. 1b & 2b ) .Defects in fig.8 (c & d) are still closer and hence merge into one. But slight path difference between the pulses reaching the probe changes the conditions of interference and result in drastic change in the resultant signal amplitude. In an unknown case the reason for this change in amplitude is not obtainable using conventional instrumentation and the interpreter's measurement of the resultant size of the defects is turns out to be wrong. To know more about the nature of interference the frequency mode of these signals were extracted and displayed in figs.3. From this spectral display it can be seen that the resultant signal shown in fig.1c of defect in fig.8c is affected by destructive interference of some frequencies to result in reduction in the resultant signal amplitude. These sort of simulation and experiment enhances the correctness of interpretation of unknown results.

Fig 3

Example (5)
During the conventional application of Ultrasonic NDE , the resolving ability of the transducer is checked using the calibration block as shown in fig.7. This is like "Go - No Go gage" and is applicable for only normal beam. Further understanding is not possible. In this area simulation can be done by the presently developed software to measure the resolvability of different transducers in applicable material. It can be used for angle beam inspection probes too. The only input required is the digital A-Scan signal using the probe under evaluation in the material at the required material distance. This application is shown in fig.7.

Fig 7

Example (6)
Application for simulating signal pulse of defects having different sizes is verified in this example. Side Drilled Holes of different lengths has been made on Graphite and compared with simulation.(see fig.4). In this example , the basic signal used for simulation is the smallest SDH.

CONCLUSION:

The computer integrated ultrasonic NDE facility developed in house have the capabilities for generating standard artificial signals, simulation of defect conditions and analysis. It is found quite useful in studying the interaction of sound with defects, for improving the accuracy of inspection and interpretation and for the performance check of analysis software systems. This can be used in situations where fabrication of certain defect conditions is difficult and also to gather more information from fabricated defects by changing the conditions artificially .

REFERENCE:

1. "Simulation Of Ultrasound Pulse Propagation In Lossy Media Obeying A Frequency Power Law" Ping He. IEEE Transactions on Ultrasonic, Ferroelectrics and frequency control.,Vol.45,No.1,January 1998.

© AIPnD , created by NDT.net

|Home|    |Top|