Acoustic pulsing - some questions answered

TWI Bulletin, May 1985

Peter Bartle

Peter Bartle, CEng, MIM, MWeldI, is Section Head; Structural Monitoring Studies in the NDT Department.

Some basic questions relating to the novel technique of acoustic pulsing for monitoring defect growth in large structures are posed and answered, and the technique is compared with other NDT processes.

News that The Welding Institute had devised, and was actively developing, the acoustic pulsing technique, ^[1] generated much interest. Questions are often asked relating to its capability or to obtaining a perspective on its application. Many enquirers also wish to relate acoustic pulsing to alternative techniques with which they are already familiar. A selection of typical questions and answers is presented here, which include comparisons with alternative techniques. Although the illustrations used are related to monitoring a major offshore structure, to provide a consistent basis for the comparisons, many components or structures, welded or otherwise fabricated could be monitored in this manner.

What is acoustic pulsing?

Acoustic pulsing was originally conceived as a simple means of monitoring (intermittently or semi-continuously) large defect tolerant structures for significant in-service defect growth e.g. 100mm surface crack extension. ^[2,3] The sensitivity of the technique however, has proved far higher than was anticipated and consequently the potential application range is widening.

The approach ( Fig.1) is simply one of passing a constant ultrasonic pulse into a structure; letting it travel through; and detecting and monitoring the resultant wave for differences between the initial reference signal and the later signals.

Fig.1. Acoustic pulsing: input signal to each transducer in turn, and check the signals received at those transducers within range, to check whether the same as the input reference signals

A change in the detected signal represents a change in the acoustic paths being monitored, produced for example by crack growth.

This technique can be used for:

1. detecting the occurrence of defect growth;
2. locating the defect growth (using the timing along the wavetrains for two or more transducer pairs to the first changes between the corresponding wavetrains);
3. using the magnitude of the change as an empirical guide to the extent of defect growth.

How sensitive is acoustic pulsing to defect growth?

This is a difficult question to answer at present - definition of sensitivity is the subject of ongoing study which is complicated because of variation with transducer spacing, etc. The introduction of holes as small as 2.5mm diameter into a steel plate has been detected with transducers 1 and 2m apart and the sites of such holes have been located to within 20mm. The sensitivity that can be achievedwill be dependent on structure geometry, transducer locations, transducer separations and whether steps, such as the use of a range of input signals, are employed to enhance sensitivity.

The technique is essentially aimed at detecting defect growth and normally a pre-existing defect cannot be identified until it grows. A philosophy popular amongst potential users is that of locating defect growth with acousticpulsing and then applying conventional techniques to inspect that area.

How does acoustic pulsing compare with acoustic emission?

Both techniques involve 'listening' with transducers, but in acoustic emission ^[4] the listening is for stress waves generated by deformation or crack growth within a structure ( Fig.2a), whereas in acoustic pulsing the listening is for pulses deliberately introduced into the structure, whether crack growth is occurring or not. Generation of the emission is not within the control of the operator of themonitoring system for acoustic emission testing, but for acoustic pulsing the operator controls the timing and the nature of the input pulse and hence the whole monitoring process.

Fig.2. Local monitoring techniques

a) Acoustic emission;

b) Local vibration analysis;

c) Ultrasonic inspection

Fig.3. Global vibration monitoring-spectra from natural or artificially induced vibrations are compared against a reference signal

Although acoustic emission should be a more sensitive monitor of crack growth than acoustic pulsing, the sensitivity in practice depends upon the detectable emission levels per mm ² of crack growth. This varies by orders of magnitude, ^[5] and sensitivity can be reduced correspondingly. Consequently, acoustic pulsing may be equally, if not more sensitive in some cases e.g. on tough ductile steels where detectable emission levels can be low.

The amplitude levels of acoustic emissions detected are often low, again especially with the tough ductile steels, and this limits the inter-transducer distances that can be used, typically to 2-6m. With acoustic pulsing, the signals detected are introduced by transducers, and the detected signal levels are much higher than for typical acoustic emissions for tearing in ductile steels at the same transducer spacing.

The input power can be greatly increased for acoustic pulsing, allowing it to be used with much larger inter-transducer distances: 50m has already been exceeded as a propagation distance and the data available show that working at up to 100m separations should not cause detection difficulties even on heavy section structures.

Long inter-transducer distances mean that fewer transducers and less cabling are needed, which in some cases e.g. offshore installations, is a major cost saving feature.

As acoustic pulsing transducers do not need to be monitored continuously, they can be monitored in sequence and periodically. This also helps to minimise the basic equipment requirement. It also means that the monitoring team need only be on site or on an offshore structure when a planned survey is scheduled.

Another difference between the two techniques is that acoustic emission relies on the detection of stress waves generated at the moment of defect extension so if an emission is not detected properly or is misinterpreted, then that potentially important item of data is lost forever. With acoustic pulsing, not only is it possible to check repeat readings and average the data obtained, but also to use a series of different input signals to give additional checks and increase sensitivity.

How does acoustic pulsing compare with vibration analysis?

Acoustic pulsing is based on the comparison of signals collected at intervals, the comparison being made in the time domain e.g. are the successive half cycles in the last received signal of the same amplitude as the corresponding half cycles in the reference waveform. Vibration analysis, however, is based upon monitoring the frequency spectra of structures, or elements thereof, for changes from a baseline or reference.

Vibration analysis ^[6,7] can be a global technique ( Fig.3) whereby the whole structure is monitored with very few transducers and the input source is either the natural dynamic loading on the structure or, for example, an oscillating 5t weight. In this manner, a large structure can be monitored using fewer transducers than are required for acoustic pulsing. The resonances most likely to change in the event of structural damage are the higher harmonics, however, which are difficult to detect and monitor because they are weak. Consequently 'global' vibration analysis is suitable only for detecting major crack growth or perhaps only the complete separation of a structural member such as a brace.

A localised and more sensitive form of vibration analysis ^[8] ( Fig.2b) involves treating separately individual braces and the structure in immediate contact with them, by mounting a vibrator and sensors on the brace each time monitoring is undertaken. For offshore structures this involves extensive diving operations for each monitoring session whereas ideally diving should be necessary for inspection purposes only.

How closely does acoustic pulsing relate to ultrasonic inspection?

Both involve the transmission and capture of ultrasonic pulses. For acoustic pulsing the input is made at a few constant positions with the objective of producing a pulse that will radiate through large amounts of, or all the structure. In ultrasonic inspection ( Fig.2c) the input transducer is traversed or scanned across the area of interest, while the pulses are injected in the form of a narrow beam to give accuracy when sizing a defect.

The pulses used for ultrasonic inspection are short and sharp, usually only one to eight cycles in duration. The transducer rapidly damps its own residual excitation, so that there is the minimum of a 'tail' to the drive pulse to detract from the detection of components reflected or diffracted by any defect, such signals often being weak. In contrast, the pulses used for acoustic pulsing are likely to be tens or hundreds of cycles long, because the changes sought are within the detected wave train. For both techniques the transmitting and receiving transducers can be the same or different, the former being more common in ultrasonic inspection and the latter being the norm in acoustic pulsing.

The sharpness required of the input pulse for ultrasonic testing, and the need to damp the residual excitation combine to limit the drive energy that can be injected into a structure. This added to the problems of beam divergence severely limit the range of ultrasonic inspection such that a transducer to defect distance of 150mm would be considered 'quite long.' In acoustic pulsing it is hoped that 'quite long' will prove to be well in excess of 50m.

What other techniques closely approach acoustic pulsing?

If only a small percentage of the potential of acoustic pulsing is realised, in terms of the combined capability of detecting defect growth of relatively small magnitude (and locating the defects), using large inter-transducer distances, then it has more to offer than virtually any other monitoring technique yet devised.

Two techniques exist which have some similarity to acoustic pulsing, however, both of which may be termed 'long range' ultrasonic procedures. One of these ^[9] was studied in the USA at Drexel University and probably comes nearest to acoustic pulsing. The differences lie mainly in the analytical technique applied to the received signal. The success of The Welding Institute approach stems from progressive comparison of the received signals which gives high sensitivity plus a location capability. The Drexel system involves an 'en bloc' comparison, which reduces sensitivity to a degree where not only does major defect growth need to occur for a change to be recognised, but also the defect growth needs to be in or close to the line of sight between the transducers. Inter-transducer distances are also likely to be much shorter with the system developed at Drexel.

The other system examined at Harwell ^[10] used the transmission of Lamb waves over long distances for the inspection of tubes. Lamb waves are the basic propagation modes of the acoustic pulsing technique, but the Harwell approach sought to detect a discrete defect reflection, rather than changes within the rf signal from the Lamb waves.

What equipment is required for acoustic pulsing?

In essence the equipment requirement for acoustic pulsing is a signal generator with suitable measurement and control of its output, possibly a power amplifier, transmitting and receiving transducers (which in many instances can be of identical construction) leads, and a capability for recording and possibly averaging the signals ( Fig.4). Signal analysis can be performed manually, but is tedious so computer analysis is preferred.

Fig.4. Basic equipment requirement for acoustic pulsing (computer control and analysis strongly recommended)

The problem with using standard laboratory equipment for acoustic pulsing is the need continually to check the drive pulse for amplitude and frequency. Consequently, a purpose built system with a computer to set and check the drive signal, switch between transducers, store and analyse the data is preferred.

Is data handling and analysis difficult?

The basic steps that need to be followed are:

1. verify the input signal;
2. collect the received signal and compare immediate repeats;
3. compare the new received signal with the reference signal to find the first change along the signal if one exists;
4. use the timing of any first change found in iii to calculate the locus of possible sites of defect growth;
5. repeat i-iv for all other appropriate transducer pairs (transmit and receive), and input signals;
6. compute any loci intersections to establish the sites of defect growth;
7. if required estimate the extent of defect growth from the magnitude of the signal change.

These are all relatively straight forward analytical operations. The most difficult are iv and vi, but these are similar to the standard location procedures used in acoustic emission testing.

The immediate difficulty is that of specifying the level at which a change of signal is considered significant. This difficulty exists because experience to date is limited. Also if the technique is to be used for estimating the extent of defect growth, rather than for locating areas for detailed inspection, then work is likely to be required, for each application, to establish the co-efficient relating defect growth to signal change for vii. If the system is to be operated under conditions where the noise level is high relative to the amplitude of the received signal, then additional steps to reject noise and/or signal average will be required. Again, however, these are standard and/or straight forward procedures.

A greater problem than coping with i-vii is likely to be reducing the volume of data that has to be captured and permanently stored to an acceptable level. The scope for enhancing the capability of the technique by repeating signals and using a series of different input signals, makes it possible to generate enormous volumes of data.

Work on the condensation of data and the definition of which data are significant is therefore important.

Could the technique be applied immediately?

The short answer is yes, but with some reservations. Apart from ensuring adequate reproducibility, setting up equipment and establishing procedures should be far from onerous undertakings. The drawback that needs to be accepted is that analysis of the data would have to be undertaken in the absence of background information on normal levels and patterns of variation in signals when defect growth was not occurring. The Institute can offer some help and guidance, but to a limited extent since some of the knowledge and understanding gained to date, are the property of the members of the Sponsor Group who have funded one of the work programmes.

What is the present position?

First, to protect the interests of Research Members, The Welding Institute has made a patent application ^[2] Second, The Welding Institute has embarked upon and intends expanding a Group Sponsored Project to evaluate and develop the techniques, which will help define the techniques capabilities and limitations, whilst establishing the procedural and equipment requirements for the successful operation of the technique (see below). This project will be supplemented by another within the Institute's Co-operative Research Programme, and by single sponsor projects for individual Member companies.

In the initial stages of the work on this technique a considerable amount of information has been amassed, which has confirmed the basis of the technique and is widening the application areas.

References