DECI NEWSLETTERS AND REPORTS

THE DECI REPORT - OCTOBER 1996

ACOUSTIC EMISSION FROM LEAKS

I postulated in my last newsletter that a high pressure leak through a plate would appear as an out-of-plane source of acoustic emission and result in a low frequency flexure wave in the plate. This hypothesis was tested recently by a simple experiment which involved drilling a 0.023 inch hole near the end of an undamped steel bar, epoxying a water hose fitting to the bar and turning on a domestic water faucet which is approximately 65 psi in pressure after screwing a hose into the fitting. Figure 1 is a photograph of the experimental setup. The SE9125-M transducer was used to determine the type of signals present, and the SE1000-HI with a 4.5 inch long, and a 6 foot long pointed waveguide was also used to show the utility of pointed waveguides for leak detection.


The SE9125-M transducer (false aperture transducer) which has been used in past experiments on this same bar has shown the capability of identifying the difference between an out-of-plane (OOP) and an in-plane (IP) source of AE (reference 1). This is accomplished by measuring the peak voltage ratio of the AE signal after passing it through a 100Khz hipass filter, and a 20-70Khz bandpass filter. An OOP source for this transducer exhibits a high frequency/low frequency ratio of 1 or less while an IP source gives a ratio of greater than 1 depending on the source depth in the plate.


Figure 2 shows the results from the SE9125-M and the SE1000-HI. Note that the high frequency/low frequency peak voltage ratio for the SE9125 is greater than one, which indicates an IP source very near the surface of the plate. Most of the AE from the leak is probably generated just prior to the water exiting the hole into the atmosphere.

Note that the low frequency signal from the SE1000-HI is higher in amplitude than the high frequency signal. This is to be expected from this small aperture mass loaded sensor which is primarily most sensitive to surface displacements due to the low frequency flexure wave generated by the leak, and fairly insensitive to IP signals. The 1 volt signal after 60dB of amplification, and consideration of the calibrated sensitivity of the SE1000-HI used for this experiment, indicates that approximately 1 picometer of displacement of the surface of the bar was taking place due to this leak.

I mentioned in my previous report that the SE1000-HI with a pointed waveguide should be a good candidate for leak detection in pressurized bodies, since no couplant is required, as well as useful in monitoring small displacements occurring in a borehole due to earthquakes. Two waveguide configurations were prepared. In both cases the waveguide material was 1/4 inch diameter 316 stainless steel rod. An SE1000-HI was mounted on one waveguide 4.5 inches long. The other was mounted on a waveguide 6 feet long in order to determine the attenuation of the displacement wave in the rod. In both cases the waveguides had a 45 degree angle sharp point at the end which gives the equivalent of a zero aperture. The waveguides were attached to the same bar used for the leak experiments with a DECI Mag80 magnetic holddown shown in figure 3. It provides approximately 10 pounds of spring compression for the waveguide and 80 pounds of magnetic force.


Figure 4 shows the data from both waveguide configurations when placed at 24 inches from the leak.

High frequency components show some attenuation when going from the 4.5 in. long waveguide to the 6 foot long waveguide. The low frequency components of the signal seem to be unaffected by the difference in length.

AE signals from a leak show a great deal of variability from one moment of time to the next. No two signals captured during this experimental work were alike, even though the leak rate appears to be constant when measured over a period of a few minutes.

DISCUSSION OF RESULTS

I anticipated that the AE leak signals would have a high frequency to low frequency ratio much less than "one". This would indicate that the leak was an OOP source. This was not the case, the ratio was slightly greater than "one" which indicates that the source was IP but very near the surface of the plate. There was a significant low frequency flexure wave present which generated enough displacement of the surface that a displacement transducer such as the SE1000-HI could easily detect it with only 60dB of amplification. The surprising result was that a pointed waveguide up to 6 feet long could be attached to the transducer and detect the picometer displacements in the bar without any additional amplification.
These results could have important utility in detecting leaks in underground pipelines. It has been shown in previous DECI reports on damped and undamped plates that the fundamental antisymmetrical lamb wave that produces these low frequency despersive flexural waves in plates is fairly independent of damping material attached to the plate, while higher frequency modes of this same wave are easily damped. Therefore one would expect that a leak in an underground pipe that is damped by fluid inside and earth or insulation on the outside would not have much influence on the attenuation of these low frequency waves. It is shown in this report that a 6 ft. long waveguide with a pointed end is effective in detecting displacements due to a leak. It is anticipated that a much longer waveguide could be used if necessary and still be effective. Therefore one could envision putting a carbide tip on a pointed waveguide of this type, and with the help of a small vibrator, push it through the ground until contact is made with an underground pipeline to be checked for leaking.
Another application where the pointed waveguide concept could find a place is in monitoring high temperature pressure boundaries for leaks. One of the problems encountered presently with using AE transducers attached to waveguides for monitoring high temperature structures is the difficulty of coupling the waveguide to the structure to be monitored. The pointed waveguide eliminates this coupling problem, since no couplant or weld attachment is necessary.
A displacement sensitive transducer is required in order to achieve good results. If one attempts to attach a standard AE transducer to a pointed waveguide for this type of monitoring, the results will be disappointing

NEW PRODUCT

We are happy to announce a new transducer to further broaden the product line of DECI. For those of you that are testing small specimens and require a very small high frequency transducer the PICO-Z should do the job for you. This transducer is also works well for burnish and glide testing of disk media and head-disk interference studies. The very flexible low noise cable allows large movements of the transducer attachment without introducing noise.

REFERENCES

1. Dunegan H.L. "Use of Plate Wave Analysis in Acoustic Emission Testing to Detect and Measure Crack Growth in Noisy Environments." Proceedings of Structural Materials Technology-An NDT Conference, San Diego California February 20-23, 1996.