Re: R.Freemantle: Multilayer automotive polyamid fuel tubes :
: : Richard,
: : Do you have experience in testing the layer thickness
: : of those particular tubes and some suggestions for it?
: : The total thickness is 1 mm and containing 3 up
: : to 5 Layers. The thinness layer can be in a range of
: : 0.1 mm.
: If the minimum layer thickness is in fact 0.1 mm (and not thinner), there is a chance this could be done with a conventional precision thickness gage and a high frequency broadband transducer (20 to 30 MHz). We have had some experience in this area.
: The usual problem is that the barrier layer is bounded by even thinner adhesive layers on each side, which are too thin to resolve even at these frequencies. Depending on the thicknesses are relative acoustic impedances of the barrier layer and the adhesive layers, you get a a pair of echoes that represent some constructive/destructive wave combination at each boundary. Thus, the echo patten sen by the gage becomes
: outer layer/varrier plus adhesive layers/inner layer. Because of the unpredictability of the interference effects as adhesive thickness changes, there can be some imprecision in the measurement of the middle layer when doing peak-to-peak echo timing with a conventional gage. Accuracy needs to be established with the aid of reference standards of known thickness.
: : About a similar theme the April UTonline issue reported about a new patent from
: : Ford. What is your opinion about the methode used there?
: This fuel tube situation differs from the fuel tank application described in the Ford patent, because the polyamid tubing is both much thinner and less attenuating than polyethylene fuel tanks.
: We have had experience in the auto fuel tank application with direct pulse/echo measurement of barrier layers at 10 MHz. This did, however, involve somewhat thicker barrier layers (0.5 mm nominal) than those described in the Ford procedure.
: I'd be happy to discuss specifics with anyone interested.
: --Tom Nelligan
I agree with Tom, you probably could use welldamped high frequency probes in this application. The problem of the thin bondlines may be overcome by applying some signal processing algorithms. One possibility is cepstral processing - I have seen
good results with this technique and it is possible to get significant improvements in resolution.
An excellent paper which tackles this problem is:
P.N. Dewen et al. "Improving the resolution of ultrasonic echoes from thin bondlines using cepstral processing"
J. Adhesion Sci. Technol. Vol. 5, No. 8, pp 667-689 (1991).
For layer thickness characterisation, or at least determining when the thickness of the structure goes out of specification, it may be possible to look in the frequency domain at the resonances excited in the structure. This would also reduce sensitivity to the
adhesive bondlines. I did some work on charactising thin polymer layers ( < 50 microns) where the wavelnegth of the ultrasound was less than the polymer thickness, and I found frequency domain technqiues to be of greatuse.
The method desribed in the patent is interesting and highlights the problem of pulse-echo measurents on attenuating media. The tone burst idea is not new and I think similar results could be achieved by studying the through thickness resonances.
Some time ago our laboratory did a lot of work on characterising very lossy materials (mainly rubbers) many of which were like acoustic black holes !! Pseudo random sequences (Golay Sequences) and cross-correlation methods where used to characterise the
attenuation and velocity. The main benefits of this technqiue are: you can propagate low frequency signals but get wide bandwidth/ high resolution measurements, you get excellent signal to noise improvements, and you can pre-filter (highpass) your input
sequences to de-emphasize the low pass effect of the sample attenuation. I don't have any references to hand but I could dig them out if any one is interested.