SOUND INTENSITY MEASUREMENTS

[COBWARD]

Hi
Does anyone in this forum know of or have any opinions on the use of sound intensity measurements for derivation of sound power levels?
Also does anybody know about the validity of this technique for use in connection with the machinery directive?

Thanks

 

[GregLocock]

Opinions I've got by the bucket load.

SI was a fad we all lived through about 20 years in the automotive world. We learned a lot, including that it was all a bit hard!

1) Whatever the theory and salesman says you cannot easily make good measurements in a noisy environment.

2) keep the mics well away from the radiating surfaces, otherwise you measure the bizarre loops where energy is pumped out by one panel, and then absorbed by its neighbour, where your estimate of overall sound power will be more difficult. A pint of virtual Guinness to the best sounding theory of why they are there. I have a theory...

3) Generally the easiest and most reliable sound power estimates are from 6? microphone free-field hemispherical measurements

4) What's the machinery directive? Cheers

Greg Locock

 

[MikeyP]

I used SI to measure the sound power of a 30 MW gas turbine backup generator at a power station in Malta a number of years ago (let's face it, you are not going to turn down a 1 week job in the southern Mediterranean in the 1st week of January when you are living in Scotland!). On the whole it worked rather well.

In this instance, a free-field measurement was not appropriate as the main oil fired turbine hall was located in a building close by and the whole facility was built inside a near semicircular quarry. I followed the ISO standard for intensity measurement using the "scanning" approach as the structure was so large. ie moving the probe over a pre-defined discrete area of the structure whilst averaging for say 16 seconds (as opposed to the point-by-point surveying approach). This gives a much coarser representation of the radiation behaviour but is generally less prone to error.

I also used the best equipment. Just lashing any old pair of mics together is never going to work. Perhaps the most important aspect is the callibration, which was re-done every hour or so, as this defines the limits of what is real usable signal and what is noise in a much more stringent way than for a normal pressure measurement. I was told that the calibrator I hired cost as much as the probe and analyser put together.

BTW. Malta has only 2 power stations; not much of a national grid. On the last survey of the day when it was nearly dark, we radioed the control room to turn up the generator from idle to 15 MW and the lights across the whole island got brighter!

M

 

[MikeyP]

In response to Greg's comments:

I thought the whole point of an intensity probe was that it only measures NET power flow. Your "bizzare loops" (near-field radiations) are an entirely rective phenomenon, for every Watt heading away from the structure in the nearfield zone, ther is 1 Watt heading the other way in the next half of the cycle. This enables the intensity probe to distinguish between the near-field and far-field.

Here is one of my infamous handwaving explanations of how nearield radiation occurs...

Think of the ideal case of a simply supported rectangular panel. When it is vibrating at a particular frequency, the mode shape of the panel comprises a sine wave with an integer number of half wavelengths in both the x and y directions. This forms a rectangular grid of peaks and troughs across its surface. As you say, the the positive pressure "source" at a peak is cancelled out by a the negative pressure "sink" at an adjacent trough. The pressure fluctuation passes from one to the other and never "escapes" to the far-field.

HOWEVER, in the next half of the cycle, each source becomes a sink and each sink becomes a source (peaks become troughs and troughs become peaks). The pressure fluctuation travelling beween a source and a sink is a wave like any other. It can only travel at the speed of sound. THEREFORE IT CAN ONLY BE CANCELLED OUT IF IT REACHES THE SINK BEFORE THAT SINK BECOMES A SOURCE AGAIN IN THE NEXT HALF OF THE CYCLE.

Slightly more technically, the cancellation can only occur if the wavelength of the bending wave in the panel is shorter than the wavelength of the acoustic wave at the same frequency.

Now if we plot a graph of the speed of waves vs frequency for both the acoustic wave and the bending wave, the line representing the acoustic wave is flat (a constant 343 m/s). However, the line representing the bending wave increases with increasing frequency (in fact it increases as the square root of frequency). So at some point on our graph, the lines must cross. The frequency where this occurs is the "coincidence" frequency also called the "critical" frequency. Below this frequency there is cancellation and hence reduced radiation as energy is tied up in the near-field. Above the coincidence frequency there is no cancellation and hence strong radiation.

Note that there is still some radiation below the critical frequency. The corners (and in some instances the edges) of the panel have no neighbours with which to cancel. In this case we have 4 small radiating areas which do allow some energy to escape to the far-field, but the radiating area is much smaller than that of the whole panel and hence the amount of radiation is reduced.

Make mine an extra cold one.

M

 

[GregLocock]

I will.

OK here's mine

In the far far field the machine behaves as a point source. In the near near field the machine is a collection of tiny pistons, each radiating the same frequency, but with almost random (but stationary) phase and amplitude compared to its neighbour.

Somehow we have to generate a nice coherent spherical wavefront from a collection of tiny sources.

The bizarre loops are an adjustment in the energy flow so that the spherical wave can be generated by a pulsing spherical piston of air.

I am sure in practice that this is the same as your theory. I got a few splinters in my fingertips before I realised that the coincident frequency between discrete components is when the speed of vibration waves is the same as the speed of the acoustic waves between them, although since the bizarre loops have a variable path length (some come quite a long way out of the engine) I think that may be a variable feast, to get theological.
Cheers

Greg Locock