it depends on the task. For example, one have to use different distances and positions for sound localization (see Dr. Fredö's answer) and noise suppresion.
This implies that you want to measure acoustic particle velocity which can be done using two microphones, where you should first read this document to get your bearings
http://www.bksv.com/doc/br0476.pdf
or, better, using a microflown U probe
http://www.microflown.com/
There are other ways as well but the two above mentioned are the more common techniques.
Just to sum up - I guess you are aware of the fact there are three kinds of pressure and that these are very different beasts?
Line Pressure (Ambient) which is a scalar and static, i.e. affects density but does not shift mass
Kinetic Pressure (wind) which is a vector and provides a net transport of mass.
Pulsation (sound) which has two components, pressure as a scalar and an oscillating flow (acoustic particle velocity) without net transport of mass where this flow is a vector.
As you operate a high pulsation as compared to the line pressure, I suggest that you keep track of the %LP = 100*Pulsation/LP as this is a measure of linearity and, as, it affects shock waves from wave steepening.
You see, a shock wave forgets what it was doing upstream its collapse. This implies that any frequency response function will have zero correlation when shocks collapse - which can be irksome.
Gas is not as sensitive as is water as its B/A factor is 0.2 whereas it in water can go all the way up to 10,000.
Still. wave steepening matters in music instruments when you have standing waves and moderate losses so I am guessing they might matter also for you.
Hope this helps
Claes
P.S: - You find lots of good reading material here
If you looking for a basic starting point I highly recommend reading John Eargle "The Microphone Book" there is a difference between measure particle velocity and pressure gradient. As mentioned Microflown make PU probes which measure particle velocity but are very expensive. Using two microphone you can simply subtract one from the other to get an estimate of the pressure gradient or pressure derivative with respect to sensor spacing see (Gradient Microphone) Chapter 4. As with many beam forming techniques be aware of white noise gain which you will suffer as that sensor spacing gets smaller.
FYI - If I understand things correctly, Rifat wants to measure with particle velocities of 8 m/s to derive a relation between flame power and input flow.
For a free wave, that would be in the ball park 160 dB re 2E-5 Pa or 3.2% of the ambient pressure, i.e. things are starting to get non-linear.
To me this suggests that one needs robust techniques. That said, I am all for cheap technologies and as things are nonlinear anyhow, one might perhaps do just as well with an approximate sensor like a pressure gradient mike?