I don't think there is an official rule, but my practice has always been that the pressure recorder have a range that puts the test pressure in the middle 1/3 of the calibrated range. So, with a 600 psig test I would be happy with a gauge calibrated either from 0-900 psig (0-1000 psig is ok) or 0-1800 psig (0-2000 psig wouldn't bother me).
Most pressure measurement devices have a sweet spot around the middle of their calibrated range that has the lowest uncertainty that the device can record. It is nice to be in the sweet spot, but the uncertainty of today's devices is so good that being outside it isn't really a major problem. Just try to stay out of the first and last 10% of the range (don't use a 0-10,000 psig calibrated range for a 100 psig test, resolution gets to be a problem)
Zdas, my understanding is that most sensors or instruments are rated for accuracy as a percentage of full scale. Thus the greatest accuracy as a percentage of actual reading would be near full scale of the instrument.
Compositepro,
That is just wrong. If an instrument has an uncertainty of +/-0.5% of calibrated range, and the range is 0-10,000 psig then every point between zero and 10,000 psig is +/- 50 psig. A reading of 100 psig is 50-150 psig. A reading of 9,000 psig is 8950-9050 psig. Both have a dead band of 100 psig. Saying that the 100 psig dead band is somehow less important at 9,000 psig than at 100 psig is an odd statement.
Accuracy is a very imprecise term. Pressure measuring devices are rated on "uncertainty" and "repeatability". The rating is based on the worst uncertainty coupled with the worst repeatability in their possible calibrated range. So if a gauge is rated at +/-0.5% of calibrated range, that is the worst you will see, but the number might actually be +/-0.001% in the center of the range (but you can't rely on it).
+/- 0.05 means you don't know the actual pressure, but there is nothing to say that the instrument could not be indicating dead on at 501 psig, and at 9999 psig too. Its just you can't be sure. The uncertainty is not necessarily constant, its a maximum. For example, the +/- 0.05% is the guaranteed maximum uncertainty, but nothing to say its not +/- 0% at 2/3 scale, +0.05% at full scale and -0.05% at low range, or any combination thereof.
**********************
"The problem isn't working out the equation,
its finding the answer to the real question." BigInch
+/- 0.5% of 10,000 psi full scale is +/-50% of a 100 psi reading and +/-1% of a 5,000 of a 5,000 psi reading. That's pretty basic. What that means is for greatest accuracy/precision/repeatability in measurement, you select a gauge where the measurement will be close to full scale. There may be other factors involved in scale selection like maximum process pressure which require you to compromise on accuracy.
If that is the way you want to look at it, I really hope that you don't design any tests around where I am. Your logic says that for a 100 psig test you want to use a 0-100 psig gauge to maximize "accuracy". What it actually does is fail to register transients over the gauge range. Not a good idea.
The range effectively doesn't matter, as long as the gage covers the pressure you're interested in reading. If the gage is +/- A, as far as reading it is concerned, and that's all you'll ever going to know its saying, is its X +/-A anywhere and everywhere on the gage from 0 to N.
If you need a more accurate reading, buy one that's +/- 0.001
**********************
"The problem isn't working out the equation,
its finding the answer to the real question." BigInch
The code doesn't call this out in detail, but it's commonsense. What function does the hydrotest serve? How precise is 1.3x or 1.5x the MAWP x the ratio of stresses? Does it really make any practical difference to its validity if the test is carried out at 1.295 or 1.29x, or 1.305 or 1.31x the MAWP?
If you're using a commercial bourdon tube pressure gauge, conventional wisdom is to pick one where the test pressure falls between 1/3 and 2/3 of the gauge's maximum range. That's adequate even if the gauge is only +/- 5% accuracy. But that's unnecessarily restrictive IF you choose a gauge or transmitter which has better accuracy, has been calibrated in the recent past (and you can prove it) etc.
What they don't want to see is a +/- 5% gauge being used at 1/10th of its full-scale, such that your test pressure is now uncertain by ~ 25%.
If you're using pressure loss as an indication of leakage, the stability of the measurement device and its readability or resolution (the ability to detect small changes in measured value) are also important.
The specifications you referenced do not give details on pressure gage accuracy; however, Section III (NB-6412) does.
The range of analog gages must be not less than 1-1/2 nor more than 4 times the test pressure.
Digital type gages may be used without range restriction provided the calibration and readability error does not exceed 1% of the test pressure. As Compositepro mentioned, pressure gages are rated by manufacturers at a % of full scale. The III-NB requirement of 1% of test pressure sets a lower bound on use, but would allow the digital gage to be used to full scale.
Temperature rise - the pump operator slips, the switch sticks, the relief valve doesn't, the .... breaks and the pressure rises past the expected limit.
Do you "really, really, really", not want to be able to prove you didn't go to 1500 psi on your system when the test gage max'ed out at 1250 psig - just because you're too cheap to buy a gage accurate at mid-range for your 1000 psi test?
If you use a 1050 max-reading gage at a test pressure of 1000 - what max pressure did you actually go to when it hit the stops?
