Orifice plate delta P fluctuations
Orifice plate delta P fluctuations
(OP)
I apologize in advance for the long post, but have had no luck finding good info on my problem so far.
We use a flow bench based around a Daniel Simplex #073C orifice plate holder to test air flow through production valves. We have recently upgraded our data acquisition system and are finding unexpected results of the instantaneous readings from the delta P transducer.
The flow bench is constructed in accordance with ASME PTC 25 and ASME "Fluid Meters" 6th Edition, 1971. Pipe size is 3". The orifice plate holder uses square edge orifices and flange taps and most of our testing will use orifices with a beta of .200 or .339. Control valves are at least 28D upstream and straightening vanes are at least 11D upstream of the orifice plate. Test readings (over a period of about 10 secs) are taken only after all flowing pressures and temperatures have stabilized (which make take 5 minutes or more, depending on the valve under test).
The delta P transducer is a 4-20ma, 5-psid unit with a nominal 0.1% accuracy. We calibrate the unit over a range of 0-125 in H2O using a water manometer and precision regulated air supply. Data acquisition software samples the transducer at 166Hz and then averages every 83 readings in order to provide a display update to the flow bench operator every 0.5 secs.
However, when flow testing we find the displayed delta P is not very stable, even though we are dampening the fluctuations by averaging (as described above) and all other system pressures and temperatures are rock steady. This instability makes it difficult for the test bench operator to make final adjustments during the test.
If we look at the raw (not averaged) data over a 10 second run, we see even more surprising variations. For example, the 1,660 (10 secs at 166Hz) delta P data points collected from a recent test (.339 beta, 175.5 psig, 54.5 deg F, 160.8 SCFM air) showed an average of 9.9in H2O, but a range from 14.1 to 5.8! The standard deviation for the set was 1.25. A frequency plot of the data points looks like a standard distribution.
We have done a lot of work to eliminate the causes of electrical or software program "noise" and feel confident that they are not contributing to the issue. We have also disassembled the Simplex holder as far as we could to make sure there was no dirt, wire edges, etc anywhere that could affect our readings.
There are two flange pressure taps 180 degrees apart on both the high and low side of the orifice. We connected both high side taps (using a tee) to the high side of the delta P transducer (and did the same for the low side taps) thinking we could average pneumatic noise, but there was no improvement.
We also tried installing gauge snubbers on both sides of the differential pressure transducer and reran the same test as above. In this case, our average delta P was still about 10in H2O, but our max was only about 11 and our min was about 9. The standard deviation was 0.31. Therefore, the gauge snubbers have shown that the delta P is in fact pneumatically "noisy".
Way back when manometers or bourdon tube gauges were used to measure delta P, these fluctuations were not noticed. Our previous data acquistion system had much lower resolution and accuracy than our new setup and that is why we have not seen this until now.
We have talked with Daniel and they believe it is normal given the obvious turbulence in the flange tap areas. They said the spacing of our upstream control valves and flow straightener are slightly under the API 14.3 standards and may also contribute to our measured fluctuations.
I have done a fair amount of research into this topic, and have found only a couple of papers on the subject. One paper discussed orifice errors at low delta P, and the other was regarding signal noise ratio comparison between an orifice plate and a proprietary flowmeter. Both papers used a ratio of the standard deviation of the data sample to the average delta P as a way to characterize the pneumatic noise. Both papers showed typical ratios to be well below 1%, albeit their testing was done at larger betas. Our testing has this ratio at over 10%.
So, my question is should we be looking for some other reason for the pneumatic noise? Or, is this just the way these devices operate?
If in fact the pressure signals are this noisy, has anyone every used snubbers (or the like) to dampen the signals? Due to hardware and software limitations, we cannot easily increase our sample rate to electronically dampen the fluctuations.
Thank you!
We use a flow bench based around a Daniel Simplex #073C orifice plate holder to test air flow through production valves. We have recently upgraded our data acquisition system and are finding unexpected results of the instantaneous readings from the delta P transducer.
The flow bench is constructed in accordance with ASME PTC 25 and ASME "Fluid Meters" 6th Edition, 1971. Pipe size is 3". The orifice plate holder uses square edge orifices and flange taps and most of our testing will use orifices with a beta of .200 or .339. Control valves are at least 28D upstream and straightening vanes are at least 11D upstream of the orifice plate. Test readings (over a period of about 10 secs) are taken only after all flowing pressures and temperatures have stabilized (which make take 5 minutes or more, depending on the valve under test).
The delta P transducer is a 4-20ma, 5-psid unit with a nominal 0.1% accuracy. We calibrate the unit over a range of 0-125 in H2O using a water manometer and precision regulated air supply. Data acquisition software samples the transducer at 166Hz and then averages every 83 readings in order to provide a display update to the flow bench operator every 0.5 secs.
However, when flow testing we find the displayed delta P is not very stable, even though we are dampening the fluctuations by averaging (as described above) and all other system pressures and temperatures are rock steady. This instability makes it difficult for the test bench operator to make final adjustments during the test.
If we look at the raw (not averaged) data over a 10 second run, we see even more surprising variations. For example, the 1,660 (10 secs at 166Hz) delta P data points collected from a recent test (.339 beta, 175.5 psig, 54.5 deg F, 160.8 SCFM air) showed an average of 9.9in H2O, but a range from 14.1 to 5.8! The standard deviation for the set was 1.25. A frequency plot of the data points looks like a standard distribution.
We have done a lot of work to eliminate the causes of electrical or software program "noise" and feel confident that they are not contributing to the issue. We have also disassembled the Simplex holder as far as we could to make sure there was no dirt, wire edges, etc anywhere that could affect our readings.
There are two flange pressure taps 180 degrees apart on both the high and low side of the orifice. We connected both high side taps (using a tee) to the high side of the delta P transducer (and did the same for the low side taps) thinking we could average pneumatic noise, but there was no improvement.
We also tried installing gauge snubbers on both sides of the differential pressure transducer and reran the same test as above. In this case, our average delta P was still about 10in H2O, but our max was only about 11 and our min was about 9. The standard deviation was 0.31. Therefore, the gauge snubbers have shown that the delta P is in fact pneumatically "noisy".
Way back when manometers or bourdon tube gauges were used to measure delta P, these fluctuations were not noticed. Our previous data acquistion system had much lower resolution and accuracy than our new setup and that is why we have not seen this until now.
We have talked with Daniel and they believe it is normal given the obvious turbulence in the flange tap areas. They said the spacing of our upstream control valves and flow straightener are slightly under the API 14.3 standards and may also contribute to our measured fluctuations.
I have done a fair amount of research into this topic, and have found only a couple of papers on the subject. One paper discussed orifice errors at low delta P, and the other was regarding signal noise ratio comparison between an orifice plate and a proprietary flowmeter. Both papers used a ratio of the standard deviation of the data sample to the average delta P as a way to characterize the pneumatic noise. Both papers showed typical ratios to be well below 1%, albeit their testing was done at larger betas. Our testing has this ratio at over 10%.
So, my question is should we be looking for some other reason for the pneumatic noise? Or, is this just the way these devices operate?
If in fact the pressure signals are this noisy, has anyone every used snubbers (or the like) to dampen the signals? Due to hardware and software limitations, we cannot easily increase our sample rate to electronically dampen the fluctuations.
Thank you!





RE: Orifice plate delta P fluctuations
First, I'd like to complement you on an excellent post, it clearly states the problem and the steps you've taken. Good job.
Second, Daniels was right--it's not broken. The process starts with Bernoulli's Equation which had a long list of assumptions that allowed a closed-form equation. The assumptions put the overall repeatability in the range of +/-5%. We get around that by averaging, taking infrequent samples, and using pretty wide pens on a chart recorder and generally pretending that we don't know that "steady state" in a flowing stream lasts a couple of miliseconds at a time. Pushing a pipe full of gas through a little hole and then letting it relax back to the full pipe is quite chaotic.
I had a project once that I collected 1000 Hz data. The differential latency in the measurement devices made sure that none of the three inputs ever reported on the same physical data point as the other two at any given time. My variation was a lot more pronounced than yours.
What is the uncertainty on your pressure, dP, and temp devices? If they are each 0.5%, then the combined uncertainty is 1.5% (1-0.995^3), any variation less than 1.5% is meaningless. For example, if your dP transmitter is 0-125 inH2O then any reading within 0.6 inH2O of any other reading is exactly the same number. With the gauge snubbers your standard deviation was half your uncertainty--you can't expect to do better than that. Without the snubbers, the standard deviation is only twice the uncertainty. I figure that if my standard deviation is less than 10 times my uncertainty then I have a statistically relevant result. You are way better than that.
I can think of two things that should help some (but they will not get rid of the variability)
1. Throw all of your 0.2 beta plates away (the uncertainty below 0.25 increases exponentially)
2. Move your meter to upstream of the control valve you're testing. It has to be introducing its own flow perturbations so you can remove that source of uncertainty.
Good luck.
David Simpson, PE
MuleShoe Engineering
www.muleshoe-eng.com
Please see FAQ731-376: Eng-Tips.com Forum Policies for tips on how to make the best use of Eng-Tips Fora.
"It is always a poor idea to ask your Bridge Club for medical advice or a collection of geek engineers for legal advice"
RE: Orifice plate delta P fluctuations
One thing not completely clear in my original post is that the control valves are used by the operator to adjust the air flow going into the orifice plate. The valve under test is mounted in a large receiver well downstream of the orifice plate. The control valves are used to attain a certain pressure in the receiver while the valve under test is open and flowing to atm. The flow rate across the orifice is thus the same as the flowrate through the valve under test. Therefore, we can't do much about relocating those control valves per your suggestion.
Regarding the gauge snubbers, have you ever heard of anyone using them in a situation like this one? The folks at Daniel did not seem real excited about this, saying they were not aware of anyone doing it. Fortunately, we have a calibrated sonic nozzle we can substitute for the valve under test, so it is relatively easy to see if the snubbers have much effect on the calculated flow rate across the orifice plate.
Thanks again,
Frank
RE: Orifice plate delta P fluctuations
I've never seen an actual snubber on the impulse lines from a secondary element to a recording element , but I've seen a LOT of needle valves that had "slipped" off their back seat to an intermediate position that was pretty near to closed. I know why this is not strictly according to Hoyle, but on EFM I haven't seen it hurt anything (and on charts I've seen it dampen a painted chart to the point of being legible).
If I was you, I'd run the sonic nozzle against your tube and compare 1 min cumulative production. I bet you find that on that scale you are well within published tolerances. If you try to match sub-second instantaneous rates you'll get really discouraged.
David
RE: Orifice plate delta P fluctuations
RE: Orifice plate delta P fluctuations
RE: Orifice plate delta P fluctuations
Pressure 175.500 psig(190.196 psia )
Temperature 54.5 Deg F
Ambient Pressure 14.696 psia ( msl )
Base Conditions (STP) 14.696 psia / 60.0 Deg F
Molecular Weight of Gas 28.960
Specific Gravity (G) 1.000
Flowing Specific Volume (Vf) 0.9944 ft3/lb
Specific Volume @ STP (V) 13.0967 ft3/lb
Z Factor (Redlich-Kwong) 0.993
Z Factor @ Base Condition 0.999
Isentropic Exponent 1.459
Viscosity 0.017 cP
**** FLOW DATA ****
normal | maximum
Flow 161 | 161 Scfm
Pipe Reynolds No. (Rd) 87343 | 87343
Mean Fluid Velocity (Pipe) 4.0 | 4.0 ft/Sec
Line Loss 0.01 | 0.01 (psi)/100 ft
Friction Factor (f) 0.0210| 0.0210
Meter Loss(Installed) 0.30 | 0.30 psi
**** METER CONSTANTS ****
Meter Bore (d) 1.0407 @ 20 Deg C
Pipe i.d. (D) 3.0700 Inches
Beta Ratio (d/D) 0.3390 Inches/Inch
Metal Expansion (Fa) 0.9998 (316 Stainless)
Expansion Factor (Y1) 0.9995
Reynolds Factor (Fc) 0.9966
Drain/Vent Corr. (Fh) 1.0000
Discharge Coef. (Cd) 0.59963
Spinks Factor (S) 0.06961
**** TRANSMITTER DATA ****
normal | maximum
Meter Differential (h) 9.8 | 9.8 in w.c.
Flow Rate 161 | 161 Scfm
Q (Scfm) = 5.982*S*D^2*Fa*Fc*Fh*Y1*V*sqrt( h / Vf )
**** ACTUAL FLOW CALCULATIONS ****
min | normal | maximum
Meter Diff (h) 5.5 | 9.8 | 14.1 in w.c.
Flow Rate 116 | 170 | 192.6 Scfm
try moving the flow straightener or at least inspecting it
you are getting velocity fluctuations of +/- 1 fps
RE: Orifice plate delta P fluctuations
However, I am wondering about the derivation of the flow rates under "Actual Flow Calculations", especially the "normal" result of 170 SCFM at 9.8 inH2O dP. Why is this not the same as the 161 SCFM of above?
I believe you provided this info to show what flow rates (and thus velocities) are seen at the plate to cause the dP range I am seeing, correct?
I did inspect the flow straightener, a tube bundle conditioner about 8" long in good shape about 31" upstream of the orifice plate. It would be a chore, but I could probably move it up to another 24" further away from the orifice plate. I have (hopefully) attached a sketch of the arrangement.
Thanks,
Frank
RE: Orifice plate delta P fluctuations
How far is the receiver from the orifice plate, and could you lengthen that run? If nothing else, just to de-tune the receiver from the orifice. Or run a dead-headed pipe somewhere on that run to act as a quarter-wave damper...if the noise is harmonic...
RE: Orifice plate delta P fluctuations