For head type flow meters the rules are generalizations.
Fifty years ago the mercury manometers were mostly ranged 0-100 inches water column. That has been the throw-down starting point since.
However, if low pressure and compressible than one guide is the maximum inches water differential corresponds to the maximum operating pressure. Thus, if the operating pressure is 20 psig, start with 20 inches water. You don't want the permanent loss to be a large percentage of the inlet pressure.
For high pressure applications you can use higher differentials. Where necessary people increase to 200 inches water. That is typically the maximum for a new design. One might use still higher differential on a plant modification.
Normally I just specify 2 psi maximum pressure drop. However, it seems like any orifice plate that someone from instrumentation buys for me ends up being greater than that - more on the order of 3-4 psi. It really depends on how big of a range of flowrate you are trying to measure.
The flow calculations (in gas phase) fall apart when the ratio of DP (in H20)divided by flowing pressure gets too high. There is a table in AGA report 3 and it's something like 1.5 to 2.0 So a 100"/100psia = 1 is ok, but 50" at 20psia will yield unreliable numbers.
On liquids, you limits are how much force you can exert on a plate. Again, AGA 3 publishes that limit. If you want to go higher, you can use a a thicker plate. I've seen as high as 1000" of H2O used.
My personal preference for orifice plates, which is the result of much feedback from instrumentation folks over a long period of time:
I like to use 200" W.C. as my "throw-down number".
Admittedly, this is on the high side.
I quickly will drop to 100" or even 30" W.C. when 200" might not be available.
I don't go as quickly down below 30" W.C., but will do so if pushed.
I consider 10" W.C. to be close to a practical minimum for an orifice plate dP flow measurement. It is too easy to introduce errors into your measurement at lower dP.
The above numbers are for the range of the dP transmitter, not the operating dP.
why does a 10" diferntial have more errors than a 100", a rosemount transmitter states it accurracy as percent of span. If the process introduces errors, then its a process error not meter error.
It is easy to not appreciate the loss of acccuracy that occurs with this type of change. We tend to think that there's an adjustment screw that you can just rotate to go from 0 - 10"W.C. to 0 - 500"W.C. (or the equivalent in software). In fact, the working parts of the instrument have a particular accuracy; they will respond to a change of (say) x"W.C. If your range is 0-10", then the likely error can be thought of as x/10; if it's 0-500", then the errror would be x/500.
Actually, things are more complicated than that. I've seen vendor's literature that states incredible accuracy for their instruments. In fact, a transmitter on a test bench under ideal conditions does have a very high level of accuracy. Field instruments set and then neglected for long periods of time don't do nearly as well. There are many potential sources of unexpected errors as well. If you're measuring a vapor flowrate using an orifice, for example, you might get a small slug of condensate that forms in your instrument's impulse line. Capillary atractive forces could easily hold it in place, and it may deaden the transmitter's response. With a large differential, this might go unnoticed, but with a small differential it could cause an appreciable error.
As you might infer, I believe that it is much better to think of instrument errors in terms of +/- some value and NOT a % of span.
djack77494, wrong, the newer instruments can be adjusted, I suggest you visit rosemounts web page or the others honeywell, yokagawa, ect and get caught up.
Doog's point goes beyond the instrument. A Rosemount 3051S is very accurate and very rangeable in terms of pressure differential. However, the orifice plate and associated fluid dynamics invoke much greater error than the transmitter accuracy for measuring the flow.
JLseagull, not according to API, if you follow the standard.
The coefficient of discharge will be less than .5%. The DP transmitter and pressure transmitter will be less than .25% each (including calibration and drift errors.) The square root sum of the squares of all parts of the metering will be less than .75% for a typical meter
THere seems to be a general lack of appreciation that in a field installed measurement system of ANY kind, you will never get anywhere near the accuracy the forum entries here would lead me to believe. I like to think of such a system in terms of an orifice plate with pressure differential transmitter used to infer flowrate. Please humor me here and let's talk in terms of this system. What level of accuracy do you think is achievable?
WRONG.
You will not get very good accuracy no matter what Fisher Rosemont, Yokagawa, Honeywell, or anyone else says. You will have moisture and scale and unexpected contaminants in every part of your sensing system. The installers will not have done all that you expected. Connections will have leaks and obstructions. Valves may even be inadvertently closed. The sensors will drift and change with time. The fluid's pressure, temperature, and composition will differ from what you have calibrated it for. Your velocity pattern will not be fully developed. Erosion, corrosion, deposition, and maybe even plate bending will alter the geometry of the orifice. All of the above will vary with time. I think you're self dillusional if you tell me you can measure fluid flowrates in industrial settings within fractions of a percent in accuracy. And the smaller the range of the physical parameter you are measuring (dP in this case), the greater the error as a percentage of the measurement. I know you can turn a screw on an old Foxboro dP cell and change the range of the transmitter to nearly anything you'd like. The diaphragm remains the same and is no more sensitive because the range is changed. Why would modern electronic instruments behave differently?
Current electronic dp transmitters are quite accurate compared against the Foxboro model 13. One can expect the system accuracy on an orifice flow measurement to be near 1% within the top 80% of the dp range, perhaps better than 2% within the top 80% flow rate range. As the Reynolds number diminishes (sounds technical, right?) the accuracy diminishes more rapidly.
I agree that depending on the application high degrees of accuracy may be required and are achievable...at a cost. To maintain high accuracy requires a lot of care in the design and installation, and it requires a significant effort to maintain that accuracy. "Run of the mill" measurements often fall short of the accuracies mentioned in this thread.
Orifices flanges are acceptable by the standards, you just need them machined correctly and have centering pins. A "Jr" or simplex fitting isn't much more than a set of flanges.
