by Terry_Zakis » Wed May 26, 2004 1:29 pm
Hello Fellow Flow Bench Enthusiasts,
I'm new here, but new to flow, engineering, and motor work.
I think that jsmith is really onto something here with the concept of the flow inside of a pipe, as compared to plenum flow, especially when using an orifice plate. The same would apply to all other bernoulli devices (venturi, pitot tube, anubar, etc).
The previous unregistered guest was also exactly correct, in stating that the Superflow style (and MSD type) benches will show differences in flow, with changes in head orientation on the bench. This is due to the proximity of the orifice to the flow discharge, and the fact that the flow is not stabilized before entering the orifice plate in relation to the plane of the pressure taps.
I've been a practicing Engineer for 15 years (power plants, steam turbines, heavy-duty gas turbines, and more recently aero-derivative gas turbines of Pratt & Whitney), and from all of my research, there are only two industry recognized codes for measuring flow with orifice plates in closed pipe. There is a separate code for plenum flow that the vacuum motor people use, but that's a separate case, and I don't think it can provide the level of accuracy of a properly set up orifice system.
The most recognized code is the ASME MFC-3M, which deals with measuring flow inside of closed pipes, with orifice plates. Another piping designer which I read comments from (in another post here) was also onto this with estimations of straight lenghts of pipe before orifice plates.
Estimating, compressible gas flow accurately is the most difficult thing to do in process measurement, as there are numberous chances to introduce uncertainty into the measurement.
The beauty of the MFC-3M code, despite it's complexity, is that measurement of flow within closed pipes, that follow the code requirements, can determine flow within +/- .5 to .8%, depending on the Beta Ratio (Orifice ID / Pipe ID), and most importantly, do that without a separate calibration.
Frankly, and with all due respect to those running benches, I don't believe all the hype circulated about being able to measure overall flow to within 1/2%, as so many claim. The +/- .5% to +/- .8% that you can acheive from the MFC-3M code is also based on an U95 Uncertainty. Meaning that there is a 95% probability that the "true" flow value is within a band of +/- .5% to +/- .8%, but this is also dependant on the beta ratio, the uncertainty of the differential pressure transmitters, and gas temperature measurement (i.e. Superflow type benches that heat the air stream).
The need to be so close on determination of flow is typically called "custody transfer" and is the highest level of flow measurement accuracy available, and is used for flow measurement of fuel flows (for power plant and gas turbine acceptance testing). So if specially designed flow sections that determine pass/fail criteria on multi-million dollar contracts can only measure within 1/2% at best, we shouldn't go on fooling ourselves that we can do better.
On a practical side, the key of course is repeatability. I think we can all live with a bench under or over reporting flow by 1%, as long as the data is repeatable. That way you can determine changes to port flow on an accurate basis.
As we comtemplate these flow determinations, keep in mind that flow straighteners are a big help. Watch for recommendations that call out to have 10 upstream lengths and 5 downstream lengths, without disturbance, from your orifice plate. Most deisgners are aware of this, but the fine print states that these distances are required when suitable flow straighteners are in alredy in place, and when all all bends are in the same plane.
There are published data, ASME Fluid Meters, Part-I, which you can't even purchase anymore, that show flow can be upset for as much as 80 diameters, after out-of-plane turns, without flow straighteners.
For these reasons, I think use of some type of plenum is beneficial, so that your flow measurement is not affected by the discharge from the port. Transitioning back into a pipe smoothly is another factor, and then you still need to get your straight lengths in.
Sorry about the long post. Wish it were more simple. Best case is to develop a flow bench, and use a separate flow "prover" system to tell you what the bench is doing. This is how I'll be setting up my flow bench/lab. I'll have one wall of my shop (26 ft.) which is have ASME MFC-3M, code compliant flow sections, with four pipe diameters, four different orifice sizes to optimze Beta ratio. Air flow will be routed in series, from through the provers, and then through the flow bench. This way I'll be able to try different designs and see which more closely match the flow measured by the provers.
My flow bench? Not there yet. Almost. I've spent over $7,000 in the last 1.5 years acquiring all of my components and doing research. The only thing left for me now is to decide on the geometry of my design. I will be using orifice plates.
Well back to work!
Best Regards,
Terry_Zakis
:O