Suspect orifice flow meter installation 1

[jimhokie]

I have a question about an orifice flowmeter installation being used to verify a pump is putting out its design flow and head. It has a 2.68" orifice in a 4" pipe, with 58" of straight pipe upstream and 15.75" straight pipe downstream. The question deals with the first downstream fitting: a spec flange with a 1.93" orifice prior to discharging into a tank. The pressure upstream of the flowmeter (pump discharge) is about 700 psi with a flow of ~80F water at supposedly about 2000 gpm. Can the downstream spec flange orifice be interfering with the DP across the flowmeter orifice and throwing off the flowrate measured by the orifice flowmeter? The DP across the flowmeter should be about 150psi for 2000 gpm, but it's only 135 psi corresponding to about 1875 gpm. I'm thinking the flowmeter is not indicating correctly as opposed to the pump not operating on its normal curve. I'm far from a flowmeter expert beyond knowing you want at least 10D or more straight pipe upstream and at least 5D downstream. But could an even smaller orifice in close proximity downstream of the flowmeter throw it off?

 

[zdas04]

Problems with orifice plates in liquid flow always make me wonder "is the pipe running full". The empirical equation for testing if a pipe is running full (10.2*ID^2.5, with ID in inches and the answer in gpm) says that you need 326 gpm to be full so that isn't the problem here.

The downstream orifice is unlikely to be a problem because at your flow rate communication of flow pertubations upstream is pretty difficult. Not impossible, just unlikely.

Your beta ratio is a little on the high side, but should give you reasonable uncertainty.

I'd check for a backwards plate before I did anything else. Next I'd check the water analysis in the RTU.

Finally, I'd look for a way to divert the pump discharge into a tank for a few minutes to do a bucket test.

David Simpson, PE
MuleShoe Engineering

Law is the common force organized to act as an obstacle of injustice Frédéric Bastiat

 

[LittleInch]

If I've worked this out right you have a velocity of about 14 m/sec in your 4"pipe. That's some speed.

I'd be very interested to see your orifice plate. If it's perfectly round and the same size as you think it is I'll be surprised at that sort of velocity.

I suspect if you remove the 150psi(!) DP from your system aa result of the orifice plate it might flow a bit more as well...

If you only want to check the flow have you thought about using a strap on UT meter out a mag flow to avoid creating any impact on the flow you're trying to measure?

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[hacksaw]

You have not described an installation that would be considered suitable for a meaningful flow measurement. Required accuracy? Duration of the test? Budgeted cost?

 

[Duwe6]

"is the pipe running full".

It is unusual for horizontal process pipe to be full -- go watch some glass pipe in a pharma plant. Numerous 'standing waves' and a little over half full is the norm, not the exception.

Switch to a Mass Flowmeter, you will be impressed with the accuracy.

 

[zdas04]

Duwe6,
I think that you are trying to enter a conversation that you do not have adequate preparation for.

Pipes run full. Or they don't. It is not magical or mystical. It can be calculated (that is sort of what Engineers do). A 4-inch pipe with less than approximately 326 gpm will not run full. A 4-inch pipe with more than about 326 gpm will run full. Not magic. Engineering.

Any flow meter will increase its uncertainty dramatically if the pipe does not run completely full ("accuracy" is a meaningless term, the concept that uneducated people usually mean by "accuracy" is covered by people who do fluid measurement for a living as a combination of "uncertainty" and "repeatability"). Those "glass pipes" you are talking about in pharma plants (they are not glass by they way) are carrying less fluid than the equation I included would predict for the minimum flow rate to run full. It is common for lines to not run full. It is also common for other lines to run full. 2000 gpm in a 4-inch line will run full.

As to "mass flow meter", that is a marketing term, not a measurement term. There is no meter in the world that measures "flow rate" (either mass flow rate or volume flow rate). It cannot be done with our knowledge of fluid flow (and it possibly can't be done at all, we just don't know). Any meter will infer a velocity from some parameter that can be measured.
[ul]
[li]A differential producer (like an orifice meter) will infer a velocity from a dP across a known orifice with known fluid characteristics and convert it to a volume flow rate by assuming a flow profile.[/li]
[li]A turbine meter will infer velocity from the rotational velocity of a set of blades.[/li]
[li]An ultrasonic meter will infer a velocity from either time of flight of sound waves (requires assumptions about fluid characteristics like density and inherent speed of sound in that media) or the Doppler shift in the sound waves (requires the same assumptions).[/li]
[li]PD meters assume that the chamber is totally full every time the disk rotates and counts the disk travel.[/li]
[li]etc.[/li]
[/ul]

Coriolis meters evaluate the vibrations of a pipe to infer a density and the shift in the position of a bent pipe to infer a velocity, add in a bit of arithmetic and a few dozen assumptions and you can infer a mass flow rate from these readings. They do OK in many applications, and they have a way with constant chemical properties to infer a change in density with some amount of change in temperature and pressure. The are not magical, and are certainly not a silver bullet in this application. By the way, the uncertainty of a "mass flow meter" in a line that is not full is approaching a perfect random number generator--the numbers are not related to the flow rate at all.

In any liquid flow, you are better off with a volume flow meter than with a "mass flow" meter. In gas flows, there are times that mass flow is better than flow rate at standard conditions, but not many. The difference is marketing not fluid measurement.

David Simpson, PE
MuleShoe Engineering

Law is the common force organized to act as an obstacle of injustice Frédéric Bastiat

 

[LittleInch]

I assume the pipe running full debate is valid for a horizontal pipe with an open end? Many pipes can be filled and then run much slower if both ends are kept under liquid level or pressure control and don't vapourise. Vertical pipes flowing up will surely run full regardless of flow, but vertical pipes down with a free end have a different formula?

Anyway I agree with zdas04 that for the vast majority of flow measuring devices, they only work in single phase fluids. Measuring two phase in a pipe is nigh on impossible except with special two phase meters and their accuracy leaves a bit to be desired.

Getting back to the OP, who has gone rather quiet, he doesn't say what the fluid is. A change in temp or density will impact the orifice calculation also, but I continue to think that a meter which has such a big impact on the pressure and creates a flow restriction of reasonable proportion, isn't normally a good idea. Nearly all meters to some extent affect the flow, but usually only to an insignificant degree.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[jimhokie]

Thanks for all the feedback. A little more background...I'm just the guy cognizant of the pump in the system while others have designed the temporary piping system and instrumentation to verify the pump flowrate prior to complete system construction and certification, and still others are running the testing. So only now that there appears to be a "pump problem" via an apparent under flow do they come to me with details looking for answers. My short answer is that the pump is fine, but I need to lead them to righteousness regarding their system and instrumention to prove to them that the pump is fine.

No doubt the pipe is running full...2000gpm in a 4" pipe! I think the beta ratio flagged by David might be the problem. I've also suggested they verify correct orifice orientation (I've seen dumber mistakes made here). I think they may have used an orifice more typically used in more common (to our applications) schedule 40 pipe, while this instance is sch. 160. For the 2.68" orifice they are using in a 4" sch. 160 pipe, I get a beta ratio of .78 vs. 0.2-0.6 recommended for liquids (water in this case). So I've suggested they revise that to a 1.375 to 2.0" orifice (beta ratio of 0.4 to 0.58, shading to the high end given our flowrate).

About the flowmeter causing a significant restriction mentioned by LittleInch...this is just a temporary instrument for flow measurement to ensure the pump is operating on its expected head-flow curve near its maximum intended flow, but it's a good point that if they need to use a smaller orifice in the flowmeter to get the correct beta ratio, they should also revise the orifice in the spec flange at the pipe exit to the tank, such that the combination of flow restrictions properly simulate the final as-built system resistance so the pump is still running near its desired flow rate.

I've also been told now that they are adding an ultrasonic flow meter. I'll report back once I know more.

Thanks again for the input everybody!

Jim

 

[bimr]

Your flow rate equates to a velocity of around 50 ft/sec which is extremely high and will be associated with high Reynolds numbers that will throw the readings off.

In typical industrial flow rates, the maximum liquid velocity is at 30 ft/s (10 m/s).

Where do people come up with statements like this: "A 4-inch pipe with less than approximately 326 gpm will not run full." I would like to see the source of that statement. Velocity has nothing to do with a pipe flowing full as long as the minimum velocity to push bubbles downstream is reached (3-5 ft/sec depending on slope).

 

[bimr]

http://ballots.api.org/copm/cogfm/ballots/docs/Chp22-2ballot1939.pdf

 

[hacksaw]

What a discussion...

Idiotic orifice plate installation, idiotic d/p of 150 psi for flow measurment no less, and claims that you do not have to worry about full pipe installations.

Oh yes, to be sure the pipe will be full until you get to the orifice plate...

The OP installation will have an measurment accuracy of +/- 25-30% on a good day.

Some real tips today...

 

[zdas04]

bimr,
You are confusing velocity and volume flow rate. I have seen partially full pipes flowing at very high velocity, and whether the pipe is full or not has nothing to do with "pushing bubbles". It is about how much fluid is trying to get through a horizonatal pipe. The empirical equation that describes the phenomena is in my first post in this thread. If you want to read more about that equation see thread124-177112.

David Simpson, PE
MuleShoe Engineering

Law is the common force organized to act as an obstacle of injustice Frédéric Bastiat

 

[jimhokie]

At such a high flow rate through this size orifice, is cavitation likely occurring at the orifice, which I presume would throw off the DP to flow correlation?

 

[LittleInch]

Cavitation, erosion, bending of the plate... I've never seen anything like this level of DP and expecting to get any sort of accuracy or repeatable data.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[btrueblood]

I'd add in the question, what flow conditioner is used between the pump (or last fitting) and the orifice plate. Swirling flow from a pump (or other source) can persist for a lot more than 10 pipe diameters, and can throw a good 10-20% error into orifice flow measurements, and similar errors into most other types of meters.

 

[bimr]

zdas04

Here is what you posted above: "I think that you are trying to enter a conversation that you do not have adequate preparation for." Ditto.

That equation that you reference is just a quick and dirty estimating tool that you are misusing in this application. The equation is not a hydraulics principle or theory.

katmar stated "Perhaps the reason for the difference is that Durand and Márquez-Lucero are answering a different question to the others. The Durand and Márquez-Lucero article seems to be looking at the flow required to ensure that an open ended horizontal pipe will be completely full at the discharge point."

http://eng-tips.com/viewthread.cfm?qid=289211

Also refer to the attached reference.

Note theat when a pipe has a flooded suction with a fluid that will not offgas, it is possible to have full pipe flow at much lower fluid velocities.

Every municipality in the world is pumping water at economical fluid velocities of 3-5 ft/sec. When was the last time that you observed air being forced out of your tap?

 

[zdas04]

bimr,
I guess a masters degree with an emphasis on fluid mechanics and thermodynamics, a Master's Thesis on fluid measurement, 3 years of working as a Measurement Engineer for one of the 10 largest corporations in the world, consulting to the AGA 3 committee, a half dozen projects at CEESI, and several presentations to SwRI don't qualify me to talk about fluid measurement. Some would call that adequate preparation, but I guess you need something additional in the "Civil/Environmental" arena. Do you want to share what that might be? Remember we are not talking about cross-country municipal water pipes at 125 psig, we're talking about an industrial, in-plant, horizontal, liquid measurement application.

The two questions that Katmar talks about in your link are horizontal flow vs. vertical flow. For vertical flow the Froude number method works well (I've used it in predicting wellbore flow). For horizontal flow (like this thread is talking about) it is grossly inadequate. When Katmar talks about the equation I listed above being to ensure that an open-ended pipe is full on exit, he is kind of right (as he is usually right), but the continuity equation would say that the flow any distance upstream of the exit would have the same mass flow rate as the flow rate at the plane of the exit so this equation works for any horizontal pipe.

For cross-country pipes, vertical sections can run full when horizontal sections don't. I find in real flows that if my mass flow rate is high enough to satisfy this empirical equation then I get rebound on the downhill side of hills and the manometer calculations work (i.e., the only elevations that matter are the start point and end point). If the flow is less than that amount, then I don't get rebound on the downhill runs and the manometer calculations do not work. The equation that you are sneering at has proven to be a very effective predictor of the performance of something like 30 produced water gathering systems around the world that people have hired me to evaluate. It works pretty well to predict pipe performance in real flows.


David Simpson, PE
MuleShoe Engineering

Law is the common force organized to act as an obstacle of injustice Frédéric Bastiat

 

[bimr]

One suspects that it is the difference between fluid mechanics and hydraulics.

Fluid mechanics is basically a theoretical foundation.

Hydraulics on the other hand is a branch of science concerned with the practical applications of fluids, primarily liquids, in motion.

There are millions of fluid applications that will flow pipe full at less than the 8 ft/sec that you have stated whether in a industrial plant or municipal water main. Are you saying that they are all incorrect?

The same reference where you obtained your equation states a typical fluid velocity in an industrial plants is 4-6 ft/sec.

Regarding "For cross-country pipes, vertical sections can run full when horizontal sections don't." I would say that this is an example of poor engineering design.

 

[zdas04]

One suspects that there is less difference than you might imagine. I've worked with real flows a heck of a lot more than I've been in classrooms. The tenets of fluid mechanics do actually work in real flows. What you call "hydraulics" I suppose means "flow of water at fairly low pressure and variable rate/variable demand". What ME's call hydraulics generally involves Stokes' flow.

Most real liquid flows are markedly slower than the 13 ft/sec (API's errosional limit for pure water). Which means that most real flows will not have full pipe on horizontal runs. The equation that you are sneering at simply points out where it is not reasonable to depend on a pipe being full (e.g., in a flow measurement station). I'm not sure what your point is, and I'm becoming progressively less interested in digging out what it is.

I'm just going to ignore your "poor engineering" comment.

David Simpson, PE
MuleShoe Engineering

Law is the common force organized to act as an obstacle of injustice Frédéric Bastiat

 

[bimr]

zdas04,

You are confusing the concepts of sealing flow and pressure pipe flow.

Sealing Flow: If the flow in a pipe is great enough, the pipe is said to be “running full” or is sealed at its discharge point. This precludes the possibility of air entering at that point and getting beyond the liquid seal.

http://books.google.com/books/about/Practical_process_engineering.html?id=89JTAAAAMAAJ

http://www.globalspec.com/reference...ing-sealing-flowrates-in-horizontal-run-pipes

Pressure Pipe Flow: Refers to full water flow in closed conduits of circular cross sections under a certain pressure gradient.

http://www.learncivilengineering.com/wp-content/uploads/2012/05/closed-pipe.pdf

Sealing flow has nothing to do with full pipe flow in pressure conduits. That is why it is possible to have full pipe flow at low velocities in pressure conduits.