Possible causes for error in Coriolis Meter measurements

On 8 Nov 2006 19:17:39 -0800, "Stew Peake" proclaimed to the world:

So you have two identical meter measuring the flow in the same pipe? What is the configuration of the pipe? Why are you using two measuring devices?

I'm assuming you are transferring fuel and the metering is for quantifying the transfer. I have not worked much with this type of meter, but with flow meters in general, I have found that the manufacture will be very helpful in assisting you with finding the cause of the error. The flow meter industry rely on their meters performing as advertised.

The up and down stream conditions effect the accuracy of any flow meter and these errors increase with flow rate. I suspect that the errors might also be due to the lower density fuel having volatiles in it that are coming off in the low pressure pockets formed in eddies around any connection or rough spots in the piping. I don't know what the rated accuracy is for these meters. From your information, it looks to be 0.5%. Remember that one meter can be 0.5% high and the other 0.5% low, so you could have a 1% error between the two meter and they would still be within the rated accuracy. If this is the case, then there is a 0.2% additional error. That 0.2% of total flow by the meter primary element could be in the form of vapor, not liquid. 1500 L/min is moving pretty good.

If you can slowly increase the flow rate and record reading at several points, then graph the points. If cavitation is happening, you should see the point where this starts. Limit your transfers to a rate lower than this.

This is all assuming that I am guessing your application correctly.

Are you correcting for temperature at each end? Usually fuel pipeline flows and densities are referenced to 15degC...

Cameron:-)

"Stew Peake" schrieb im Newsbeitrag news: snipped-for-privacy@b28g2000cwb.googlegroups.com...

We once used a setup like this for leak detection in a product pipe line. This flow comparison works only if you have steady flow of the same liquid. During transition phases (change of flow rate or different liquid) we expirienced a so called "line pack", meaning that for a while it looks like there is more stuff flowing out than into the pipeline and vice versa, which led to many false alarms. It was decided to put a time delay on those alarms and another leak detection system was added.

Dieter

Mass (Coriolis) flowmeters usually have integral RTDs to compensate for temperature, which is needed for mass flow measurement. Find out if the model you're using has temperature compensation. I would also check to make sure that the meters are both configured for the same minimum flow measurement. I've worked with Micromotion flow transmitters that had the capability to ignore low flows, assuming they were an empty tube. Ensure that both meters are set up the exact same way. And it's absolutely critical in any flow measurement application that the tube itself is always full during flow. Perhaps the piping arrangement and meter runs aren't up to spec and the lower density fuel is evidence of it. HTH.

Stew, What is the composition of the lighter fraction? Are there air bubbles at the tranmitting end? Entrained air bubbles can cause the mass flow meters to be less accurate. Also have you checked for vibrations in the pipe line? Regards, Raj

Thanks everyone for your advice.

I now have a number of avenues of investigation available including:

- check configuration of pipework & pressures to ensure volatiles should not be present.

- check configuration of pipework & meter location to ensure vibrations are not a problem.

- check temperature compensation within meter.

- check that both meters are zeroed correctly.

- perform some tests gradually increasing flow rate (if possible) to see at what rate the error starts to become unacceptable.

- consult with the meter vendor for why their meter is out of spec.

To answer the queries that were made along the way:

1) The meters are not identical, one is a DS300 series Coriolis Meter from Micro Motion (+/-0.15%) at the outlet, while the other is a CMF300 from Micro Motion (+/-0.05%) at the inlet. I therefore expect them to read at worst +/-0.2%. We are using the two meters so we can detect leaks of fuel.

2) The pipe is single inlet, single outlet, starts sea level goes over a hill of ~300m height and ends back at sea-level.

3) We are correcting the measured flow for temperature at each end.

4) The flow discrepancy still appears well after there is the same grade of fuel in the whole pipeline. We intend to compensate for line-pack, but will still be using time-delay to allow start-up conditions/transients to settle.

Thanks again for your help.

Regards,

Stew Peake.

On 12 Nov 2006 14:58:56 -0800, "Stew Peake" proclaimed to the world:

It sounds like you have a good plan. Since the two meters are of different manufacture, you might have a problem of each manufacture pointing at the other without having anything to back this up. You also might run into the problem of them wanting to do things that are not too feasible in your situation, like packing the meter up and sending it back.

Another possible tool to use is a portable ultrasonic doppler meter. It has a lot less rated accuracy but repeatability is good enough to get some ideas of what is going on in the pipe. Having a third meter to compare with might give you the clue you need.

Also, few meter manufacture rate their meters so simplistically. Most likely the full specs make reference to upper and lower limits on accuracy.

I have done some projects where leak detection implemented by comparing two flow meters was tried. The only way it works is by developing an algorithm that evolves over time. For instance, I would start sampling the inputs only after a steady flow rate was established for a period of time. Then I found that my acceptable error value had to change at different flow rates. Once you gather data you can minimize if not eliminate the errors due to metering. Within a few months, the algorithm evolves into something pretty free of false alarms while still giving alarm during a real event in a time span acceptable. If leak detection is your only goal, then it is doable.

What seems to happen tho is all this data makes you aware of the errors that were always there and it drives you crazy. Particularly in the case of metering that is being used to base money exchange on, people start worrying about how much money they were shorted. I hope this project does not open that can of worms up for you. Best of luck.

Stew, on a 7km Jet-A1 pipeline we are currently commissioning, the issues that made us forget precise flow monitoring at both ends and to revert to flow measurement at the inlet compared against tank level measurement at the outlet, were:

- if a rain cell moved across the middle of the line during a transfer our flow alarms would go off due to localised cooling (and we are using old custody-approved SmithMeter turbines, not expensive mass flowmeters). The operators started referring to them as the "rain alarm"!..

- We found that the pipeline was very often under vaccuum (as much as -60kPa!) whilst not in use, which added to the line packing issues on start-up as described by Dieter and made it difficult to determine when (if ever!) the pipeline was in steady-state flow.

..and of course a neat side effect of this system is that it's not possible to do a Buncefield ;-)

If you are serious about pipeline leak detection, may I suggest installing a 'Hansa TCS'

or similar Static pipeline leak detection system. The static systems are extremely accurate and, in my experience, Dynamic leak detection is more trouble than it's worth. It doesn't take many false alarms before the operators turn the alarm off forever and go back to watching TV...

I hope this helps, Cameron:-)

I'd reconfigure the meters to output mass flow and see how well they agree.

Micromotion flowmeters calculate volumetric flow by measuring mass flow and density at the meter, so any change in density will affect the volumetric reading.

Of course, the comments above were made assuming you have the transmitters configured to display volumetric instead of mass flow. If you have ProLink or the handheld programming unit you can easily monitor any of the outputs (temperature, density, mass flow, or volumetric flow.