Coriolis mass flow meter problem


Thread Starter

Alejandro Posada


I have a main line of methanol that splits in two identical lines. The flow on those two lines is measured by two identical Coriolis mass flow meters. The methanol passes then through two separate pre-heaters before entering two vaporizer exchangers.

My problem is the following: when the temperature drops below -10ºC (14ºF), the HI-flow alarm on one of the meters is triggered once or twice per week, shutting down the entire plant. The problem occurs ALWAYS on the SAME LINE and it lasts for 30 seconds max.

We switched the two meters but the problem remained the same. The two meters are also fed by the same power line from a UPS. Finally, a technician from the manufacturer confirmed that both devices are in perfect condition.

Can anyone help me. I considered cavitaion in the pipes, but it's doubtful.

Alejandro Posada

Chris Jennings

I have problems with chlorine flow measurement with coriolis flow meters if the temperature gets too low. What I found was the problem was liquid chlorine formed in the pipe and the mass flow meter was set up for chlorine in gas phase. Because the flow is now two phase flow the flow meter is unable to calculate the correct mass.

Determine if the location of the flow meter is at a low spot or somewhere liquid can form. I found our problem only occurs after a start-up which has been during winter at night. This allowed the liquid to form and then on start-up the two phase flow caused the measurement problems.

You could also check the lagging on the pipe, or use thermal imaging to look for cold spots. Remember that thermal imaging cameras don't work on stainless steel, you will need to coat it with something thermally conductive to get a good image.

Tell me how you go with my suggestions.

Chris Jennings

Alejandro Posada


Thanks both of you for your suggestions.

Methanol is liquid at temperature ranges of -97ºC (-143ºF) to 65ºC (149ºF), so two phase flow shouldn't be a problem.

We called someone in to analyse the vibrations in the lines because Coriolis mass flow meters are sensitive at certain frequencies. According to a study on the effect of mechanical vibrations on Coriolis mass flow meters published in the Journal of Dynamic Sytems, Measurement and Control (March 2003), vibrations at the meter drive frequency will always cause meter errors and the magnitude of the error is linearly dependent on the intensity of the vibrations.

We realized a 250 HP blower produces strong vibrations and there are some harmonics around the drive frequency of the meter which is approx. 580 Hz.

Now we'll analyse this and see how the pipelines are affected by cold weather by making other measurements under cold temperature conditions.

Finally, we'll also test the plant's ground connections for faults... we have a few doubts about its integrity.

Thanks again

The problem needs careful evaluation and that means checking the basics and testing assumptions.

Having eliminated a problem with the flowmeters, it remains either a problem that both flow lines are not exactly identical or that the process conditions are not identical.

The most obvious next step is to log as much live data on both flow streams as is possible and to log events.

This means logging data from the meters such as mass flow rate, density, temperature and pressure; logging other data from other line instruments such as valve position, other process and environmental temperature and pressure measurements and to log events such as operator actions, pump switching etc.

This process may need to be extended upstream and downstream.

I have found that it is often only by graphing the data that a solution becomes apparent and it can often be surprisingly simple e.g. at low temperatures, contraction opening a gasket just enough for an air pocket to be drawn in via pump suction ...
If the systems are identical, and you have switched the meters around, check the upstream section of system for inconsistent operation. It sounds like the problem may be occurring during a cycle in processing from your statement below.