Help - Difference Between Two Level Transmitters (DP diaphragm seal) 1

[PedrohsPrado]

Dear,

I am having problems with level measurement in a vessel.
In this vessel, we have 2 level transmitters (one for safety and one for control) and both are DP transmitter with diaphragm seal.

Those transmitters are connected in different process connection, being them:
- Control Level transmitter connected directly to the vessel.
- Safety Level transmitter connected in a StandPipe (Bridle)

The superior and inferior connections of both transmitter have the same elevation, in other words, they supposed to have the same level indication.

ENGINEERING DATA:
The specification of the transmitters are:
- Dp cell Transmitter + Diaphragm seal.
- Process connection 2" FLG 600# RJT + Flushing Ring
- Capillary fluid: Syltherm XLT (0,85 SG)
- Control Transmitter: Foundation Fieldbus (Rousemount 3051CD2F22A1AS2M5B4E2Q4 + 199DAB59 AFCWG4DAA9H)
- Safety Transmitter: Conventional 4-20mA (3051CD2A22A1AS2M5B4E2Q4 + 1199DAB59 AFCWG4DAA9H)

Our vessel (vapor-liquid separator vessel) receive a stream of Natural Gas (GAS IN) and operates with 3 phases:
- Gás (lighter Hydrocarbon components). - GAS OUT
- Liquid HC (Heavier Hydrocarbon components). - OIL OUT
- Glycol + Water - WATER OUT

Vessel operation conditions:
- Pressure: 39,5 kgf/cm2G
- Temperature: -28 Celsius Degrees (Below zero temperature).
- Gas Density: 43,31 kg/m3
- Liquid HC Density: 530,00 kg/m3
- Glycol+Water Density: 1074,2 kg/m3

Boths instruments are installed to measure the LIQUID HC phase (SG 0,53).
- Distance between Superior and Inferior Level Connections: 1000mm
Range: -850mmh20 to -320mmh20. (0% and 100% - without gas weight compensation)

The vessel and the standpipe have insulation.

MEASUREMENT PROBLEM:

Although the transmitters are equal and have the same elevation in their process connection, they present different percentage level indication:

The instrument that is connected to the standpipe (Control - Green Line ) always show its indication below the instrument that is connected directly to the vessel (Safety - Blue line ).

MY HYPOTHESIS:

1) Calibration mistake

Once they have a different system (Foundation Fieldbus and 4-20mA) maybe occurred any mistake in calibration process (range selection and parametrization).
I saw the configuration and both have the same calibrated range (-850mmH20 to -320mmH20).

Any of you can see any calibration mistake can cause this kind of error between this transmitters?
Can some AI block parameter cause this difference?


2) Measurement Difference Cause by Process (StandPipe x Vessel)

After I went to the field I started to think that this error could be caused by a Convection Flow between the standpipe and the vessel.

If the standpipe has a higher temperature, can it cause the difference between the transmitters?

As they are DP transmitters, I think just density variation (static analysis) does not explain the difference between them, once they are calibrated with same range.

Is the vessel work in a thermodynamic equilibrium?

The Standpipe has a Magnetic indicator (304L body - not insulated) and seemed to be at a hotter temperature than the vessel. I used a pyrometer and the temperature of the Magnetic Indicator surface was 20 Celsius Degrees (positive and have way more temperature than -28 degrees of the vessel).

Once the magnetic indicator is not insulated, Could it induce error caused by excessive heat transfer? Can it make the temperature in standpipe hotter than the vessel?


Can the evaporation in the standpipe be causing a flow due to the convection phenomenon?


If yes, I think it causes a different pressure in both transmitters. See figure below.

Thank you very much for your attention.

If you need any more information, please ph.prado@gmail.com or SKYPE: pedrohenriquesousaprado
I have some prints of the configuration.


Best Regards,
Pedro H. S. Prado

 

[danw2]

Welcome to Eng-Tips. You posted a very nicely marked up photo and diagram. That's quite unusual for a first time poster (minimal info is more the usual). Nice work.

The configured range looks good; with a difference of 530mm H2O from LRV to URV. 530mm/0.53 = 1000mm HC.

If the HC boils at whatever the temperature the standpipe is at, then yeah, you have flow as you indicate as the tank refills the standpipe as HC boils off. There might be convection flow between the standpipe and magnetic level bridle

My guess is that a major component of the difference in level is that every 'closed-system' pressure instrument is also a thermometer (Charles or Boyles Law) and that the differences in level can be the difference in temperatures of the high or low side capillary impulse tubing. If the capillaries are isothermal, any temperature effect should null out between high and low side, but temperature difference creates a pressure error.

There are no units on the trend chart. I assume it is mm, so the third point shows a difference in level of approximately 10mm?

A 10mm error is easily achievable with difference in capillary temperatures. Buried in the vendor's specifications for any capillary remote seal transmitters is a note that accuracy if x.xx% at reference conditions. It is unlikely that both transmitters' capillaries are maintained at reference conditions or ever at equal conditions.

It is not clear which trend is for which device
Your statement >standpipe (Control – Blue line) always <snip> below <snip> the vessel (Safety - Green Line).
does not match the graph because the graph's blue trend is always a higher level (not lower) than the green trend.

 

[PedrohsPrado]

Danw2, thank you very much for your attention.

The graph unit is percentage.
The difference betwen the transmitters vary from 5% to 10% (most in 8% - 80mm in level as we have 1000mm between process connections).

In the graph, the yellow trend is the percentage difference betweem them.

As our meters are balanced (same capyllar lenght and diameter), I think the most significant temperature error caused by remote seal is the "Head temperature effect" - "caused by changes in the weight of the capillary fill fluid due to density variation with varying ambient temperatures exerting backpressure on the sensing diaphragm."

But Ive calculated it with Emerson Process software (Instrument Toolkit) and it would be 1%.
So Iam not sure if it is causing the difference betwen the transmitters.

Observation: You were right. I made a mistake with the trend colors. I corrected it.

Best Regards,
Pedro Henrique Sousa Prado

 

[danw2]

Out of curiosity, what temperature difference value produced a result of only 1% error in the software?

 

[PedrohsPrado]

Hello Danw2,

I used the following data to calculate the head temperature error:

See attached the error calculation from Rosemount Instrument Toolkit.

Best Regards,
Pedro Henrique Sousa Prado

http://files.engineering.com/getfil...47d3-8709-1c2506441494&file=ToolkitReport.pdf

 

[georgeverghese]

Some critical info is missing
a)Pls describe or add a sketch of the internals in this vessel which show the overall liquid level control and interface level control.
b) Are these LTs' in total liquid level detection service or interface level detection ?

Though this may not be relevant, we see a tubing line in the photo coming down to the bottom of the photo - this tubing line is pocketed - is this for one of these dp cells ?

If these 2 dp cells are in interface level detection service, I can see a design error - there should be an equalising line from the vessel's hydrocarbon phase to the standpipe to make the interface level in the standpipe the same as in the vessel.

 

[PedrohsPrado]

Georgeverghese,

Both instruments work for total level (level of liquid hydrocarbon).
They are installed in a position that is expected to have condensed hydrocarbon. So they are not designed to measure the interface between liquid HC and Glicol+water.

The tubing line that you mentioned is the drain line of the magnetic level indicator. During normal operation, It is closed.

See the drain with details.

Be aware that the photos that I've posted here were taken before they install the Diaphragm Remote Seal Transmitters. This is why I drew the transmitters in those pics.

Thank you very much for your reply.

Regards,
Pedro Henrique Sousa Prado

 

[georgeverghese]

My earlier suggestion that there should be an equalising line if these LTs' are in interface detection service is not correct in this case since these are dp cells and not external chamber float type LTs'.

This readout anomaly is strange indeed. Could it be that the level standpipe ( which leads to the control LT) has a vapor space vessel connection located downstream of the vessel vapor demister, while that for the safety LT (which is directly mounted on the vessel body) has a vapor tapoff upstream of the demister?

By the way, the tubing line mentioned earlier is the one near to where you have superimposed the vessel direct mount LT. This is the tubing line which you can see is pocketed as it comes down to grade.

 

[danw2]

Your calc shows no temp difference between high side and low side.

Keep the high side values the same, try re-calculating with the ambient low side values at a 3 degree difference from the high side:
Last zero: 30.00 (alternatively 33.00)
Minimum: 18.00
Maximum: 38.00
And see what the error is.

 

[djs]

Are the diaphragm seals mounted directly to the vessel and the stand pipe? If not you might try draining/priming the impulse lines. Also are the diaphragm seals exactly the same? If not, because of the low differential pressure a difference in mechanical resistance will cause a difference in pressure between the two systems. Because of this for very low pressures I try to avoid using diaphragm seals.

 

[roydm]

Updated Dec 4th PM
Looking at the first diagram it shows a low point in the standpipe where water and glycol would collect at the low point skewing the level reading in the sight glass making it read lower than the level in the vessel due to a slug of high specific gravity water/glycol. This should be easy enough to prove by draining some fluid out of the low point.
I wonder why this piping arrangement, normally the bridle connections are horizontal with no low spots?
Where is the interface between the Hydrocarbon and the water/glycol, if it's too high of course water/glycol will flow into bridle

However in the photograph it shows a line off the bottom of the standpipe, I assume this drains any water collecting in the bottom of the bridle but again if there is any significant flow in this line it will cause the level to indicate low. I call this the coffee pot effect, I'm sure you have noticed how the level gauge drops to zero as you fill your cup (only half in jest).
This could be proven by briefly stopping the flow.

Why are you using filled systems, it would seem that just regular DP cell would do?
There is a significant temperature difference in the diaphragm seals, that could also account for a zero shift especially in the SIS transmitter.

We have tried to use filled systems for measuring interface with limited results, similar quality instruments effected greatly by ambient temperature changes.

Do the cells read identical when the system is shut down, zero flow in and out?

I don't think the evaporation rate in the bridle would cause any significant flow but what is the pressure inside the vessel?

Again, great information provided.

 

[georgeverghese]

If the LTs' are located one on either side of the vapor space demister, that would explain this difference in readings. The 2 LTs' would then be reading correctly, with the level higher downstream of the demister than the level upstream. This is caused by the difference in vapor space pressure across the demister. The times when the 2 readings are nearly the same would then be when vapor throughput in this separator is low and demister dp is also low.

 

[JJPD2]

Pedro,

Are these transmitters having exactly the same heights on its taps?

I mean the high and low taps are at the same reference on the vessel.....

Your error or difference between the transmitter outputs is almost constant, so I believe that issue maybe is not created by the process itself

Remember that you are using transmitters with capillaries, so that is a small tubing perfectly filled with a fill fluid, therefore any diference of heights or how the capillaries are fixed on its travel from the transmitter to the taps will have effect over the output (mA)

(-850mmH20 to -320mmH20)<------- This range is on a perfect world, In a capillaries world I suggest the following:

* Close both root valves
* Carefully, using the drains put both (low and high) seals to atmosphere pressure
* Look how much pressure is showing the transmitters
* Compare both reading (Xmtr A and Xmtr B) and check if the error is similar to the showed on the DCS
* Recalculate your range based on the pressure showed by the transmitter when was opened to the atmosphere, I guess that you will see some value near to -850 mm H20 but not exactly...maybe -854.3, -853.15.....

By the way as mentioned the capillaries will be affected by temperature changes.....

Regards

 

[roydm]

Discounting the effect of temperature look at the Bridle connection, a pocket that is a natural trap for water/Glycol mix which is much heavier than the oil.
A) The oil in the vessel applies it's head to the water in the bridle but being heavier it only pushes it half the distance up the standpipe.
This IMHO is poor piping design. the pipe between vessel and bridle should be horizontal all the way with no pockets.
You should be able to guarantee what is in the bridle.

B) As I pointed out earlier there is a pipe at the bottom of the bridle, we don't know what that is doing, any flow in the bottom connection will cause a significant level difference in the bridle.
Would the original poster explain that line?

Is it possible to drain the pocket by opening whatever valve is in the line, this would prove A)
If there is a steady flow through the small line close it for a minute or two proving B)