Control valve good design practice 2

[98769876]

When installing new filters (cartridge) in an existing process with a pump and control valve do you place the filters before or after the control valve. You need to be able to measure the dP across the filters but the problem with the control valve is that it will change your dP measurement.

 

[TD2K]

Why does the control valve change the dP? You say this system has a pump so I assume we are dealing with a liquid. The dP across the filter is a function of the flow and that's the same upstream and downstream of the control valve, assuming you don't have flashing occuring.

I would normally likely put the filters before the control valve because of the disturbances caused by the control valve but if you locate the filters a ways downstream, it's not really an issue either location. In either location, I would design them for the maximum source pressure.

 

[98769876]

The control valve controls the level in the pump tank where we pump from. At times the level drops and therefore the control valve cuts back - changing the feed pressure to the filter. - Yes you will probably say that the exit filter pressure will then alos drop and the dP should therefore remain constant. But remember the Control valve now reduced the flow because it cut back the pressure drop across the filter is now less ( compared to initial) therefore dP is not áccurate since you don;t have a consant flow through the filter.
The filter pressure drop is calculated by delP= KQ*visc. (K constant - for filter design at certain flow) Q flow changes therefore delP across filter changes.

Therefore the position in the control vlv will affect the flow which will affect the dP.

 

[DRWeig]

You almost have your answer there in your 2nd post, 98769876. Unless you are also measuring flow rate, you can't compare the current pressure drop of the filter at the current (unknown) flow rate to the pressure drop of a new, clean filter at that (unknown) flow rate.

If you do know the flow rate, you can calculate what the pressure drop would be at the design flow rate and compare it to your standard for time-to-change-it pressure drop. But pressure drop across the filter is not linear. The pressure drop will rise in proportion to the square of the flow. At 2X flow rate, differential pressure will be 4X. At 1/2 flow rate, differential pressure will be 1/4. I'm assuming here that your liquid is incompressible and there is no flashing at the control valve.

If you don't know the flow rate, it really will be hard to do. I don't think you can assume that the flow rate even at wide-open valve will not vary, but if it is constant you can measure your pressure drop across the filter (new) at wide-open valve, then periodically during tank fills at wide-open valve. Or, if you know the valve's characteristics it might be possible to estimate flow rate by observing the differential pressure across the control valve. Best would be a reasonably-accurate flow measuring device.

As for position of filter relative to control valve, TD2K covered that nicely.

Good on ya,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[TD2K]

"Therefore the position in the control vlv will affect the flow which will affect the dP."

Wrong. Draw a sketch of your system and the pressure drop across each component. For a single phase liquid system you'll see that the location of the filter does not affect the dP (the absolute inlet and outlet pressure will change but not the dP). For a gas system, the pressure drop would be greater downstream of the control valve because of the decrease in system density due to the control valve's dP. That could be significant or could be minor depending on the pressure drop across the control valve relative to the total system pressure.

 

[DRWeig]

Or are you using the word "position" for two different purposes in your post?

It might be more clear if we refer to the "location" of the valve (no real consequence) and the "percent open" or "percent stroke" of the valve actuator (changes flow, and, of course DP).

Good on ya,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[GHartmann]

Maybe lets try it this way:

The flow through the filter varies with the flow to your tank as set by the inlet level control valve.

Therefore you always have varying flow and reulting varying dP through the filter regardless its upstream or downstream of your valve.

You just need to decide if its better to have the filter on the higher or lower side of the control valve for the particular fluid you are pumping.

 

[98769876]

My point that I was trying to make was:

For 1st hour pump tank is 60 % level. THis is = level SP and the control valve remains at 30 % since SP requirement is achieved.
Flow is measured at 0.8 l/s (we have a measurement but it is very erratic because of process).
dP across the filter is for instance 10 kPa(trends).

Next Hour:l
Pump tank level increases. Control valve stroke now goes to 90 % Output to control the level. Flow on the flowmeter goes to 1.4 l/s and the dP across the filter which is after the control valve now reads 50 kPa.
the process then reaches steady state and valve remains at that position.

Next hour:
pump tank level drops.
Control valve now tries to control the level and valve stroke goes to 10 %. Flow drops to 0.6 l/s to the filter and filter dP drops again. (this is what is observed on our online trends).
Process now reaches steady state and control valve is at 10 % - untill next disturbance.

we only change out the filters every 10 hours -sll these changes in the flow affects the dP across the filter. We want to use the dP to see when to change filter cartridge but because the flow is not constant - cant see dP increasing like we expect. dP measurement is not helpfull.
System layout/line-up:
We have the pump tank then the pump then the control valve , the flowmeter then the filters across which we measure dP.

 

[GHartmann]

Unfortunately you have to try and correlate the average filter dP versus the average flowrate and that appears to be problematic with your system flow variations.

Perhaps you can correlate trends of flow vs dP or % valve open vs dP. Establish a "normal" dP and then check to see how you are doing vs. the normal.

One thing I would recommend is to move the flowmeter upstream of the control valve which is the more common installation practice.

Is there any way you can have a more steady state operation for the tank level by having a more steady flow of fluid into the tank?

Good luck.

 

[SeanB]

You could build a calculation block. From your trends you know the clean filter pressure drop at a given flow rate. Using your measured flow rate you can calculate the theoretical clean pressure drop. This can then be compared against your measured pressure drop. When these start to drift apart you get a deviation alarm. Then it is time to change your filter.

 

[TD2K]

Along the lines of what Sean suggested, build a constant flow dP estimate. You have the actual flow and actual filter dP, correct your actual dP at any time (dP is going to be approximately proportional to the flow squared) to a 'design' or 'baseline' flow through your filter.