Erratic Flow Control


Thread Starter

Bill Dolan

Problem Statement: Erratic Flow Control In Auto or Manual Control

A transfer line exists between two pressurized tanks and tanks are the same size and are at the same elevation. A simple single loop flow controller sits on the transfer line to move a water based fluid from one tank at a higher pressure to the other. The flow meter is a T-series Micromotion and the valve is a Samson sanitary angle globe. Flow setpoint is 100 Liters/Hour. The valve is in a very controllable range of approx. 45%.

Flow control is excellent until the upstream tank is topped off, which then results in a higher static pressure plus overlay pressure relative to the valve inlet pressure. Both tanks are at the same elevation, thus the need for the overlay on the upstream tank to transfer to the receiving tank. When upstream tank is 1/4 full the flow control is good, but once the upstream tank is topped off, then flow control becomes erratic even with the flow controller in manual mode.

There is a local PI at the flow meter and control valve inlet and the pressure does not change coincident with the erratic change in flow. The downstream tank vessel pressure is rock solid and so what is causing the flow to be erratic? The only process difference is the increased upstream static pressure with a full tank. Filling the upstream tank changes the valve dp from approx. 8 psid to 12 psid. Fluid temp is a constant 25 degc.

Any helpful hints would be appreciated.
You mention no pump.

So I assume that your pressure 'overlay' is the pressure in the upstream tank that provides the force to drive the fluid through the flow meter, valve and into the downstream tank against the head pressure of the downstream tank liquid level and its overlay pressure?

Is this pressure overlay in the upstream tank held constant, regardless of the level in the tank?

Or does filling tank increase the pressure because the overlay volume decreases with increased level?

These are pressurized tanks. Each must have a pressure relief valve. Is either relief valve popping, releasing built-up pressure, and causing the erratic flow changes you're seeing as the controller's PV upsets so quickly do to pressure relief? That's the only thing I can think of that would cause erratic flow in manual mode.

The controller's tuning constants could be too aggressive, which could cause cyclical flow as the DP changes overtime, but in manual mode, you should have a continuously, slowly decreasing flow for a fixed valve position as the upstream tank loses pressure as it drains and the downstream tank gains pressure as it fills. That's not erratic, that's flow responding to the available DP with a fixed valve position.
Down stream tank vents with a modulating control valve and pressure is held constant at 2.5 psig. No issue there. It's understood that valve dp will decrease as upstream tank level drops with constant overlay of 10 psig. Therefore, control valve will open more over time to maintain the same setpoint. With controller in man mode or auto mode flow will suddenly drop to 88 lph from 100 lph and then recover. Micromotion has taken data from the core processor and has determined this drop is real. It's rather periodic in that it happens every 3-5 minutes. There are no detectable process disturbances anywhere. It's too simple to not work, but obviously I'm missing something.

If the erratic upset (spike to down 88 lph) occurs with the controller in manual mode, AND the valve position holds during and after the upset event, then the pressure perturbation must be elsewhere in the system, it is not being initiated by the flow controller's output.

What else could cause pressure upsets?

If the flow valve's actuator is air-to-open, and there's a momentary, periodic air supply 'event' that significantly drops supply air pressure to the actuator, the valve could momentarily close, lowering the flow.

Positioners can malfunction. I'm stuck for the moment on a known periodic, repeatable malfunction, but there's probably one in the loop. Smart or analog? known good or scavenged?

What controls the downstream tank pressure modulating valve to hold that overlay pressure at 2.5 psig? Another PID controller?

100 lph is approximately 1.6 lpm, not a huge volume of air to displace in the filled downstream tank. More along the lines of what I would think of as 'bleeding' than 'flowing'.
Is that downstream tank valve have a sufficiently small enough Cv to continuously bleed off that volume of air?

What controls the upstream tank pressure at 10psig? A diaphragm (proportional) pressure regulator? What relieves the pressure when the upstream tank is re-filled ?

There's numerous points of pressure control here, not just the control valve to regulate the flow, and something is changing . . .
Hi there,

You need more information to find out what the cause is of this problem. With a intermittent problem like this you need as much long term data as possible to see what is happening when. Basically what you want to do is monitor the level and pressures of both tanks as well as the output to the control valve as well as the flow meter reading. If you draw trends and study them for a period of 48 hours I am sure you will begin to see what is constantly happening when. For instance you might find that when the level in the upstream tank is at a certain level the problem starts to occur but only if the pressure is so much in the downstream tank or anything else that forms a constant patten. Once you have a suspect, you need to simulate the problem manually to confirm that is definitely the problem. It might not be possible to simulate the problem so the next step is to start eliminating the most likely suspects first and work you way down. You might also find that nothing is happening at exactly the same conditions and that it seems like a random problem. Then you need to move on to dealing with it as random. So you need to start thinking outside the box.

To me it seems you have a problem on the control valve. The first thing I would look at is the control position of the cv at various points and also at the point when your problem starts. I am sure you know as well as I do that you do not want to control in the high or low end of the stroke. Ideally your control should always be between 25 to 75% but I am sure you know this and have made sure this is not the problem. Next I would suspect the positioner. Replace it with a new one as a test since sometimes this can be your problem but it might look like something else.

If you have the controller on manual and the problem is still occurring the problem is at the control valve and not the PID Cont or flow meter. Especially if you have already confirmed that this is a true flow meter reading.

This confirmation of the flow meter true reading helps you to move onto another possibility.

Have you tried to open up the control valve and check if all is ok with the plug and seat. It can also be that the plug have come loose from the stem or a piece of metal or plastic is stuck inside the trim and block the flow from time to time, so this is something you need to confirm. Also blow the transfer pipe clean once you have removed the valve for inspection since the intermittent blockage might be in the transfer line somewhere. Is it possible to confirm the upstream tank bottom is clean and no sludge or loose pieces of whatever could block the transfer line inlet.

Once you have checked all these things let us know we will see if we can think of anything else you can try.
Thanks for the response, but I have checked all the above twice over. It's a biotech application, so all equipment is CIP'd and SIP'd before use. I have two identical systems and they both behave or rather, misbehave identically. Given the low flow and a slightly viscous fluid, I'm wondering if the Reynolds number of the fluid is running between laminar and transitional flow through the vena contractra area of the valve seat. Dropping into a laminar range could account for the sudden drop in flow for a fixed valve output. I can't change flow rate now since the system is in production to further test this theory.
Is it possible to maintain level in the upstream vessel rather than to refill it each time in batch style. If you keep inlet level head uniform you will get a very good flow control in the end. I think that whatever device you have for pressure control in upstream vessel is not fast enough to cope with changing levels due to liquid discharge and additional effect from the vapor pressure inside vessel
Pressure control for upstream tank is very fast, but then I'm only drawing off 100 LPH from an 8000 Liter tank. Pressure is very stable and is not the source of the disturbance.
Just a little followup: I did finally identify the source of the disturbance and it is attributable to valve positioner tuning. The valve positioner, when initialized for the first time, sets the tuning based on the flowing fluid. This was performed in air and the tuning selected was much too aggressive for liquids. The positioner gain was reduced and the flow control is now perfect. Thanks to everyone that weighed in on this issue.

Ray R. Powell

What type of valve positioner can be tuned? It seems to me that the gain/proportional band was too high/narrow and had to be adjusted in the controller. Remember that on a fast acting loop no derivative can be used, such as flows and pressures. Also sometimes no reset is required.
This is a Foundation Fieldbus positioner, which has PI tuning for the position feedback loop as do others. I did not need to touch the controller tuning. This is really a cascade, flow onto position. The position or inner loop was too aggressive.
Yes thanks for the feedback, I wish more people will send us the conclusion or final end result. I had a similar problem once on a very erratic pressure control and eventually I started playing with the positioner tuning sets and the control become much better.

Ray, for your info the Fieldview DVC series is one of the positioners that have preset PID tuning sets and you can change from a fast to a slower tuning sets with the HART. Very easy to do and if I remember correctly they go from set D to M. Cannot remember which is the slow and fast side but it is a matter of try one side and see if the problem gets better or worse.
> Bill, Thanks for the update. It's a rare event to get an update or a conclusion.

It's nice to know that people are interested in hearing the feedback.