Hydraulic control valve scheme not cutting the mustard

P

Thread Starter

Phys Ed

* Hydraulic control valve system manipulating a 12" ball valve to regulate pump discharge and suction pressures on a gas pipeline.

* PLC output to the control valve system is -8 mA to +8 mA; where a negative signal closes the valve, a positive signal opens the valve, and 0 mA maintains the current position.

* Currently using Proportional only control (assuming bias is "built-in" to valve; it stays put at zero output).

* Problem 1: When opening or closing the valve and a change in valve direction is required, the mA signal must come all the way back to and cross over 0 mA to get the opposite action from the valve. Sloppy!

* Problem 2: At times, line conditions dictate that the discharge pressure cannot be maintained at setpoint. The control valve will open to 100%, but will have not have much effect on the pressure. When conditions change and the pressure rises again, not only does the mA signal have to decrease below 0 mA to begin closing the valve (see problem 1) the ball valve must travel from 100% to about 35% before any change in pressure is seen. Sloppy! Limiting the maximum open position of the valve is not an option because of the wear and tear it would cause on the ball.

Currently, intermediate setpoints are used to prevent the pressures from overshooting (discharge) or undershooting (suction) at startup.

Any suggestions would be greatly appreciated.
 
R

Richard Belknap

> * Problem 1: When opening or closing the valve and a change in valve direction is required, the mA signal must come all the way back to and cross over 0 mA to get the opposite action from the valve. Sloppy!

It sounds like the ma signal to the valve corresponds to the speed at which it opens and closes. This is typical of a hydraulic servo valve. I don't see why this is a problem.

> * Problem 2: At times, line conditions dictate that the discharge pressure cannot be maintained at setpoint. The control valve will open to 100%, but will have not have much effect on the pressure. When conditions change and the pressure rises again, not only does the mA signal have to decrease below 0 mA to begin closing the valve (see problem 1) the ball valve must travel from 100% to about 35% before any change in pressure is seen. Sloppy! Limiting the maximum open position of the valve is not an option because of the wear and tear it would cause on the ball.

I assume the valve opening is in response to more flow demand on the system discharge. Meaning as the flow demand increases the pressure drop across the valve increases causing lower system discharge pressure. Thus opening the valve decreases the pressure drop across it and therefore increases the system discharge pressure.

If this is true(I don't have a schematic, I'm guessing), then the system would work as long as the pump can support the flow at the required pressure. Hence, it sounds like you are operating beyond the flow capabilities of the pump. In other words, as the flow increases the differential pressure across the pump decreases(check pump curve).

If you open the valve all the way(then you have essentially an open pipe) and your pressure downstream is still too low then your pump must be too small.
 
> It sounds like the ma signal to the valve corresponds to the speed at which it opens and closes. This is typical of a hydraulic servo valve. I don't see why this is a problem.

Richard, first of all, thank you for replying. Maybe I used the wrong word, a typical hydraulic servo valve may not be a "problem", but this application is certainly a "dilemma" that I am faced with. This is my first encounter with such a beast and I am not quite sure how to attack it. Is there a better approach than using a Proportional Only controller? I could really use some help on this one.

> If you open the valve all the way(then you have essentially an open pipe) and your pressure downstream is still too low then your pump must be too small.

Richard, in my attempt at brevity I mislead you on the valve opening all the way. In this application, it is okay that the valve opens all the way and the pressure is not maintained. The "dilemma" is when conditions change and the valve has to travel from 100% open to 35% before making a difference on the rising pressure. Having to travel so far lends itself to the pressure overshooting.

Thanks again.
 
R

Richard Belknap

It sounds like you need D(Derivative) added to your P to move the valve quickly from 100% to 35%. The D will respond to a fast change on the input of the PD controller by giving a large pulse for a short time to the PD Output. Thus, allowing the valve to move faster. With small changes in the PD input, the D will have little or no effect on the PD Output.

Since you get very little change from 35% to 100%, is it possible to use a valve with a smaller Cv? This would give you more control over the output of the process.


Good Luck and hope this helps.
Richard
 
Richard,

> It sounds like you need D(Derivative) added to
> your P to move the valve quickly from 100% to
> 35%.

You may be right about adding derivative. Originally, I had the same thought, but was concerned about the small amount of lag time of the pressure. I've most often only used derivative on processes with a large lag time.

During development, I devised a scheme which would ease into the full derivative effect. I called it adaptive rate. So, things like a noisy signal or brief pressure change would have minimum derivative action. It worked great with the simulator.

When I got to startup, I had trouble tuning the loop with that scheme. It seemed like any rate value at all would cause the valve to start oscillating.

Stopping and re-starting the station pumps on this cross-country pipeline is not something the customer enjoys doing, as you could imagine. Because of that, after about 3 failed attempts with that scheme (which I don't think was a fair shot at it), they encouraged me to go to the Proportional-Only scheme, which I had used at another station.

> Since you get very little change from 35% to
> 100%, is it possible to use a valve with a
> smaller Cv?

I wish that was an option. Changing the valve size and putting an I/P on it... Man, life would be good!

One other thing. The PLC that I'm using has 2 choices for PID loops: Position and Velocity algorithms. Position is the one that most of us are familiar with. I've played with the velocity type before, but it never made any sense to me.

I tried experimenting with it again last night and I have to say it looks promising. I will do some more testing to be certain.

Have you or anyone else had any experience with the velocity type PID loop? Any thoughts or suggestions would be appreciated.

Thanks,
Phys Ed
 
G

Guy H. Looney

I couldn't find the original post, but I'll add my 2 cents based on what I see below. I've used PID motion controllers (Galil & Parker) w/ hydraulic servo valves in 2 totally different applications.

The first system was a large system (well over 20,000 pounds of force) & moved very slowly. This system was extremely easy to tune w/ proportional & derivative gains.

The second system was about 3,000 pounds of force, and had me at a loss for quite a while. I spent about 3 days trying to get a responsive system. The problem was the trade-off between overshooting & stability. I couldn't get both stability & responsiveness, despite using proportional, derivative, & integral terms. Finally I introduced feed forward into the system & limited the integral term. This made a HUGE difference & the system behaved nicely. It was a flying shear for cutting sheet metal. The cylinder had to ramp up to match the line speed of the metal, stay at the speed while the metal was cut, then ramp down to 0, and finally reverse to the "home" position & wait for the next cycle.

The basic premise of feed forward & integral limit is quite simple, but was necessary to achieve the performance this system required. Feed forward basically anticipates the upcoming accel / decel & injects voltage directly to the system instead of waiting for the PID loop to calculate the error signal. The integral limit is essential because during the move you're
building up error & at the end of the move the integral term wants to try to compensate. This attempt to compensate is the reason the system rings a lot at the end. No matter how high you try to raise the derivative gain, stability won't be reached. The derivative term only "kicks in" during motion. The stability problem I was facing was at steady state which is an integral term.

Hope this helps. Bottom line is that proportional only controllers can't always handle the application. Mine required PID as well as some extra bells and whistles. By the way, the product I used in the second example is from Galil and costs less than $1000...well worth the money if you think about capital costs versus opportunity costs.

Regards,
Guy

Guy H. Looney
Motion Control Engineer

A.C.E. Systems, LLC
170 Medearis Drive
Old Hickory, TN 37138
email: [email protected]
website: www.acesystemsllc.com
work: (615) 754-2378
fax: (425) 944-5017
cell: (615) 330-0044
 
L

List Management Account

---------- Forwarded message ----------
From: Phys Ed <[email protected]>
To: <[email protected]>
Subject: Re: PROC: Hydraulic control valve scheme not cutting the

Richard,

> It sounds like you need D(Derivative) added to
> your P to move the valve quickly from 100% to
> 35%.

You may be right about adding derivative. Originally, I had the same
thought, but was concerned about the small amount of lag time of the
pressure. I've most often only used derivative on processes with a large
lag time.

During development, I devised a scheme which would ease into the full
derivative effect. I called it adaptive rate. So, things like a noisy
signal or brief pressure change would have minimum derivative action. It
worked great with the simulator.

When I got to startup, I had trouble tuning the loop with that scheme. It
seemed like any rate value at all would cause the valve to start oscillating.

Stopping and re-starting the station pumps on this cross-country pipeline
is not something the customer enjoys doing, as you could imagine. Because
of that, after about 3 failed attempts with that scheme (which I don't think
was a fair shot at it), they encouraged me to go to the Proportional-Only
scheme, which I had used at another station.

> Since you get very little change from 35% to
> 100%, is it possible to use a valve with a
> smaller Cv?

I wish that was an option. Changing the valve size and putting an I/P on
it... Man, life would be good!

One other thing. The PLC that I'm using has 2 choices for PID loops:
Position and Velocity algorithms. Position is the one that most of us are
familiar with. I've played with the velocity type before, but it never
made any sense to me.

I tried experimenting with it again last night and I have to say it looks
promising. I will do some more testing to be certain.

Have you or anyone else had any experience with the velocity type PID
loop? Any thoughts or suggestions would be appreciated.

Thanks,
Phys Ed

*****************************************************************
YES! You can get digests, or limit messages by topic. See
http://www.control.com/control_com/alist/ for details.
Before posting, please read http://www.control.com/control_com/alist/faq_html.
The Automation List is managed by Control.com Inc.
 
Check out REXA at www.REXA.com. They make self contained electrohydraulic actuators that are being used very successfully on control valves in pipelines. The control and tuning are quite remarkable.
 
Top