PID Control of Non Linear Ball Valve

[drbitboy]

...
Here in the UK we are no better, 50 years after decimalisation the mechanical industry still use imperial especially in pipe sizes.

Great stories, but if I had done that I would have said it's on me, if I never checked what a "gallon" of water would weigh with that calculator - "a pint's a pound the world around."


"No better?" Units do not matter; communication matters irrespective of the units chosen. And it was a Mars mission, not Hubble, where the communication failure over units occurred.



Sorry, nothing personal, But this pushed a button as it is one of my rant-topics: unit conversion involves the second-most fundamental, and easiest operation of technical work i.e. multiplying by one, followed by the most fundamental and important operations i.e. thinking about the result. I'm not saying anyone is a bad person for making a mistake: those things happen and I've done more than my share. If anything, parky showed superb insight and skill there in challenging an assumption (it's an imperial gallon) and moving on to the right answer. But I disagree that it can be blamed on the unit system used.

 

[MorphuisOGrady]

what is the required accuracy? How about a strapping table to define the output and then an add/sub function to give fine control. You could feed that into the PID in man mode and keep the faceplates

10% would be sufficient. Strapping table sounds interesting. I think there is an AOI in the PlantPAx library for a Strapping table application. Might be worth a shot.

 

[Tom Jenkins]

... or just using a simple control where the motor is turned a little open or close to get the desired flow.

Now I will make an assumption about the motor. The motor only opens or closes at one speed. In which case you can bump the motor open or closed until the desired flow is meet. If the motor moves to quickly then sometime between bumps will be necessary to see the effect.

That is exactly the way I have done it hundreds of times. If you have a flow meter you don't need the valve position. Slow travel time on the valve also helps get precise control. 60 seconds per 90° works well for me.

A couple of notes. Air is harder than liquids to control, and ball valves, like butterfly valves, are more non-linear than spools. In my opinion, the savings from omitting a small diameter flow meter won't offset the aggravation and angst of trying to get by without feedback. Going cheap, like taking shortcuts, usually bites you in the butt in the end.

 

[kamenges]

Originally posted by drbitboy:

...followed by the most fundamental and important operations i.e. thinking about the result.

With the increasing availability and proliferation of easy-to-access engineering tools this is going to become a lost "art". You need to have some amount of feel about whether the number you are given makes sense in the context of the question or you really risk coming up with the wrong answer.

"Why did you use that number??"
"Because my computer/smartphone/web app/Magic 8-ball said that is the number I should use".

I get this already today. It is only going to get worse with time.

Keith

 

[Peter Nachtwey]

When I use Mathcad I don't worry too much about the units because Mathcad will keep track of them for me. When I program in python, I convert everything to basic MKS units were possible because mixing and matching units is error prone. After doing the calculations, I then scale the results back into the units the user wants.

This thread has been a good example of what I was complaining about in the "responses" thread. Too many people ask for help without putting much effort into even understanding the problem.
--------------------------------
If I was MorphiusOGrady's engineering manager, the first question I would have asked is what the goal is and why. What is the return on investment? Then we can get into details.

When I was in the navy I learned to have the answers to the questions the senior officers would ask. They made it clear that if they don't have all the information necessary for making a decision then I was wasting their time and I was to find out the answers. It was best to have the answers. Sometimes it would good to have a recommendation. I think it was their way of checking if I would make the same decision as them.
--------------------------------------------
Why is the accuracy of the flow meter so bad? It seems to me that would be the place to start. If the flow meter is inaccurate due to humidity and temperature changes then why would using your own pressure sensors be any different? There was no mention of compensating for temperature and humidity. Usually these numbers are relatively constant if in doors.

 

[drbitboy]

I'm with Peter: SMH; there is so much variability in the answers that I wonder if any two of the respondents are considering the same process. Or mebbe it's just me being t'ick in da 'ead.



I assume

• a PLC program going to maintain a stable differential pressure at a fixed value, or within a fixed range.
• a PLC program is somehow going to change a valve position to maintain that differential pressure
• that valve position is controlled by a motor
• changing that valve position will change the differential pressure across the valve,
  • That change will be reverse-acting i.e. increasing position (opening valve) decreases differential pressure

Is the PLC going to control the valve using discrete outputs to the motor, either

• running the motor in direction "+" to increase the valve position and decrease the differential pressure,
• or running the motor in the opposite direction "-" to decrease the valve position and increase the differential pressure,
• or doing nothing if the differential pressure is already in the target range?

OR


Will the PLC send an analog output to the valve "system," and the valve system will run the motor as needed to set the valve to the target position corresponding to that analog output*, with no further interaction from the PLC?


* e.g. 0/5/10V analog output => 0/50/100% open valve, or 4/12/20ma analog output => 0/50/100% open valve, etc.




There are many other questions, like, if this process is "traditionally controlled on flow," how does controlling on pressure differential across a valve provide a similar overall objective? Perhaps differential pressure has always been the target and traditionally a second controller compares measured and setpoint differential pressures and then sends a flow setpoint to the valve/flow controller, and the OP wants to eliminate the cascading setup and control pressure directly with valve position?



Also, the non-linearity of the valve is possibly a red herring, as it is the linearity of the whole system response (change in output value vs. controlled value) that matters.

 

[drbitboy]

What is the return on investment?

This. This right here. This is engineering (even more important than multiplying by one).


To the OP: what is the return on investment of controlling the pressure differential across this valve?


Because, although I am not saying there isn't any, I suspect that controlling some other process variable will have a more direct and more easily explained relationship to ROI. It may work its way back to the dP across this valve, fine, but I am not at all sure dP is worthwhile pursuit.


My brother worked at a plant in an industry with very low margins, and the company would have cash flow issues from time to time. Whenever that happened they would declare that they were in contingency mode and all capital projects had to be more strictly monitored and selected based on ROI. Whenever they did that, he would say "Does that mean when we're not in contingency mode, we can fund booze and parties? If I cannot justify on an ROI basis what I am doing every hour of the workday then you should fire me." And so he would ignore the contingency measures.