Process Application Help

Hey there Guys. Need a bit of help here. I have an application where I have water running into a commom header and then the piping splits into two pipes. Each pipe has a magmeter and a modulating control valve to regulate the flow based on the feedback from the magmeter. My task is to balance the flow between the pipes to the valve that is most open. I know it sounds strange but thats how the requirement calls for. By the way, I'm using a SLC 5/05 PLC to run all this. I know I'm gonna have to use a PID to run the valving with feedback from the magmeter. I'm just having a brain **** when it comes to the balancing act. Anyone run across this before? Anyone with some sample logic I can look at just to jump start my brain?

Thanks

Hmmmmmmmm

We need a little more information here. Is the flow input to the header variable or constant? Does the flow out of the two pipes have to be variable and balanced or just balanced? What determines the total throughput of the system?

Dave

The requirements won't work.

Both valves will end up increasing output just to keep up with the other valve until both valves are wide open. You need to control both valves to a set point. This means you need to use two PIDs.

IMHO you cannot do it.

You must first isolate one pipe from the other with a set of pressure regulators.

Then the measured flow from one pipe become the setpoint of the other valve/pipe system.

It will have fluctuation and dead time but if you keep the 2 pipes connected it will just not work.

One cannot control speed AND position at the same time. Einstein said it... Frank of course!

What's this?

Pierre said:

IMHO you cannot do it.

Click to expand...

Eric's requirements can not be done for reasons I stated above. One can use two PIDs tracking a common set point.

Pierre said:

You must first isolate one pipe from the other with a set of pressure regulators.

Click to expand...

I don't need pressure regulators to do this and the two pipes and valves do not need to be isolated.

Pierre said:

Then the measured flow from one pipe become the setpoint of the other valve/pipe system.

Click to expand...

That will work, but not as well as as having both track the same setpoint.

Pierre said:

It will have fluctuation and dead time but if you keep the 2 pipes connected it will just not work.

Click to expand...

There is no problem suppling two different pipes and valves from one header. Yes, the slave will definitely lag the master. Dead time would be only the time the PID takes to process each update.

Pierre said:

One cannot control speed AND position at the same time. Einstein said it... Frank of course!

Click to expand...

I control speed AND position of hydraulic actuators at the same time, ALL THE TIME! That is what I do!

What do Einstein and Frank have to do with this?

Eric,
I dont know if this will help relieve your brain gas or not. I have a setup that is just the opposite of yours. Maybe it will jump start an idea for your application. We have 2 seperate water pipes feeding into a common tank header pipe. One feed pipe is "soft" water the other feed pipe is R.O. water. We mix the two to obtain a desired alkalinity.
There are two flow meters and two I/P controlled air actuated butterfly type control valves. There is a small PanelView device where the operator can select the desired flow from each source. Example: 35 GPM from source "R.O." & 15 GPM from source "Soft". There are two PID rungs. The setpoints for each PID equal the number the operator enters on the PanelView. When the blending begins the valves will modulate as needed to achieve the desired mix. This setup isn't perfect, it works well enough for the application we have. I can see a potential problem if the operator enters a number higher than the supply pipe can deliver, or the total of both setpoints is more than the header can flow.

There is a word value in each PID that represents the amount of output(%) to the slave device, (OP) I think it has read only status.
Maybe you could use this word as the setpoint for the second PID (after massaging it as needed).
Example:Set point of Valve 1 = 50 GPM. Valve 1 = 50% OP. Setpoint of Valve 2 = (50 - 10).
Hope this helps
BD

Take a look at this. Just turn flow around. control valve is on
pipe that hogs flow.

http://www.jashaw.com/pid/ratio.html

Mickey

God, I mean Peter, is right... use 2 PID's, first one running off a setpoint, the second one having a setpoint derived from the first one's position.

If the one with the highest flow is apt to change, there will be some complications but they are workable.

That isn't what I said.

The Water Boy said:

God, I mean Peter, is right... use 2 PID's, first one running off a setpoint, the second one having a setpoint derived from the first one's position.

If the one with the highest flow is apt to change, there will be some complications but they are workable.

Click to expand...

Pierre mentioned the master/slave approach. The problem with the master slave approach is that the slave lags the master as Pierre pointed out. I would have both PIDs control to the same set point. This reduces errors to due lag an interaction between the two PIDs.

THIS IS SIMPLE, ONE FLOW SET POINT FOR TWO PIDS. IF BOTH VALVE PIDS ARE TUNED UP THE SAME WAY, THEY WILL TRACK EACH OTHER DUE TO HAVING SIMILAR LAGS AND ERRORS BETWEEN EACH PID AND THE COMMON FLOW SET POINT. NOW IF ONE USES FEED FORWARDS THE ERRORS AND LAGS CAN BE ALMOST ELIMINATED!

NEVER USE A MASTER/SLAVE PID SYSTEM IF YOU CAN GENERATE THE SET POINTS FOR BOTH PIDs FROM THE SAME COMPUTER/PLC. As Pierre pointed out, there is a problem with the lag between the master and slave PIDs. Can anyone name another?

This is an extra credit, precision pickle type of question. There is a hint above.

Eric says...

"I have an application where I have water running into a commom header and then the piping splits into two pipes. Each pipe has a magmeter and a modulating control valve to regulate the flow based on the feedback from the magmeter. My task is to balance the flow between the pipes to the valve that is most open."

Isn't that kinda self-contradicting? When you say "balance" do you really mean "equalize" the flow through each of the pipes?

If so, it sounds like you want to provide... let's say... 50 GPM through Pipe-A and 50 GPM through Pipe-B.

Then you suggest that the valves in each pipe might be open to different degrees. Pipe-A might be open 30% while the valve in Pipe-B might be open 60%. The valve in Pipe-B is "most open".

You say you want to balance the flow "to the valve that is most open".

That doesn't make sense.

Then you say "between the pipes". That suggests that the two pipes are connected somewhere beyond the manifold. Should that really say... "through each pipe"?

Ultimately, the flow through each pipe will be dependent upon the down-line head-pressure.

The valve in Pipe-A could be open 100% and yet provide no flow at all if the down-line head pressure (in Pipe-A) is equal to the supply pressure at the manifold.

So, Eric... could you clarify this a bit?

This is done all the time - by yours truly among others. However, Eric neglected to provide a few key bits of information.

Case 1) Is the water "running" into the common header from a stand pipe, city water supply, or other more or less similar source? In other words, is it a source with very high flow rate capacity and with a fairly slow changing pressure in the common header? If this is the case, then your problem is fairly simple. Any good analog control loop, PID or other, will be able to modulate the valves to maintain set flow based on feedback from the magmeter. Setpoint is desired flow, feedback is magmeter signal, and output is valve position.

Case 2) If you are pumping into the header, you need to add flow control of the pump. The most common way of doing this is with a pressure control loop, with desired header pressure the setpint, a pressure transmitter for feedback, and pump speed or discharge valve position as the output. This effectively makes case 2 become like case 1 as far as the two controlled flow valves are concerned.

Both cases are contingent, of course, on having a decently sized control valve and a valve operator that is sufficiently responsive and not hunting.

Case 3) You have a constant rate flow source (like a positivie displaement pump) and simply want to maintain a constant ratio between the two flows. In that case your open valve sets the flow rate, and the setpoint to the other flow control loop is a multiple of that flow. You don't want to manipulate both valves here, because it just creates more back pressure without changing the total flow.

Now, to me Most Open Valve (MOV) implies one of several algorithms used to minimize system pressure by always maintaining one or more valves at a maximum position to minimize throttling losses and energy consumption. There are a number of ways to do this. For case 2 it usually involves manipulating the pressure setpoint to force one valve further open until it is at a predetermined "Open" position. For case 1 MOV is not necessary. For case 3 the most open is either a designated valve or a comparison and the most open valve selected at current positions at any given time.

update

Hey guys. Thanks for your help. Let me give some more info. I have 4 low lift station pumps providing raw water into a common 30" header. From the common header, the flow is split east and west into individual supply pipes. Each supply pipe is then supplying water into a claricone. In line of each supply pipe is a control valve and a magmeter as stated in the beginning. The claricone is a huge 3 story high salad bowl. The purpose is to have water come in at the base and mix with chemical injectors at the base also. The top is open and a trough is used to exit the treated water from the structure. This is how it was explained to me. My guess is that the client requires the flow to be balanced/eqaul between the two claricones. The spec I'm reading is very vague and the client is a bit ****y when it comes to explaining more. I think the client has a hard time explaining what they really want. I do like the idea of having a single setpoint driving both PID blocks. Anyways, you guys really did help me out with this. I now have a better path to take when doing my programming. Just needed a little nudge in the right direction. I'm definitely not a know it all and pride myself in asking for other peoples advice. As its shown here, many good ideas have sprung up. Thanks guys.

Eric

Re: What's this?

Peter Nachtwey said:

...What do Einstein and Frank have to do with this?...

Click to expand...

Frank was Albert's well known brother!!!

Your pumps are probably going to be putting out variable flow rates to maintain a constant wet well level. In that case you probably want to designate one valve as the most open, probably the furthest from the pumps if you have a constant header diameter. Measure the flow rate through that unit, use it as the setpoint for the loop controlling the other valve, and you should be in business. You probably want to have tolerance and slow response in your loops, or you are going to have some hunting problems.

You can probably get some help from Walker Process - here is a site. They can not only clarify (no pun intended) their requirements, but may even have some suggestions for logic. It is in their best interests to have your controls work, or their priduct could be blamed for poor performance.

http://www.chicago-bridge.com/cbi/products/index.html

flow control

Hey there Tom. Thanks for that URL. I think you nailed it right on the head with the wetwell scenario. I finally got clarification from the client as to how the control strategy works. What they want is to match the plant flow through both effluent pipes. For example if the plant flow is 4 mgd from the raw water pumps, then they want 2 mgd flow through each pipe.

I do have a flow rate signal from the raw water pump flow meter to tell me the plant flow. I can then simply split the flow through both pipes equally. Each control valve will then run independently to keep its half flowing at the rate needed. The only drawback is that I have to have a message block from the one PLC to the other to retrieve that flow. As a safety precaution, I plan on driving both control valves wide open in case of comm failure.

I just kind of wished I could do the same with the wetwell level instead. Just having brain lock into how to do that. I would rather use the wetwell level because the plc monitoring the wetwell is the same running the control valves. Communications loss then becomes a moot point.

Although I still like everybody's suggestions on how to do it. Ton, thanks again for that URL. Appreciate it.

Any comments or suggestions will be very well looked into and taken into consideration.

Cheers, Eric