Two VFDs one discharge

[ptonsparky]

Customer with two irrigation wells feeding one Pivot. Neither pump is able to supply the water requirements of the Pivot. Both pumps have VFDs and are located remotely from each other. This is the first year that both will feed the pivot. I feel that unless one pump is made the primary source and allowed to run at full speed both pumps will "search" while trying to maintain 35# of pressure.

As one pump increases output the other will drop with the PID loops of both acting against each other.

Yes/No or a case of HUA and no need to worry?

 

[steve66]

In theory, the PID's would see the exact same inputs, and would provide identical outputs, so the VFD's would run at the exact same speed. (Assuming all the PID parameters are set the same.)

In reality, I'm not sure what you will get. It seems like differences in sensors and the PID gain will make the setpoints slightly different. That means each PID will be trying drive the output in different directions.

Is there any way to use a single PID loop, and connect the output to both VFD's??

 

[gar]

110405-1352 EDT

Where do you want to control the pressure? At the pump or pivot? And what is a pivot?

Consider the following approach or a variation on it:

One pump is connected to the pressure sensor wherever that might be. The speed of the second pump is servoed off of the speed of the first pump. Or a little more sophisticated approach is the same except there is a flow meter on each pump and the second pump has its flow servoed to the first pump's flow. If you can stabilize this system, then it might provide a more precise balance of the flow from each pump.

If the pumps are not positive displacement, then in the speed controlled approach you might have additional compensation for different pipe lengths to the pivot.

.

 

[ptonsparky]

Form a triangle with 1/4 mile sides. At the top sits a Center Pivot Irrigation system which covers a 1/4 mile radius for irrigation via sprinklers every 50' of so along the pipe. Bottom left corner has a pump/vfd as does bottom right corner both sending water to the Pivot. Pressure feedback for each vfd is at its respective location.

Pumps are centrifugal @ 300-600 gpm I would guess. Water required changes in the system as the end gun or corner system starts/stops if they are next to a road. Covers more acres that way.

 

[sii]

110405-1352 EDT

And what is a pivot?.

http://aspenranchrealestate.com/images/centerpivotirrigation.jpg

 

[hurk27]

I have seen a well set up very similar to this, it was a plant that required allot of cooling water that was via one temperature PID into a splitter kind of controller that operated both VFDs which controlled both pumps in parallel, both pumps would run the same RPM's providing a constant velocity of water flow per the cooling load requirements, I would think a parallel running of the pumps would reduce the ware and tare on the motors and pumps by letting both share the load and speed, trying to run one faster then the other would cause that motor to use up it's life faster and would require more replacing.

I would belive the company who design this system would be a good source for the required info?

 

[Jraef]

Pressure or flow is the only way to do this, speed would get you all fouled up. It should work, each loop will be looking at the pipe pressure and regardless of what the other VFD is doing, each one should be reacting ONLY to the situation in the pipe. If you think it through a bit, what's the difference between that and having a varying input pressure on the pipe and having one pump trying to deal with that with a PID loop? So if you substitute a natural variation situation with one produced by another pump in the system, what's the difference?

But tuning the two PID loops would be a bit of a PITA. No two pumps, and the wells they are in, and the pipes they feed into are going to be exactly the same. So you are going to have to tweak the slew rates on the PID loops, most likely by messing with the accel / decel rates on the VFDs in accordance with the differences in water table, pump set depth, pipe friction etc. I wouldn't want to be the guy to do this, these kinds of things never go well for me.

So a word to the wise:
Don't use the PID capability inside the VFDs. I know most of them have it built-in now, but here's my reasoning, and it ESPECIALLY applies to a PITA situation like this. The problem is, if the VFD dies and has to be replaced, you loose the PID programming in it and have to do it ALL OVER AGAIN! Whereas if you use a separate PLC or even a name brand loop controller, if the VFD dies, all you have to do is match the basic programming of the drives. All the complicated stuff for the PID loop tuning stays behind and doesn't change. It's a BIG pitfall people fell in to when VFD mfrs sought to "simplify" installations by integrating PI or PID control into the VFDs. The hidden "gotcha" is that unless you are willing to go through all of the tuning again, the only way to recover from a disaster or drive failure is to buy an IDENTICAL replacement drive from them, you can't go out and buy the fastest one available ! Then if, as is usually the case now, the original one has been replaced and no longer available, you are screwed.

Been there, done that, got the t-shirt and the hat.

 

[ptonsparky]

Customer had a pair of DanFoss that when one died it was just easier to modify the control and put in an AD vs deal with the BS of buying from them. They were overkill anyway.


As far as the company that designed this...so far just the farmer, a pivot guy and maybe me.

 

[Cow]

I had to do something similar once with a 150HP pump and a 60HP pump both feeding into a common waterline. They were about a 1000' apart. I tuned the 150HP and ran it, worked like a champ. Drove down the road, entered similar PID setpoints as the first drive(they were both ABB's) for a starting point, then spent half the day trying to tune it and get it to smooth out. Frustrating to say the least. When that 60HP would start up, sometimes it'd really push the water in the pipeline and run near 60Hz, other times it'd start up and only try to run around 40Hz. It was never consistent and was a complete pain to tune.

So if the thought process is to use PID on both motors so they both run and wear out the same, I'm not convinced they will. If one gets a jumpstart on the other it seems like it becomes the lead while the other will lag behind and not run full speed.

It might make more sense to make one 60Hz ALL the time and the other fill the gap using PID. Every so often you could switch it around.

Personally, I'd think a farmer would rather wear one pump out at a time instead of having two worn out pumps at the same time.

 

[gar]

110406-1108 EDT

ptonsparky:

I believe you are indicating there is a pressure sensor for each pump and these are at the respective pumps. Also that the pressure set point at the pumps is 35 PSI. Each pipe from each pump can be considered as a constant resistor. The sprinkler pipe and nozzles is a load resistor with some degree of variation. Pressure is like voltage and flow rate is like current. A pump in combination with its pressure sensor and feedback loop is like a constant voltage power supply with some internal resistance much smaller than its load resistance.

What is the expected pressure drop in the 1/4 mile supply pipe. Unless the pipe is rather large I suspect a substantial drop. Last night I put in some wild figures at the site http://www.efunda.com/formulae/fluids/calc_pipe_friction.cfm
100 PSI, 10 FT/SEC (6.82 MPH), 4" dia., roughness 0, length 0.25 miles, elevation difference 0, fluid density 1 kg.l, viscosity 1 cP. Results pressure drop 38.4 PSI, 392 GPM.

Need better figures from you actual installation and the expected pressure at the sprinkler pivot input point. If the pressure drop is large in the supply pipes compared to the sprinkler pressure drop, then the system with constant pressure pumps would reasonably provide equal flow distribution. I think knowledge on the pressure distribution is the first thing you need to look at.

.

 

[iwire]

Just throwing this out there .......

Lock one pump at 100% and let the second pump adjust to pressure signals. This seems to me like it would be less complicated. Maybe set it up so that each new start up alternates which pump is the lead to keep wear and tear equal.

But maybe I am just crazy.

 

[Jraef]

Just throwing this out there .......

Lock one pump at 100% and let the second pump adjust to pressure signals. This seems to me like it would be less complicated. Maybe set it up so that each new start up alternates which pump is the lead to keep wear and tear equal.

But maybe I am just crazy.

You are not crazy, several of the VFD mfrs, like ABB, have a built-in or optional "macro" control program that will do something similar to that, including alternating the drives. The only difference is that they don't "lock" the first one at 100%, they ramp it as necessary and only bring in stage 2 when the first one can't satisfy the demand. The only caveat is that one of the dives becomes the "master" that is running the macro control program and the other(s) become slaved from it. That necessitates that the VFDs themselves are fairly close to one another and exactly the same brand / model. This problem had the VFDs far apart, and as we just found out, different mfrs. But for sure if you had one separate PLC to talk to both drives, you could do that AND deal with the issue I mentioned earlier. That's how I would do it, but not everyone is comfortable with writing PLC programs like that.

 

[iwire]

But for sure if you had one separate PLC to talk to both drives, you could do that AND deal with the issue I mentioned earlier. That's how I would do it, but not everyone is comfortable with writing PLC programs like that.

I wrote a PLC program to control an overhead door, partially because it was a solution to a problem, partially because I had the small PLC and partially as an exercise in programing. The unit I had could only handle 100 lines of programing.

 

[Jraef]

I wrote a PLC program to control an overhead door, partially because it was a solution to a problem, partially because I had the small PLC and partially as an exercise in programing. The unit I had could only handle 100 lines of programing.

It gets easier after the first one

 

[Electric-Light]

Using a pressure tank on the output of each pump along with a check valve on both the input and the output of pressure tank(air bladder) prevents each pump from feeding back into each other and provides dampening and each pump's VFD will simply operate at speed to maintain the pressure at its respective tank.

Water is not compressible so lacking adequate pneumatic cushion, the pump will hunt.

It's commonly done with well water pumps.

 

[gar]

110418-1311 EDT

ptonsparky:

Have you finished the project?

If so what solution did you use and how does it perform?

.

 

[ptonsparky]

110418-1311 EDT

ptonsparky:

Have you finished the project?

If so what solution did you use and how does it perform?

.

Customer has decided not to connect the two pumps to one pivot at this time. IF I am involved in the future we will go with one pump as primary and let the second maintain pressure as needed. We may have to make some adjustments on the primary vfd to ensure that both pumps run fast enough to maintain air flow for cooling.

I think this will come closest to KISS, for what he has in mind.

FWIW my post #8 was referencing a different project.

Thanks