Web Unwind and Rewind Control

I have R&D project that I am working on. It consist of an unwinder and a rewinder. The goal is to unwind the material from the unwinder and rewind the material at a specific tension and web speed on the rewinder. The material being wound is a strong plastic that can be stretched.
Web Width=30"
Core Dia.=3"
Max Roll Dia.=30"
Max Web Speed 650fpm

The hardware consist of siemens micromaster 440 drives that have encoder feedback. Each unwind/rewind motor has a load cell to measure tension. PLC is a S7-1200 series with an et200s rio
speed/speed trim signals to from drive are hardwired analog.

I have developed a few control systems that were rewinders only. Material was being fed to me from by another piece of equipment that was out of my scope. The control theory was calculated roll diameter and web speed sp. determined base speed and a dancer/load cell determined speed trim.

I am unsure how to go about controlling this process. What is standard practice? I would like to operate both drives to control speed but I feel since there is no dancer to increase or decrease the web length from rewind to unwind that the control would have to be so tight that it cannot be achieved.

I have never controlled a motor via "torque mode" and I am not sure how to go about it. It would seem to me that the motors must be sized very specifically to the process. I have a feeling they are not and are over sized quite a bit.

Luckily this is an R&D project so we can go about trying different practices.

Please see the attachment for a graphical view of the process.

Any help is greatly appreciated.

Have you consider getting rid of the unwinder and just use the rewinder? I don't see why you need an unwinder

In the diagram below, you would place your plastic roll onto spindle A. You control the WEB speed with the motor attached to roller D. You control the tension of the web by controlling the brake at spindle A based on the readings you get from the load cell at roller C.

David,
I agree with you, unfortunately what we want to try to make the machine run with motors on the unwind and rewind.

I don't think you need 2 loadcells. Set the running speed to control the rewind drive, and have the loadcell's feedback control the Web tension by changing speed of unwind drive based on a tension setpoint.

So there's not a nip in the middle?
I think your going to have a tough time with this one
Without something to isolate loadcell one from loadcell two
How can you control tension without affecting the other?

And 650fpm?
Also where are you doing the diameter calculation?
As far as drives go I would look at the Siemens S120

Since you have control of both motors I would think more in terms of gearing and changing the gear ratio to keep the load cell force constant. This way the unwind is NOT simply reacting to the load cell but to a combination of the load cell and the rewind.

I agree that there should only be one load cell.

JRW said:

So there's not a nip in the middle?
I think your going to have a tough time with this one
Without something to isolate loadcell one from loadcell two
How can you control tension without affecting the other?

And 650fpm?
Also where are you doing the diameter calculation?
As far as drives go I would look at the Siemens S120

Click to expand...

I agree. Traditionally, <something> is being done in the middle, and that is the 'pacesetter'.
Of course, if you don't have a pacesetter, you are just operating what we used to call a "re-winder".
There needs to be a line speed 'reference', at a minimum.
But, without a nip, the 'tension' in the sheet is uniform at unwind and rewind.

As it an R&D project, you can start with the minimum as you have described.

Calculate Rewind/Unwind diameter from Core/StartRoll Diameter and number of revs of rewind/unwind and material thickness.

Rewind speed demand = Line Speed demand/(Rewind diameter*pi)

Unwind speed demand = Line Speed demand/(Unwind diameter*pi)+load cell trim

jdbrandt said:

I agree. Traditionally, <something> is being done in the middle, and that is the 'pacesetter'.
Of course, if you don't have a pacesetter, you are just operating what we used to call a "re-winder".
There needs to be a line speed 'reference', at a minimum.
But, without a nip, the 'tension' in the sheet is uniform at unwind and rewind.

Click to expand...

So when you make a tension set point adjustment where will you see the tension go? Both loadcells will see it right? My opinion is that an s roll is needed in the middle. This would be his master speed too.
What I would like to know is what is the process? Normally there is something in the middle anyway....coating..stretching...something

Of course , he didnt specify if these are center wind and center unwind. I'm assuming from the pics that they are.


I hope he keeps us posted on his progress. Generally these threads die

My drawing list (2) load cells but we do not necessarily have to use both. There is a dancer available on the unwind but they would prefer to just use the load cells.

As for the process in the center, there is none at the moment its as is in the drawing. The plan is to to place some sort of folding process that will fold the film over onto itself. I am not involved in this aspect.

For some reason I see a picture of a dog chasing his tail when trying to think about using (2) load cells.

JRW said:

So when you make a tension set point adjustment where will you see the tension go? Both loadcells will see it right? My opinion is that an s roll is needed in the middle. This would be his master speed too.
What I would like to know is what is the process? Normally there is something in the middle anyway....coating..stretching...something

Of course , he didnt specify if these are center wind and center unwind. I'm assuming from the pics that they are.


I hope he keeps us posted on his progress. Generally these threads die

Click to expand...

Unless there's a nip point, PSI in the sheet is the same at the unwind and the rewind. PSI = width x thickness x pounds pull.
Winder pull is a function of the torque (K * (phi) * Ia) and the moment arm, which is a function of the diameter.
If the unwind's pull is greater than the rewind's pull, the unwind wins, and the sheet will go backwards.

Jeff
I not disputing the formulas, just tuning and getting it to run,
Especially through a complete roll build at 650fpm


Stuntman

So you have a few opinions here
Do you have control scheme figured out yet?
Post a block diagram when you do
I would curious on what your going to try

This is a very interesting project

JRW-I am thinking of using the unwind to set web speed and then having the rewind control tension by trimming its speed reference. The micromaster 440 has a PID dancer function that does precisely that. I am unsure how the unwind drive will react when/if the rewind creates a pulling of the web attached to the unwind. My theory is that if the pull is great enough the unwind would go into more of a coast. This in turn would create energy which would raise the dc bus voltage and fault the drive. If that is the case somehow the unwind needs to act like an intelligent brake. I am going to sleep on it and take a look again at it tomorrow. I agree it should be an interesting and fun project.

This is is your gear ratio

[QUOTE='L D[AR2,P#0.0]
Rewind speed demand = Line Speed demand/(Rewind diameter*pi)

Unwind speed demand = Line Speed demand/(Unwind diameter*pi)+load cell trim[/QUOTE]
GearRatio = (Line Speed demand/(Rewind diameter*pi))/(Line Speed demand/(Unwind diameter*pi))+GearRatioTrim(load cell)
or
GearRatio = (Unwind diameter)/(Rewind diameter)+trim
but you don't know the diameters so use the load cell as a trim for the gear ratio.
GearRatio = GearRatio + GearRatioTim(LoadCell).
I would start with a simple proportional gain.
GearRatio = GearRatio + K*(LoadCellTarget-LoadCellActual).
This is basically an integrating process. The GearRatio will integrate up or down until the actual load cell data matches the set point.

I actually think of the master as being the unwind and the gear ration is the ratio of the slave speed to the master speed. starting out the slave speed will be slow because its spool is empty and the gear ratio will increase. The load cell will try to speed up the slave or increase the gear ratio if the actual load cell force is higher than the set point force so actually the gain, K will be negative.

After thinking some more about this I think that if it is possible to know the diameter or it is reasonably predictable it would be good to try calculating the gear ratio on-the-fly
GearRatio = (Line Speed demand/(Rewind diameter*pi))/(Line Speed demand/(Unwind diameter*pi))
If the tension on the load cell varies from the set point then use a phasing command to advance or retard the unwind a bit to maintain desired tension.

The problem I see with this is that a real motion controller is required.