Extruders have traditionally used either forced air, or on-off controlled water cooling on barrel zones.

In our situation water cooling is used due to the amount of heat that must be removed, and this is done with the time-honored Asco valve used in on-off (timebase 10 to 20 seconds) PID control.

For most materials and process conditions this is fine, but there are some combinations (usually, involving high throughput jobs) where we'll see cooling oscillation regardless of how the process PID parameters are set up.

In those cases, very judicious adjustment of the flow needle valve can have a remarkable effect in stability improvement. I've noted sensitivities of 1/4 to 1/8th turns out of 9 total turns - usually somewhere around 2 to 3 turns open - between satisfactory vs. oscillating control.

The other concern is that (especially with extruders running APET material, where barrel temperatures in the 500 to 600 degree range are not uncommon) valve failure tends to be one of the leading causes of process variation and/or downtime. A hardier control actuator could well be justified for this alone.

It seems to me that there could be several different approaches to extruder barrel cooling, for instance, stepper or servo driven needle valves (in conjunction with shut-off valves for when nearing 0% output).

Has anyone familiar with this or a similar application ever seen anything other than on-off Asco valves used?

 

I have used proportional valves successfully in a similar application where heat removal rate had to be carefully regulated across a wide variety of products on the machine.

 

I've used Fisher control valves, for a similar application in the steel industry. The process speed is up to 120 meters/sec. The major components are 1)Control Valve: contact Fisher rep(Emerson Process Mgmnt). 2)Flow meter(Rosemount Magflowmeter). 3)Temperature device-( I assume you already have this in place, if not, Ircon make an extreemely rugged pyrometer that will work well for this application. 4)PLC, for PID control. It is critical that the valve is sized correctly. Another concern is if there are different size spray nozzles. If you get a good Fisher rep, he can lay out the components for you, it's pretty straight forward.
Good luck,
EG

 

Check "www.sterlco.com":http://www.sterlco.com
They have a line of cooling systems used on plastic extruders.

 

We had a simular problem with extruder throat cooling continualy diving under temp when the solenoid valve opened. I tried to tune it out with PID's with no success. I ended up reducing the water inlet & outlet line and valve size from 1/2" to 1/4". Put valve on outlet line and added a flow control. I moved the valve away from the extruder and mounted the valve body to the extruder platform where it acted as a heat sink. Another means of cooling the valve down is to bolt a piece of square aluminium tubing to the valve body. I did this over two years ago, acturacy is perfect and we have not had one valve failure since.

 

Greetings:
In a similar situation with cooling a "guide" which was built many moon ago, I modified the scheme. (The age of the machine is truly a testament to the quality and craftmanship of some of the older equipment. In this case an averaging thermal was installed in the cooling jacket). I used two TE's, one on the inlet and one on the outlet. This gave me my average temperature across the machine. As previously mentioned by the other contributors, determining the proper valve size is critical, and needs you consideration. Work to get that aspect as close as possible. As well as the two TE's for the inlet and outlet give some thought to another two for points within the jacket.
Good Luck!
Regards,
Paul

 

On extrusion chill rolls, a 4-20ma analog water valve has been used in many applications. It would work on a barrel zone as well, if the zone can not be controlled by other methods. An example valve is at "www.ciicontrols.com/new/ea81.htm":http://www.ciicontrols.com/new/ea81.htm .

 

Hello all;

In one of the previous posts, someone talked about extruder control forced air versus water cooled. I have both types of extruders, I am using Micrologix 1500 PID to control the heating and cooling of the barrels. I do not have much trouble with the forced air cooling. I simply wait for the heating output to go all the way to zero, before I start the cooling fan. However, with the water cooling it is much more difficult task. I am applying the same concept, waiting for heating out put to go to zero before water flow to start. The difficulty is because of the latent heat of vaporiztion of water, the barrels cools very quickly. As a result the heating output comes on. This makes the control very unstable, switches between cooling/heating. I know some of temperature control supplier such as Eurotherm, they do the water cooling very well. Water cooling is considered non-linear and forced air is linear.
Does anyone has done water cooling using PLC PID to do this? Any guidence would be appreciated

Thanks

 

For water cooling you have to set the PID gain much less than for heating. You need a gain-scheduled PID controller with different settings for heating and cooling. I assume you are using PWM. The water cooling valve would only pulse open for a fraction of a second every 10-20 seconds.

You also need a deadband between heating and cooling or you will cycle between heating and cooling and waste energy.

Warren
http://www.pc-pid.com
"the PC-based PID Controller people"

 

I'll concur with what Warren says. I'd farmed out some programming work for extrusion barrel heat/cool control on a SLC500 platform, and the scheme has separate heating and cooling PID control blocks (which allows the proportional bands to be different), and a deadband crossover range.

However, the cooling side is straight PID output into the duty cycle timebase (i.e. 50% cooling output and a 20 second timebase yields 10 seconds
on/10 seconds off).

I've found, for a 4-1/2" extruder, the I (and D terms, if you are so inclined) are approximately the same value between heating and cooling
blocks, and, depending on your cooling specifics (plumbing, water pressure, metering valve setting, etc) the cooling prop band will be anywhere from 30 to 70% of the heating prop band value.

If I ever delve into it (it works well enough, and spare time to hoke around with things that are working is in short supply) I would do the following, although I doubt it fits into the strict definition of gain scheduling.

1). I'd keep the basic programming the same, but add a second "inner" timebase. The first, "outer" timebase would remain, in this example, 20 seconds, and the second timebase would be a much shorter duration (say, 5 to 10% of the outer timebase, or 1 to 2 seconds). PID control block output would work based on this second timebase. For instance, if the inner timebase was 2 seconds, then 100% output would be 2 seconds valve closure, and the outer timebase would mark the overall cooling cycle time of 20 seconds. For 50% output valve time on would be 1 second, 25% output yields 0.5 second, and so on.

2). Add programming to automatically change the inner timebase preset value from the original 'seed' setpoint up to the value in the outer timebase in case the loop saturates. That is, if you have a setpoint of 450°, PID cooling output is hammered at 100%, and measured temperature value is consistently higher than setpoint then the inner timebase duration is
gradually increased until the loop drops out of saturation. Instead of 2 seconds go to 4 seconds, wait 20 minutes for the zone to settle, then, if
needed, go to 6 seconds, etc... In case the loop satruates due to an actual problem (for instance, an open solenoid coil) instead of due to changes in process conditions add a means to return to the original 'seed' inner timebase setpoint. Programming could also be added to detect when
band-crossing cooling oscillation is occuring, and shrink down the timebase setpoint until it stops.

3). This isn't likely to be an option (I'm guessing you are either using a small HMI, or maybe none), but if you have sufficient screen real estate this might be a useful observation. Eurotherm EM-1 extrusion controllers circa early '80s listed both the current setpoint, and last setpoint for all loop parameters on the loop setup screen. I'd always found this useful to
catch cases where someone inadvertently (or unadvisedly) changed one or several setpoints to poorly performing values.

Another thought is to set aside a block of registers to save PID, timebase, et al. values that are found to work well during commissioning, and add 'default' pushbuttons to the HMI that swap these values into the working setpoint registers. This way, unless legitimate field device problem exists, it will be possible for process techs and/or operators to regain a
sembalance of normal loop operation in the event an otherworldly setpoint has crept in.