Temperature gradient in 3 zone laboratory furnace?

[scottydoesntknow]

This is a surprisingly difficult control question with required details. I am putting together a stress rupture testing laboratory that uses 3 zone furnaces for elevated temperature testing. I have run into a roadblock with some nonbelievers that refuse to accept that temperature gradients occur in the furnace in certain temperature renges.
Here is the application:

1) Test coupons instrumented with 3 thermocouples evenly spaced on the coupon.

2) Coupons are typically 2-3 inches long.

3) The thermocouples also serve as control points to run the furnace.

4) Each Thermo controls its corresponding zone.

5) The control mechanism for each zone is an independent PID loop.

6) Coupon is attached to loading rods out of both sides of the furnace the do conduct heat out of the furnace.

When run with tuned PID's they perform marvelously. I have temperature deviations on the order of half degrees at +1800F temperatures. No problem here.

The problem is that it was decided to convert said 3 zone furnaces to 2 zone and run them in much the same manner. When run with 2 control thermocouples, it also performs marvelously, HOWEVER, when outfitted with a monitoring thermocouple in the center of the specimen, the temperatures are 3-5 degrees F higher.

Is there a resource that I can use to explain heat gradients to said nonbelievers to convince them that YES there can exist a 5 degree difference in temperature over 0.75 inches in a furnace?

 

[MintJulep]

How is your furnace heated, and what view of that heat source do the thermocouples have?

Thermocouples can be sensative to radiant energy, causing them to read high.

 

[scottydoesntknow]

They are 21 amp Satec 3 zone split tube furnaces, so they are radiant furnaces. The thermos are wire tied to the specimens and in full view of the furnace elements. I suspected that they would be sensitive to that kind of energy.

 

[IRstuff]

If the two control sensors are in zones 1 and 3t then zone 2 would be driven by the mean value of 1 and 3, so it would essentially output the same heat as the outer zones. But the outer zones have ends at lower temperature while the center does not. This would result in a higher temperature in the middle, since it's heat loss is lower than the outer zones. When running as 3-zone, the middle zone should normally see a slightly lower heat input to account for the difference in heat loss.

TTFN

 

[hacksaw]

two sensor control insures that the peak is in the middle, as you've found out.

how are you matching the sensors and what sort of stability are you seeing?

 

[scottydoesntknow]

For 3 zone control, obviously, top sensor goes to top zone, middle to middle and bottom to bottom. This control mode gives fantastic control. Control gets better at temperatures above 1000. From what I understand this is where radiation starts replacing convection as the primary mode of heat transfer according to our heat treat guy.

For 2 zone control on a 3 zone furnace, what we have been doing is shunting the output signal from the center zone to the missing outter zone. This works well as long as the specimen is shifted away from the outside zone that gets the shunt.

Some other furnaces that we use are in a 2 zone configuration, so we rewired the 3 zone to be 2 zone. Hence the 2 thermocouple control configuration. Control again was fantastic, we had over a 70 hour period a maximum deviation from the setpoint of 0.7 degrees at 1100F.

 

[IRstuff]

There's a damn good reason for designing the furnace as a 3-zone furnace. Your bypassing of a zone results in the non-uniformity, which should have been expected. You made a symmetrical system asymmetrical; the result has to be non-uniformity.

This has nothing whatsoever to do with the ability of the servo system to maintain its setpoint.

TTFN

 

[scottydoesntknow]

It was expected. Don't get me wrong, I'm not trying to understand what's going on here. I'm trying to convince my superiors (non-engineers who do not quite understand heat transfer and controls) that this is how it works. I'm not a thermodynamics expert by any means, I'm a computer engineer. BUT I did pay attention in physics and I DO adequately understand controls. I was thinking that someone here who understood the problem, which all you guys who have responded to this post obviously have no trouble with, could possibly help me dig up some resources to prove to these guys exactly what we are saying. Thanks to everyone who took time to respond. I very much appreciate it.

 

[IRstuff]

OK, I get the picture. The bottomline is that the control accuracy is irrelevant to the thermal profile. You could easily run all 3 zones from the center control signal. Your control accuracy would still be dead nuts, but the thermal profile will not be what the control loop was designed for and you'll get an odd thermal gradient.

Your people are confusing the ability to maintain setpoint with the other operational requirements of the system.

Obviously, the ideal thermal control system would have a much finer granularity in zoning, e.g., a 5-zone system would produce a much flatter and longer isothermal zone and a 9-zone would be even better. The limit on granularity is cost and the ability of two adjacent zones to have independence. A zone much smaller than the diameter of the furnace would probably not make sense.

Going from 3-zone to 2-zone not only increases the granularity, but makes the system asymmetrical.

A simple thought experiment would be to ask all involved if they think that a single controller could produce the desired thermal profile. If not, then ask why they would not expect that a two-zone configuration would be worse than a three-zone configuration.

TTFN

 

[scottydoesntknow]

I hadn't considered using that example... It makes perfect sense. I think that might be straightforward enough that some kind of understanding can be reached before the eye-glazing begins. I'll give this approach a go and post back with the results soon!

Thanks!

 

[scottydoesntknow]

It was the silver bullet, alright. And I didn't even need any equations or theories. Thanks a bunch. I produced power vs. time plots for all three zones and it showed most of the power being diverted to the top and bottom zones with the center eventually going to almost no power. The temperatures were well within tolerance for 70+ hours.

Regards,

Scotty

 

[IRstuff]

Cool!! ;-)

TTFN