HIGH (800C+) wall temperature measurements - experimental help

[TTM]

HIGH (800C+) wall temperature measurements - experimental help!

Hi All -
I'm seeking some advice from any experimentalist types. I'm working on a project analyzing heat transfer properties of exotic fluids. We have a simple experimental setup: laminar flow in a small diameter stainless steel tube (~1/4inch) with constant heat flux boundaries (delta T over the tube length ~50C). HOWEVER, this experiment is at such a high temperature it will glow red (800C).

We need accurate surface temperature measurements in order to create useful heat transfer correlations (Nu as a function of Re, Pr, Gz, etc).

Anybody who works with thermocouples at high temp knows they are susceptible to large error if the leads to the joint undergo a large thermal gradient. Unfortunately, that is a fact of life for us if thermocouple wires are to be routed to a room temperature junction. (see pg Z-29 http://www.omega.com/temperature/z/pdf/z021-032.pdf)

I have looked into alternatives, including very thick wall tube, EDM and spectral analysis of the tube surface. Does anyone have any experience in high temp. and could shed any light?

 

[boneh3ad]

And yet the steel industry uses thermocouples to monitor temperatures much higher than 800 C. My thoughts are that you ought to try a different type of thermocouple and look into how the steel mills do it.

 

[TTM]

I'm going to guess they don't need a very precise measurement, and if they do, they can afford to design their geometry around these measurements. We must maintain a uniform surface heat flux, so any bulky equipment which disrupts that condition will disrupt the experiment. Trust me, this problem is not as simple as it appears, many people with many degrees are scratching their head along with me...

 

[boneh3ad]

You would likely guess wrong. In heat treating and alloying metals, you have to have accurate temperature control because a degree too high or low can cause the mixture to cool in a different phase than what is expected and cause different physical properties. It is a lot more involved than you would think.

At any rate, I do a lot of work with thermocouples but never any higher than about 500 K, so I have no firsthand experience at higher temperatures. I just thought I would suggest looking into the sorts of things the steel industry does as it may offer some insight.

How accurate do you need to be?

 

[TTM]

Within 5C (or 10 even) would put us in a much better place than we are now. Your post did spur me to a closer look at pyrometers, which is generally what high temp manufacturing and PCB industry uses. They come with their fair share of difficulties.. they can give you an accurate surface measurement, however are 1000x more expensive than an off the shelf type K and much more difficult to install correctly. Line of sight access to our test section would take careful consideration as it is surrounded in 6 inches of insulation.

The price one pays for working at high temp!

 

[boneh3ad]

What about optical methods like IR thermography? I suppose that would necessitate having a clear view of the test section though and probably wouldn't solve the cost problems.

Have you looks into thermocouples other than type K? Maybe type B or something? More expensive, sure, but not a thousandfold.

 

[TTM]

All thermocouples regardless of type will suffer from decalibration caused by steep temperature gradients, since they all rely on the seebeck effect.

Robert Moffat in his Gradient Approach to Thermocouple Thermometry explains that the thermocouple voltage is actually generated by the section of wire that contains the temperature gradient, and not necessarily by the junction.

A steep temperature gradient will exacerbate any inhomogeneities in that section and give you a drifting measurement.

Cost is a big issue of these IR devices, not to mention the possibility of an oversight on our part - we drop 20K on a set and find out they don't work in our case because of some stupid small thing we overlooked. If anybody that has experience with these could chime in that would be nice

 

[brewnog]

I measure exhaust temperatures of up to 780C using good K type thermocouples, we find the accuracy is easily within 10C.

 

[TTM]

if you don't mind me asking, how do you come up with this 10C figure?

 

[AlephZero]

Anybody who works with thermocouples at high temp knows they are susceptible to large error if the leads to the joint undergo a large thermal gradient. Unfortunately, that is a fact of life for us if thermocouple wires are to be routed to a room temperature junction. (see pg Z-29 http://www.omega.com/temperature/z/pdf/z021-032.pdf)

I don't follow your logic there. Almost any high-temp measurement is going to have the measuring equipment at room temperature. All the PDF is saying is that (1) the voltage difference is actually generated across the temperature gradient (and if you think about it, where else could it possibly be generated??) and (2) if the properties of the leads have been degraded in the temperature gradient region this may lead to errors or unstable readings. So the "fix" is simply not to degrade the leads by mechanical abuse, exposure to the atmosphere, etc.

I'm with brewnog on this one. We use thermocouples up to 1000C in some pretty hostile environments (for example rotating machinery with accelerations of a few thousand G on the hot junction, and the leads going through slip ring connectors to the data logger), and they are pretty much bistable devices: either they are working and accurate, or they are broken and the output is obviously nonsense.

You didn't say whether your tube was 0.25''ID or 0.25''OD, but it could be your problem is the thermal path between the couple and the (presumably curved) surface it is attached to, and this is affecting the temperature distribution in the tube. The obvious fix for that is make the tube a bigger OD.

 

[TTM]

Some more info: its .25inch OD (.083inch ID nominal), and the thermocouples are welded to the surface. (if you can think of a more robust way of attaching it without introducing more contact resistance, i would be happy to hear)
I think we all have a different definition of "working". Sure the TCs work, they give me a great ball park answer, which unfortunately is not good enough. The experiment must maintain a constant heat flux boundary (the whole point of the project) so temperature problems are compounded by the fact heating elements are right next to them. Ohmic heating is infeasible.

More from the omega PDF:

Thermocouple wire obviously can’t be manufactured perfectly; there will be some defects which will cause output voltage errors. These inhomogeneities can be especially disruptive if they occur in a region of steep temperature gradient. Since we don’t know where an imperfection will occur within a wire, the best thing we can do is to avoid creating a steep gradient.

The wire doesn't need any mechanical stress to degrade the alloy, they already exist as a fact of manufacturing. I think anyone could agree a 800C to 25C drop over 4 inches is a steep gradient, which is ill advised. Not to mention the unknown effect welding has on modifying the calibration curve, which I suppose could be calibrated out but at what effort?

Dont even get me started on how we measure flow rates of a fluid at 800C!

 

[boneh3ad]

Welding your thermocouple will do more damage than the temperature gradient will. The temperature gradients involved in welding are MUCH greater than anything you are measuring: an order of magnitude greater actually.

 

[TTM]

Welding is common practice and if done right can give good results.

 

[brewnog]

if you don't mind me asking, how do you come up with this 10C figure?

Gauge R&R, and calibration of multiple probes against a common heat source.

 

[AlephZero]

The picture I'm getting is that you have an insulated enclosure with a heat source of konwn power inside it (IR radiation heating?), and you are trying to measure the temperature of your pipe with some fluid flowing through it. (And also the temperature and mass flow of the fluid, but that is not the question).

I would say you have two problems:

1. Unless you are doing your welding in the same sort of environment that the T/C manufacturers use, you are doing an unknown amount of damage to the T/C.

2. T/C's don't measure the temperature of what they are attached to. They measure the temperature of the T/C junction. If your T/C is on the surface of the pipe, it is receiving heat from outside. It is also probably disturbing any convection in the air around the pipe. The thermal conductivity and emissivity of the T/C and the weld is probably not the same as the pipe so it will tend to form a local hot or cold spot.

I'm assuming there is no significant radial temperature gradient through the pipe (or if there is, you can calculate it accurately enough from the material properties of the pipe and the heat flux.)

I would try burying the T/C junction inside the pipe wall. You may need to use a thicker pipe so you can machine a cavity in the wall. Instead of welding, fill the cavity and bond the T/C with something like Aremco pyro-putty (this is a thermally conducting stainless steel/ceramic composite, which is good up to about 1100C). Make the cavity small enough diameter and deep enough so the disturbance to the surface heating has a chance to even out before it reaches the T/C junction at the bottom of the cavity.

Even if it doesn't work, it won't cost much to try it.

 

[gts]

Try TaSiN electrodes and talking to QFI to calibrate your measurements optically