Measuring Temperature Effect of Thermal Barriers in Thermal Shock Test 3

[KimWonGun]

I am testing different thermal barrier coatings using a high temperature flame, a thermocouple sealed inside a boss on the specimen's back (not visible in the attached sketch), and a circulating water bath; the test specimen quenches in the water periodically.

The baseline is an uncoated, cast iron specimen. Theoretically I would expect the thermocouple temperature measurements with the uncoated specimen to be higher than measurements with coated specimens when the flame temperature is constant. Yet I observe the opposite. What am I overlooking?

 

[IRstuff]

The scale is unclear, but an uncoated surface will convect and radiate, so the uncoated specimen might have a higher heat loss.

TTFN
faq731-376

 

[KimWonGun]

Thanks for catching my oversight. The specimen has a 2" diameter and a slightly under 1" thickness.

In case it was not obvious, the thermal barrier coating is represented by the tan surfaces.

Attached is another image showing the back.

Would it be logical to conclude that the uncoated specimen always has a higher heat loss regardless of specimen geometry?

 

[MintJulep]

You appear to be assuming that a single-point temperature measurement is fully representative of the bulk average temperature. That assumption may not be true.

Could you explain the water quench?

Are you taking into account the energy required to vaporize the water after the quench? The amount of water on the specimen may be different coated vs. uncoated.

 

[KimWonGun]

I am only assuming that the single-point temperature measurement is consistently capturing the effect of the thermal barrier coating at the same location for purposes of relative comparison between the baseline and different coatings.

I fixed the specimen on an armature which holds the specimen aloft while the flame targets the front surface for a set period, after which the armature plunges the specimen just for a few seconds into a fixed volume water bath which recirculates. The cycle then repeats.

All surfaces of the specimen are submerged, both when coated and not.

I was unable to attach a photo of the set-up.

 

[IRstuff]

A coating, unless it has a higher thermal conductivity, will behave like a thermal insulator, which is what I think you're trying to make, will decrease heat loss through that surface.

A 2" diameter 1" thick object has the annoying property that the distance to your thermocouple is the same as the distance to the circumference. So, by the time you measure a temperature shift, the rim of the block has also warmed up. That would seem to be an extra layer of complexity in an otherwise complex problem.

I'm unclear whether your test accurately reflects the intent of your thread title. Ostensibly, a "shock" is a high amplitude, short duration event. Therefore, a shock barrier either reduces the amplitude, or increases the duration, or both.

Since it's supposed to be a shock, the flame is ostensibly a pulse? Have you looked at the time characteristics of your thermocouple measurements? You may want to thin down your sample to maybe 1/4", which might minimize the impact of the block's thermal boundaries.

TTFN
faq731-376

 

[KimWonGun]

You are correct; the coating is designed to behave like a thermal insulator and reduce heat loss through that surface.

Your comments about the specimen geometry and thermal conduction prompt me to mention that an infrared gun measuring the bare metal surface temperature along the width near the coated surface registers a LOWER value than the thermocouple measuring the temperature inside a boss from the back.

Attached is a plot of three tests, the baseline and two coated specimens, to address your comments and questions about shocking the specimen.

Thanks for your recommendation to reduce the specimen thickness.

 

[MintJulep]

Infrared thermometers are notoriously useless.

Your experiment seems really really complicated, with far too many interdependent variables and everything is transient.

 

[MintJulep]

If you are interested in heat transfer then isn't the slope of your curves the parameter of interest?

 

[ione]

I would approach this problem trying to minimize any effect which could affect the test. Since the thermal barrier coating is applied on a surface on your specimen, whilst the thermocouple is on the opposite, I would shield both the whole lateral surface to prevent heat loss/gain from those surfaces and practically the whole the back surface. You could embed the specimen in a thick refractory layer just leaving out the front surface, i.e. the one exposed to the flame, and the point where the thermocouple is placed. I think this way you could have a better evaluation of the “screen” effect produced by your coating.

 

[KimWonGun]

Actually, according to the previously attached plot there appears to be a steady-state maximum and minimum back temperature for all specimens after approximately 11 minutes.

As the plot shows, the initial slope is about the same across all specimens, so that variable is unhelpful for a comparative analysis. My focus, therefore, is comparing peak temperatures.

If I were only interested in the relative peak temperatures, does it matter whether the secondary surfaces are shielded or not so long as the environmental conditions are constant across specimens?

 

[ione]

One of the most important feature of a thermal barrier coating is the low thermal conductivity. In practice you have a series of thermal resistances (coating + specimen) and the purpose of applying an insulation on other surfaces is that of keeping the heat transfer process on a 1-d level.

 

[IRstuff]

I think perhaps you might take a step back and explain what end-effect you are trying to achieve, rather than dealing with your current point solution. If the objective is tojust keep the temperature down, then thermal mass will slow down the transient. High conductivity black fins and thermal shunts would dump heat away from the area you're trying to protect.

TTFN
faq731-376

 

[KimWonGun]

The test's objective is to help answer the question, "Which thermal barrier coating lowers the temperature the most on the other side of the coating?"

 

[MintJulep]

Insulation works both directions.

That is what you are seeing.

Your baseline sample cools more than the coated samples during the quench. The difference is most pronounced at about time = 250 seconds.

As a result the coated samples get hotter.

So the answer to your question is "No coating is better, because in this cycle, with this test sample configuration the barrier coatings slow down the cooling during the quench portion of the cycle."

 

[IRstuff]

OK, so from your data, the coating is insufficiently thick enough or insulative enough to do the job.

Is your sample representative of the actual application? If so, then you need to be looking at a different material/thickness, and possibly add cooling features like fins to the structure.

TTFN
faq731-376