@Antoine10FF,
We discussed that hot banding effect on another thread, with photos. ( The one that lead to this one) There seems to be at least two other causes that have been experimentally demonstrated for the higher heat bands.
What bothers me most about it being ascribed to directly to heater coils is the incredibly short wire length required. The non-coiled part of the heater wire is then nearly 50% of the total wire length, which is remarkably inefficient. Additionally, where are the other two coils? If that is all three, the shadow phenomena makes no sense, and still makes no sense even if that is one coil.
Electrically, the three coils in the patent application diagram are difficult enough to make work. Shortening the coil further becomes very hard to make work at all. The device was hot for a month, so massive current electrical ideas must at least be stable enough to do that. The device is almost a short circuit with conservative designs.
[Technical Thread] Brightness of the reactor glow in the Lugano pictures and reactor temperature
-
-
@Antoine10FF
Have you considered that the pitch you see evidence for in your analyzed photo is the "beat" between the pitch of the ridges on the outer tube and the pitch of the heater coil?The heater coil is 99% likely to be as shown in the patent drawings - a triple wound 3-phase coil that is wound with about 1-2 wire thickness spacing.
That's a very interesting point, thanks. No I hadn't thought of that. It should be easy enough to test.
Thaks for your other comment too. I'm aware of the difficulties about visible photometry you mentioned (near IR spectral response, changes of spectral response with different cameras, etc.) I have code that takes into account varying spectral responses, with data for 12 camera models, color space conversion effects, camera color temperature adjustment models, temperature-dependent spectral emissivity for Inconel and transmittance for some kind of alumina.
I understand that there are large uncertainties about the alumina spectral properties due to composition and texture. My approach is to plug those in and see how they propagate. The output will possibly be an uninformative interval, but it's still an interesting (to me) exercise.
What is missing is a geometrical model and alumina scattering/absorption. Without going full FEM, I still have to make use of basic parameters for a cylindrical geometry.
I have no plans about getting into non-visible IR emissions. -
Eric and Bob
The Lugano paper is in some respects badly written - as well as containing methodological and calculation mistakes. As you know p9? (I forget) they state that for the dummy run they put TiO2 calibration dots all over the reactor and used these to determine temperature. They say that to make their (wrong) calculations work they had to "alter" the book emissivity - but the point is that the temperature calculation here will be correct, since the cal dots have well defined published band emissivity which they used.
The dummy calculations, given exact temperature, are still not robust because of the very complex calculations of convective heat which at these lower temperatures is a very significant part of the whole budget. No-one looking at the assumptions in the convection calculation could reckon that is robust - for example what temperature is the convected air? They do not have measurements throughout the test that would measure local heating. And the calculations ignore edge effects that must be significant.
However - given we have an accurate surface temperature for the active tests, and therefore also have an accurate estimate of the heat loss from the surface, we can make some progress in calculating the wire temperature from the temperature gradient between wires and surface. That does not much help because the big unknown is wire emissivity, but it is interesting.
-
@Thomas Clarke
Be careful when you say "we have an accurate surface temperature". The surface temperature is complicated by the ridges that are molded in to maximize convection. MFMP's coarse thermocouple tests [coarse because the thermocouple bead was reasonably large compared to the ridge pitch) on molded ridges showed substantial temperature difference between the ridge roots and the tips. The temperature difference could be more than 50C - hotter at the root and cooler at the tips. Some IR cameras, such as the Optris, do not image to sufficient resolution to resolve this and instead record the average temperature in each pixel. Other near IR cameras would record the maximum value in their spot. So, for the ridged surface, you need to know the temperature's functional form on the ridges as viewed from normal incidence. The average temperature the Optris would read is probably not just the simple average of tip temperature and root temperature; though that may be a reasonable first order estimate. -
Quote
Be careful when you say "we have an accurate surface temperature". The surface temperature is complicated by the ridges that are molded in to maximize convection. MFMP's coarse thermocouple tests [coarse because the thermocouple bead was reasonably large compared to the ridge pitch) on molded ridges showed substantial temperature difference between the ridge roots and the tips. The temperature difference could be more than 50C - hotter at the root and cooler at the tips. Some IR cameras, such as the Optris, do not image to sufficient resolution to resolve this and instead record the average temperature in each pixel. Other near IR cameras would record the maximum value in their spot. So, for the ridged surface, you need to know the temperature's functional form on the ridges as viewed from normal incidence. The average temperature the Optris would read is probably not just the simple average of tip temperature and root temperature; though that may be a reasonable first order estimate.
That is a fair point, but as you say it is a second order effect. Let us estimate its magnitude. If we have 50C difference then we get a change in T of:
1010 +/-25K which is +/- 2.5%.We know that the relationship between T and band radiant power is roughly T^2, whereas that between T and total radiant power is nearer to T^4. This will give us a second or order correction:
0.5*[(1+x)^2 + (1-x)^2] = 1 + x^2
0.5*[(1+x)^4 + (1-x)^4] = 1 + 6x^2.So the effect here is that at most average temperature has an error of 3E-4 or 0.03% and average radiant power has an error of 6 X this or 0.2%.
Then for this calculation from surface temperature we do indeed need to take the ridges into account, with potentially a bigger error, but still not so large it invalidates results I'd expect - I'll leave doing this more complex error analysis till it is needed.
-
However - given we have an accurate surface temperature for the active tests, and therefore also have an accurate estimate of the heat loss from the surface,
Apart from the question of the ridges, there is the fact that the dummy run had an input of only 500 W, while the live run went up to 900+ W. Also, because the ridges prevented the TiO2 dots from being used on the body, they were instead affixed to the cable tubes leading into the reactor. Together, these two facts mean what was obtained was a calibration suitable for use against the tubes for up to 500 W input power.Can the parameters derived by the Optris under the preceding conditions also be used at nearly twice the input power of 900+W? Is it valid to calibrate the Optris camera using dots on tubes leading into the reactor and then point the camera at the reactor body during the live run?
These two considerations may be minor ones, or they may invalidate the temperature data. Do we know which is closer to being true?
