First issue: 08/02/2015
Some recent experimental measurements by the Martin Fleischmann Memorial Project (MFMP) highlighted a possible error in the Hot-Cat calorimetric measurement; the calorimetric measurement we are referring to is described in the document known as “TPR2” or Lugano Report. In particular, in the report of the MFMP experiment the following consideration is stated:
“The main revelation was that the emissivity required for the camera to correctly interpret the temperatures on the surface was very close to 0.95. When we plugged in the emissivities cited from literature in the Lugano report (0.8 to 0.4), the apparent temperature was 1200° to 1500°C at 900W in.“
GSVIT already long ago had suggested that the measures and the calorimetric analysis performed by the Lugano Report authors (in the following AA) (G.Levi, E.Foschi, B.Höistad, R.Pettersson, Lars Tegnér, H. Essén) contained a fundamental error since they considered the Alumina as a gray body, as far as thermal radiation is concerned.
The experimental measurements by MFMP were carried out on a replica device (a simulacrum of the real Hot-Cat), confirmed what was only a hypothesis by GSVIT and, where correct, they allow to affirm that the method adopted by the TPR2 AA to calculate the reactor surface temperature (and therefore the amount of heat irradiated) was incorrect.
In general, it is difficult to define the error generated from a wrong method set-up on the outside temperature and radiated power measurements, since it is not sufficient that the simulacrum be identical in shape, dimensions and materials. Actually, Alumina can assume 7 polymorphic forms with different crystalline structure and industrially the word “pure Alumina” starts to be used for 99% purity. We want to recall here the results of the analysis of samples of the Hot-Cat Alumina as reported in TPR2:
“The results confirmed that it was indeed alumina, with a purity of at least 99%. Details of this analysis will be found in Appendix 2.“
However, although the presence of small percentages of impurities can alter its characteristics (what is exploited to make it fit better various applications) we did not succeed in finding in the literature a significant correlation between impurities and the Emissivity value (ε).
Assuming that MFMP used an alumina very similar to the Hot-Cat alumina, the resulting error in the Radiated Power estimation could be close to 2. In a similar way, the surface temperature would be overestimated by a few hundreds degrees; thus, considering that a wrong temperature measurement has a direct impact also on convection transferred heat by: Q=h*A*(Ts-Tamb), the error on the estimation of the emitted Power may be even higher.
Errors of this level of magnitude would be so important to question seriously the Report content and its conclusions.
As previously highlighted in a technical analysis coming from Alan Fletcher, AAs adopt as their calorimetric method a measurement of Radiated Power based on the well known Stefan-Boltzmann law linking the Power emitted to (T)4, where T is the object absolute temperature. This measurement technique requires full knowledge of the value of the surface Emissivity of the object whose temperature one would like to estimate; the reason is that the temperature of the body is a key element for the calculation of the Power radiated through the Stefan-Boltzmann law.
The TPR2 AA decided to measure the Hot-Cat temperature by using only a thermo camera (model Optris PI160) and it is worthwhile to remember that they also said that it was not possible to place a reference thermocouple directly onto the reactor surface to establish a reliable reference of the actual temperature:
“We also found that the ridges made thermal contact with any thermocouple probe placed on the outer surface of the reactor extremely critical, making any direct temperature measurement with the required precision impossible.“
(This is a claim already disputed in the past because incomprehensible; this claim was proven wrong by MFMP that, instead were successful in placing it). The AA decided to calibrate the camera, used for measuring the surface temperature, according to this procedure:
The temperature provided by the camera is fed to a diagram that maps the Emissivity temperature of pure Alumina (Figure 6 of TPR2)
The diagram is checked to verify if the Emissivity value actually corresponds to the read temperature
When no match is found, the camera Emissivity settings are adjusted until a perfect match between the temperature indicated by the camera and the temperature on the ordinate of the Emissivity/Temperature diagram for Alumina is met
Eventually, the camera Emissivity is set to the found value
This method is correct only if the Alumina surface behaves as a gray body and the used diagram contains Emissivity data [see Note 1] for Alumina very similar to the one actually used. One has to remember that the Emissivity ε of a material is the fraction of energy radiated from the material compared to the energy radiated by a blackbody of the same temperature. Since the blackbody Emissivity is by definition unitary , any real object has Emissivity less than 1: if this value does not depend on the wavelength (λ) such material is called gray body. If instead the Emissivity versus wavelength has a completely different shape, such a material cannot be defined a gray body. This is, for example, the case of Alumina.
Experimental Determination of Emissivities. In the addendum you will find emissivity dates for various materials from technical literature and measurement results. There are different ways to determine the emissivity.
Method 1: With the help of a thermocouple. With the help of a contact probe (thermocouple) an additional simultaneous measurement shows the real temperature of an object surface. Now the emissivity on the infrared thermometer will be adapted so that the temperature displayed corresponds to the value shown with the contact measurement. The contact probe should have good temperature contact and only a low heat dissipation.
Method 2: Creating a black body with a test object from the measuring material. A drilled hole (drilling depth ≤ ⅓) in thermal conducting material reacts similar to a black body with an emissivity near 1. It is necessary to aim at the ground of the drilled hole because of the optical features of the infrared device and the measuring distance . Then the emissivity can be determined.
Method 3: With a reference emissivity. A plaster or band or paint with a known emissivity, which is put onto the object surface, helps to take a reference measurement. With an emissivity thus adjusted on the infrared thermometer the temperature of the plaster, band or paint can be taken. Afterwards the temperature next to this surface spot will be taken, while simultaneously the emissivity will have to be adjusted until the same temperature is displayed as is measured beforehand on the plaster, band or paint. Now the emissivity is displayed on the device.
The AA have made use of specific “Dots” (whose Emissivity value is known) only for the measurement of some parts (the cable rods) whose materials (at much lower temperatures) were not analyzed and verified:
“Dots” of known emissivity, necessary to subsequent data acquisition, were placed in various places on the cable rods. It was not possible to perform this operation on the dummy reactor itself (and a fortiori on the E-Cat), because the temperatures attained by the reactor were much greater than those sustainable by the dots.
It was not possible to extract any samples of the material constituting the rods, as this is firmer than that of the reactor.
It should be noted that (see Figure 1) the Alumina (Al203) Spectral Emissivity values are known in literature and in particular those within the measurement bandwidth of the Optris OP160 thermal imager, this band ranging from 7.5 to 13 µm.
The Emissivity values ε = f (λ, T) are well documented in the publication by the US Department of Commerce National Bureau of Standards, Volume 7 (1971):
Precision Measurement and Calibration – Radiometry and Photometry
As shown in Figure 2A, 2B and Figure 3, for example, for temperatures ranging from 1200K to 1600K (i.e., from about 930 to 1330°C), the Spectral Emissivity measured values, within the camera measurement range, are between 0.85 and 0.95.
In addition to scientific literature, the results of the MFMP measurements show as well how the Alumina Spectral Emissivity value of their simulacrum varies greatly depending on the wavelength λ, for wavelengths close to 10 µm, i.e. in the ordinary camera range, this value is about 0.95 as confirmed by the document : Handbook of the Infrared Optical Properties of Al203. Carbon, MgO and Zr02. Volume 1 (an excerpt of which is available at this link), while the measured value of Total Emissivity (see Figure 4) to be used for calculating the power radiated, for a T around 1000°C, is less than half that value. Note how precisely in the neighborhood of these temperatures (which are probably those taken from the Hot-Cat during the test) one has minimum Emissivity, which in some measurement results to be lower than 0.3.
Furthermore, these measures show that the dispersion of the Total Emissivity real data, at the same temperature, well exceeds the error and the uncertainty that the AA are mentioning in TPR2:
The error associated with the plot’s trend has been measured at ± 0.01 for each value of emissivity: this uncertainty has been taken into account when calculating radiant energy.
In the AA report [ref. 3], R. Morrell, Handbook of properties of technical & engineering ceramics, Part 2, p. 88, in the comments for chart in Figure A4. 10, the author suggests readers should not to take that chart too literally:
“Such data should be considered tentative because it is known that emissivity can vary with grain size, porosity and surface finish through varying degrees of translucency and optical scattering.“
(In TPR2 Figure 6, Plot 1 was reportedly derived from this table).
The main issue at this point is the fact that the right way to proceed for AA would have been to use the Spectral Emissivity values for temperature measurements (for example, 0.90–0.95 similar to the values reported by MFMP) and the Total Emissivity value for the calculation of the radiated power by means of the Stefan-Boltzmann formula (typically 0.40 at high temperature). To be noted how these issues show how awkward (and in some ways not very cautious) was the choice to adopt this type of calorimetry.
The value of Emissivity to be used for the thermal imager during acquisition is much higher than that to be used for calculating the Radiated Power. If we consider an operating temperature of 1000°C, the peak spectral emission maximum (Wien peak), is approximately 2.3 µm.
Since the Alumina Total Emissivity, as also reported in Figure 6 of the Report, decreases with increasing temperature (i.e., the decrease of the peak emission wavelength), the Total Emissivity to be used for power calculation actually coincides with the value indicated by the diagram (@ 1000°C it is about 0.4), while in the case of the temperature measurement by thermal imager, even if we are observing a body at 1000°C, the value of Emissivity to be used (i.e. the Emissivity within the camera measurement Spectrum) will still be what is appropriate for the reading range of the camera used (7.5-13 μm in this case).
As already noted in other occasions, in order to reduce the Emissivity error problem, those taking measurements in foundries or in the analysis of ovens do not use ordinary cameras (as that used by the AA), but cameras with a reading window in the field 3-5 μm. In our case, the choice of a camera with a reading window equal to 7.5-13 μm for certain aspects appears to be conditioned by the fact that only in that spectral range Alumina shows an Emissivity close to 1.
As an example, Figure 5 shows the Spectral Irradiance at a temperature of 1200K.
For an immediate comparison we calculated and report:
a) the blackbody spectrum @ T=1200K (blue curve)
b) the grey body spectrum @ T=1200K, with 0.45 Total Emissivity (gray curve)
c) the Alumina spectrum (red curve), according to Al203 Spectral Emissivity data @ T=1200K, based on the National Bureau of Standards data (see Figure 2). The missing ε=f(λ) values were obtained by interpolation.
Integration of the Black Body Spectrum @ T=1200K provides the theoretical value of Irradiance (Radiant Power RP) that turns out to be 117,573 W/m2. On the other hand, if integration is carried out only within the thermal camera measurement range (7.5 – 13 μm), one gets the RP value in the measurement bandwidth. Under these conditions, as a consequence of the used values, it is evident that the Grey Body with 0.45 Emissivity (value set in accordance with the TPR2 Plot1 @ T 1200K) has an irradiation significantly lower compared to the Alumina body while being at the same 1200K temperature.
Under these conditions the Black Body RP is 9.96 kW/m2 and, taking the ratio between the Alumina RP (9.4 kW/m2) and that of the Grey Body RP (4.48 kW/m2), one get a 2.1 factor clearly indicating the presence of a large measurement error due to incorrect use of Emissivity parameters.
In the same way, Figure 6 shows the Spectral Irradiance at a temperature of 1400K. As before, for the sake of comparison we calculated and report here:
a) the blackbody spectrum @ T=1400K (blue)
b) the Grey body spectrum @ T=1400K, with 0.4 Total Emissivity (gray)
c) the Alumina spectrum (red) according to data for Al203 Spectral Emissivity @ T=1400K, coming from National Bureau of Standards (see Figure 2). Missing ε=f(λ) values were obtained by interpolation.
Similarly, integration of the Black Body Spectrum @ T=1400K provides the theoretical value of Irradiance (Radiant Power RP) equal to 217,819 W/m2. Integration limited to the camera spectral range of measurement (7.5 – 13 μm), provides the RP value in the measurement bandwidth. Under these conditions, as a consequence of the used values, it is evident that the Grey Body with 0.4 Emissivity (value set in accordance with the instructions in the Plot1 TPR2 @ T 1400K) has an irradiation significantly lower compared to the Alumina at the same temperature.
In these conditions the Black Body RP is 13kW/m2 and the ratio between the Alumina RP (12.4 kW/m2) and the Grey Body RP (5.2 kW/m2), provides a 2.38 factor indicating the presence of a measurement error due to incorrect use of the Emissivity parameters.
To be noted that in Figures 5 and 6 the diagrams on the left side show the evolution of the spectrum of Planck as a function of frequency (f). Given the characteristic of inverse proportionality between frequency and wavelength, the left part of the diagram as a function of the wavelength (λ) [the diagram on the right] corresponds to the right side of the diagram as a function of the frequency and vice versa.
Figure 7 shows the link between Radiant Power and Temperature with reference to a measurement in the spectral range 7.5 – 13 μm. The black curve reports precisely the Black body Radiant Power that acts as a reference and calibration curve for the camera.
The graph takes the Emissivity ε as parameter in order to show how it affects the estimation of actual body temperature when measured using this IR camera methodology.
For example, let us consider the 1100K T1 temperature detected by a thermal imager with a wrongly set 0.45 Emissivity (correct value 0.90).
The RP deduced from the graph of Figure 7 for ε=0.45 (about 4 kW/m2) is in fact the one that competes to the curve with ε=0.90 but for an actual temperature T2 of only 800K (temperature overestimated by 300K).
Based on these temperature values, applying the Stefan-Boltzmann relation: P=ε*σ*(T^4-Tamb^4) to calculate the thermal power per unit area radiated by a body (σ=5.67*10^-8 [W/(m2*K4)] and Tamb 300K, i.e. 27°C), we obtain:
Pmeas(T1) = ε1*σ*(1100^4-300^4) = 40.4 [kW/m2]
Preal(T2) = ε2*σ*(800^4-300^4) = 14.7 [kW/m2]
To compare through the SB the power radiated, the Total Emissivity (for these temperatures from the TPR2 Plot1 ε1=0.49 and ε2=0.65) should consider. The ratio between Pmeas and Preal is 2.7 and this result indicates a large overestimation of the thermal power radiated from the body.
Nevertheless, the error may be even higher. Consider for example a temperature T1 of 1320K (about 1050°C) as seen by the camera which has been set (incorrectly) the value of emissivity 0.42 derived from the Plot1 TPR2 instead of the correct value 0.90.
The RP can be deduced from the graph in Figure 7 for ε between 0.40 and 0.45 (about 4.8 kW/m2) is actually the one that pertains to the curve ε = 0.90 but with an actual temperature T2 of only 880K (overestimation of the temperature = 440K).
Based on these temperature values, applying again the SB equations:
Pmeas(T1) = ε1*σ*(1320^4-300^4) = 72.1 [kW/m2]
Preal(T2) = ε2*σ*(880^4-300^4) = 19.8 [kW/m2]
As in the previous example, the Total Emissivity is considered for the SB (for these temperatures from the TPR2 Plot1 ε1=0.42 and ε2=0.59). The ratio between Pmeas and Preal is 3.6 i.e. still a huge overestimation of the thermal power radiated by the body and if, under these conditions, even higher Tmeas values were considered, it is not possible exclude that the error term (which continues to rise quickly, as well as the thermal power overestimation) would be even worse.
Coming back for a while to measurements and data in the MFMP report, it is interesting to see how what we report here is not just theory. The MFMP Test case is represented in Figure 8.
The MFMP have made a change in the emissivity value, by changing the camera settings, applying the change for comparison only to certain observed areas. The value has been changed from 1 to 0.7 for zones 7, 8 and 9.
Let us consider, for example, the area 7 of the device captured by the camera. The corresponding effect on the temperature indicated by the camera is to increase the read values from 985.7°C to 1276.5°C (a change of more than 290°C).
The following Figure 9 (where the area of interest in Figure 7 has been expanded) shows good agreement between the MFMP experimental data and the calculation of the temperature error that would be expected after the erroneous setting of Emissivity, due for example to the uncertainty on the actual value to be applied.
Finally, applying the SB to this Test case and using data of Total Emissivity (as from TPR2 Plot1), one would gain an overestimation of actual radiated thermal power by a 2.6 factor.
Experimental verification on a Alumina tube
A further experimental verification of the fact that the Emissivity of the Alumina in the field of 8-14 μm is approximately 0.95 was obtained by performing a simple test that makes use of a Pyrometer (a thermal imager was not available).
As can be verified by consulting the instrument Instruction Sheet, the Fluke 80T-IR Pyrometer has a window spectral response 8-14 μm very similar to that of the camera used by the AA in TPR2 (7.5-13 μm) and the instrument is calibrated for a 0.95 Emissivity.
Since the highest operating temperature for this instrument is 260°C, it was decided to limit the alumina temperature at about 160°C. The Pyrometer analog output signal (1 mV/°C) was sent to a milliVolt meter with full scale 200mV.
According to references, the Alumina Spectral Emissivity curve is weakly affected by the temperature so that the measurement is still significant. The AA as well made their calibration (or intended calibration) at a temperature of 450°C while the measures on the Hot-Cat were performed at much higher temperatures.
Four ceramic 68 ohm resistors (10W), in parallel, were inserted in the tube, as shown in Figures I and II.
A direct current coming from a stabilized power supply fed the resistors. To ensure good heat exchange between the resistors and the alumina tube, Portland dry powder cement filled any internal cavities, as shown in Figure III.
A type K thermocouple connected to a thermometer was placed within the alumina tube, as shown in Figure IV.
As shown in Figure V, the whole system was insulated with mineral wool to reduce the thermal exchange thus keeping the temperature remarkably uniform.
In the central area, the thermal insulation was removed to allow reading of the tube surface temperature by means of the optical Pyrometer.
The power supply voltage was adjusted to ensure that the system was maintained at a temperature of about 160°C (power approximately equal to 10W); comparative readings were then performed. For greater safety, once a thermal steady state was reached, 3 readings were performed with time intervals of tens of minutes (Figures VI, VII, VIII); the three measures provided very similar results.
The insulation was not particularly accurate since the presence of a thermal dissipation, and then a thermal gradient between the internal area measured by the thermocouple and the surface measured by the thermometer, provides an Emissivity value lower than the actual one.
The temperature read by the Pyrometer is lower than that read by the thermocouple of about 5°C, difference in part attributable to the thermal flow. Neglecting the difference in temperature resulting from this flow, and considering that the instrument Emissivity is calibrated @ 0.95, the effective Emissivity of the Alumina tube is in a first approximation equal to:
ε = 0.95*[(160 + 273)]^4 / [(165 + 273)]^4 = 0.907
At this temperature, using the TPR2 Emissivity values (0.70 at 160 °C), the Pyrometer read error would be around 34°C.
The MFMP experimental data are in agreement with those reported in the literature and confirm that the procedure and the Emissivity values, used by the TPR2 AA for measurements by the thermal imager, are incorrect. The GSVIT experimental test further showed that the pure Alumina Spectral Emissivity, in the reading field of the camera used to testing the Hot-Cat, is greater than 0.90. These data are very different from those plotted and used in the TPR2 by the AA that appear to be those related to Alumina Total Emissivity. In the 1200-1400°C temperature range, the TPR2 Plot1 considers an emissivity of about 0.40 while, according to the literature, the Spectral Emissivity, in the camera reading field, is stable around values close to 0.95. This kind of error can lead to a significant overestimation of the surface temperature and to an overestimation of thermal Power by a factor 2 or more. An error of such proportions (which appears likely in the light of the measurements) makes not reliable, in our opinion, the TPR2 measurement results of the heat produced by the Hot-Cat; on the contrary, a simple Mass Flow Calorimetry, similar to the one shown in a previous Post of ours, would have been feasible and most accurate.