The big questions are: Is the excess energy real?? And how much is real??
In short, YES , we are seeing excess energy as far as we can tell. There is still a chance that we are fooling ourselves, but that chance is getting smaller and smaller as we rule out potential error after error.
The amount of energy is AT LEAST 5 WATTS. We have reason to believe it may be more. Below is an explanation of how we arrived at the baselines we did.
This blog entry took a while to put together. It is worth noting that while our team was focused on working out some technical details to make the experiment happen and then focused on pulling the data together for ourselves and for Celani, our awesome core of followers has been providing us with as many insights as we could have hoped to come up with from weeks of our own work. The power of the crowd is working!
Below is the graph of all the calibration runs on the Euro Cell in terms of temperature rise for each power.
The first run was way below the others. The last tests, with the active wire in helium, fell close to this lowest curve.
Why are the other calibrations higher than the first?
Possibility 1: The oxide coated control wire started generating excess heat after taking on some loading during the first calibration. This is Celani's inclination. If this is the case, we can safely say that we are demonstrating approximately 5 watts more output energy than the oxide coated constantan wire did.
Possibility 2: T_Glass Out sensor was less thermally connected to glass than in later runs.
From all these calibration lines, we decided to use the highest as the most conservative number, and the lowest because that was closest to how the installed Celani Wire was working under Helium with the exact same T_glassout sensor location. See graph below.
These are the lines we fit the formulas for the P_Out from. Here are the curve fit data and resulting equation parameters.
This is another view of the same thing. Mathieu prepared this for Celani to include in a presentation in Rome this morning.
When we look at the equivalent power to achieve that Delta T_out according to each base line, we get the following:
This graph was provided by one of the commentators and shows that the total amount of radiated energy from the glass is only about 31 Watts. While it does not account for all the 48 W of input power to the cell, it is still an increase of energy output from before the very same wire was loaded with hydrogen. The remaining energy output is in the form of convection.
From Nic:
We still have a low and high estimate based on the calibration baseline we take as reference.
Low is Calib CuNi44 H2 1bar
High is Calib 360L He 1bar
Then the SB calculation gives us a Cell Coefficient (CC)
CC= Correction_Factor * BlackBodySurface * Emissivity * SB_Constant
Low: CC = 3.740958E-09
High: CC = 4.514090E-09
I took Emissivity = 1, since the Correction_Factor (CF) is modifying it anyway.
Low: CF = 1.500
High: CF = 1.810
Finally the Output Power is:
P_SB_out = CC * (T_glassout^4 - T_ambient^4)
This model work very well as the errors between the model and the calibration curves are very small. The calibration data fits perfectly with the SB model, once the correct factor is estimated correctly.
As we struggle with which baseline is really appropriate, perhaps we should extrapolate the pre-loading performance of the actual Celani Wire in which the wire location, emissivity, and sensor location are all exactly the same as during the loading and following live run. Below is an illustration of this approach.
.
According to this baseline, we are getting 64.9 Watts (64.9 - 48.0 = 16.9 Watts excess).
One more piece of data to contend with - Interestingly, the graph of T_Glassin does not show the first calibration run to be that much below the others. I (Ryan) believe this is a hint that the first run was lower because of a variation in thermal contact between the thermocouple and the glass.
The graph below is meant to illustrate how the temperatures from each run were interactive with the pressures and gas types. Each of these lines represents one of the calibration runs The power and temperature started out low. As the power stepped up, the temperature rose, which caused the gas pressure to rise also. The runs starting at higher pressure showed a larger pressure rise in bars (following the ideal gas law). The one perfectly vertical line was adjusted at each step to hold a constant pressure.
The effect we demonstrated on Cell 1 in the USA is not applicable to the current run in Cell 2 because the pressure is very close to constant at just over 1 bar.
We look forward to more advice and analysis by the many, many sharp individuals out there. We also look forward to more design suggestions for how to do the experiment in better ways. Similarly, if anyone else is interested in trying the experiment for themselves or at their institutions, let us know. Facilitating research into the New Fire is our goal.
Addendum: Mathieu has put together this nice summary of the early results in a PDF document:
https://docs.google.com/open?id=0B9qCtGOFmvhmeFF2ZzNhX3JXUTQ
Update #1 - Calibration Basis Overview
This talk through was given in Rome to help delegates at the Coherence meeting understand the data coming from the EU cell.
Correction: Bob says "same source" that should be "same type"
Comments
Check the file in the bottom of this article.
quantumheat.org/.../...
You have all the calculation for the borosilicate glass.
You can change to cell temperature in the Excel sheet.
Here is the link to the file to download (Ctrl + S):
docs.google.com/.../
1- Ok I got the graph wrong before. But I did some calculations:
the IR radiated portion for a blackbody at 600°C from 750 nm to 2500 nm (IR range for which the borosilicate is transparent) is
T= 650 °C --> IR power (range 750 nm- 2500 nm) = 12%
T= 900 °C --> IR power (range 750 nm -2500 nm) = 26%
So a consistent part of the radiated power still escapes.
Anybody please check this calculation
About the graphs you posted, I see your point, at some low pressures the T_glassout decreases (especially with Helium). But the general trend is that at high pressures (>3 bar) the temperature drops. This is consistent with Gipsel interpretation: compressed gas cools the wire by convection and hinders IR radiation. Then the thermocouple reads a lower value because it sensitive to IR radiation. By the way, the pressure dependence is not related to emissivity at all, it's just cooling by convection.
The last observation I make is that for the EU cell, when power_in was raised from 48 W to 54 W or so, the P_Xs also did not increase much, not until the resistivity of the wire began to become unusual and funky. This too is contrary to the emissivity hypothesis, as significantly higher internal temperatures did not result in a significantly altered (over basline) T_GlassOut reading (until, again, resistance began to get weird).
There is just too much handwaving and too many holes in the idea it can account for a significant portion of anything. Ultimately, I really like 123Star's idea for experiments to quantitate the effect if it exists. If it does, and you were right and supported by direct data, then the quantitation would allow us to factor it out and see if any excess was left over.
Finally, look at these two graphs: quantumheat.org/.../... and quantumheat.org/.../...
Not only do the show the behavior I'm talking about where there's an optimal pressure, but even worst for the emissive hypothesis, the higher 110 W show even less of an effect! That is absolutely opposite what your hypothesis claims. Here a hotter wire with more emissivity is giving less of an effect over a cooler wire, and this effect falls faster with pressure drops than the cooler wire (less of a hump too).
The fact the P_xs did -not- increase with increasing power in directly falsifies and torpedoes the emissivity hypothesis. Unless there is an alternate explanation.
So, the data continues to leave me unconvinced of this emission hypothesis.
Take a look at this graph, quantumheat.org/.../...
Look at the calibration curves, not the experimental curves (which yours is), as the experimental curve may have been showing actual production, it's impossible to know but obfuscates the discussion. If you look at those shapes, you'll see exactly what I mean.
Here's Ecco's picture for borosilicate transmission of IR i.imgur.com/EIZzK.png . Anything longer than 4 um is fully absorbed at our thickness. Ecco also posted a picture for quartz somewhere, but it's really buried, and is quite different than borosilicate.
quantumheat.org/.../...
Quoting Ged:That's not true neither. Borosilicate has some partial transparancy below 3.5µm and is basically fully transparent below 2.8µm. With quartz the partial transparency starts at about 4µm, and gets fully transparent (for 3-4mm thickness) below about 3.5µm to 3µm depending on the quality. Quartz just lets through a bit more at lower wire temperatures already. That's all.
Remember, the US cell is transparent to IR. IR is already making it fully through the cell. Or put another way, the thermalcoupler is already seeing the full IR radiation of the cell at all times. That is lost energy from the system that is not detected, but became detected on a very narrow band of pressure in accordance with particular types of gas. And this effect did not appear to be power-in sensitive (which your hypothesis necessitates it would be).
That is in direct opposition to the emissivity hypothesis. Then you have to factor in the magnitude of the effect in the EU cell (the magnitude of the worst case in the US cell was still much smaller than what the pressure steady EU cell saw), and the fact that the NiCr wire also saw a small anomaly of excess heat over the highest baseline -only- when the Celani wire was loaded with hydrogen, and not during the calibrations in the presence of the naive wire (or helium).
But that idea isn't actually supported, because as the gas pressure falls beneath 2 Bar, the outer heat also falls. The raise only peaks at around 2-2.5 Bar and then falls again. This is directly contrary to your hypothesis, but is in line with the hypothesis that the quartz was changed in absorption characteristics.
Furthermore, that effect happened most in Hydorgen/Argon mixes, and least in Helium. Yet all these mixes are not IR radiators and not IR absorbers, and should have had no impact on the IR. And yet, they had extreme impacts on the curve shape.
This is again contrary to that hypothesis about emissivity, but completely in line with the hypothesis that the quartz absorption bands were changed by the particular gas. And indeed, quartz does have funny absorption bands as was posted by Ecco I believe.
So, the evidence as far as I see does not support your hypothesis, but supports the null, as far as I see.
Actually, it showed that with lower pressure the measured glass temperature rised. Lower pressure means less cooling of the wire by convection. The wire gets hotter up to the point where radiation emission balance the constant heating power. So we have more radiation at a shorter wavelength at lower pressures. And the measured glass out temperature rises. This is exactly the effect 123star and I mean. The US cell proved to be really susceptible to this, the EU cell should be a bit better (but still vulnerable to it). The thermal contact issue is an additional source of error. If both play a role, it's hard to say what happens exactly.
docs.google.com/.../...
We have all the data for 4 cell temperatures. One inside the thermal well in the middle, one on the mica wire support, one on the inside of the glass (held by spring pressure of the thermocouple) and one on the outside of the the glass held in place by kapton tape.
We looked at the variation of each temperature in different gasses at different pressures. The temperatures in the thermal well and the mica were greatly affected by gas and pressure. The external glass temperature minus the ambient was more independent. The difference across the glass was explored, but showed all sorts of unexpected deviation. See the blog post
quantumheat.org/.../...
I agree we should chat directly about this and future options, but I wanted to share some of the methodology with anyone else wondering.
I wasn't addressing your points, so I'm afraid I do not know what you are referring to (nor do you make it clear in your post, so I cannot respond :( ).
@Gispel, 123Star,
I really like your test idea, 123Star. In the end, empirical evidence trumps all things. Though, Gispel and I were referring to the shorter end of the wavelength band above where the mean emissivity wavelength would be at the temperatures we have (the main power band is absorbed by the glass); so that end is stunted as your graph beautifully shows compared to the mean point you choose. The longer wavelength tail should be absorbed by the Borosilicate.
The thing is, the US cell showed the opposite of this; it showed that when pressures/compo sition changed the glass became an absorber of IR and that is what gave the measurements of slight excess, not the other way around. So, actual experiments are the only way to address this. That's science
The method requires measuring the ∆T across a STABLE thermal barrier. Nothing else matters. This ∆T MUST represent the average over the entire thermal barrier. Since the barrier in this case is glass, some energy will be lost by radiation, but this is small compared to the claimed amount of excess power being reported. You need to address the biggest error first. The biggest error appears to be in the calibration.
When we calibrated our cells, we did it by stepping up the temperatures. We watched the temperatures over time and decided that 45 minutes was sufficient for it to not be rising anymore. Additionally, the US cell went through a 16 step curve, which would make the difference from level to level even smaller. Perhaps it wasn't adequate, though. I would certainly be willing to do a cycle up and back down to see how much difference there would be. We'll set up and do that ASAP. We are using the new pyrex glass cell. The glass is 3.2 mm thick. After this calibration, we can compare the performance of this cell to the previous generation.
@Ascoli, your graph showing the variation in T_Glassin -T_Glassout seems to make it very clear that we have thermal connection issues between the glass and the thermocouples. Thank you for that. We are working on reprising the data from Cell #1 in that format.
There's a way to test if the thermocouple is sensitive to a shift in the IR spectrum.
Let's choose two (or more if possible) wires of the same material (e.g. NiCr) with very different radii and inject in them the same power (e.g. always 48W).
The thinner wire will get hotter because it has less surface area to radiate.
I predict that the thinner wire will give T_glassout temperatures above the thickest one(*).
So Ged, if I understand correctly you are saying that T_glassout will be the same in the two cases. Right? Or are there other effects you can predict?
@Ged
I understand what you're saying but notice that the intensity spectrum has a "long tail" on longer wavelenghts.
(*) Absent thermal contact instability
As I said: it depends on subtle differences how large and even in which direction this effect works (with luck it could be canceling out to a large extent). This is not easy to assess, even if you have hands-on time with the device. But everyone having worked with thermocouples (or comparable temperature measurements) can assure you, that thermal contact is very important and one can easily be several degrees (even tens of degrees) off from very minor changes. And having a semitransparent surface with a relatively bad heat conductivity isn't exactly going to help here.
Btw., heat conduction is not required for the function of a thermocouple. It does not care how the junction gets to its temperature. And the cross section is also largely irrelevant for the final equilibrium temperature caused or influenced by radiation heating. It's not that the sensor has to block some significant amount of radiation to be influenced (i.e. to deliver biased data).
I understand and respect what you're saying, but I don't find that a convincing argument. The US cell was far more vulnerable to that effect, and yet did not show excess power anywhere near what the EU cell does. Moreover, we're talking about a very small spillover of the IR spectrum, and assuming a thermalcoupler will magically be better at IR absorption than the conduction it's built for. I don't buy it.
The thermalcoupler is not that loose or we'd unequivocally know; there's not that much power in the limited amount of IR that might leak out of the cell; the coupler's cross sectional area is miniscule; you're assuming these wavelengths don't add to the base glass temperatures at lower power in when they are longer and fully absorbed, which I find false (that'd be in our calibration curve shape); and finally the US cell did not show this behavior though it should be far more likely to.
Just wanted to say that I'm very happy Doctor Edmund Storms decided to join our quest. To my mind his presence gives the project even higher credentials so I'm very happy the project managed to attract such a respected scientist.
I feel very positive about his remarks as we need some really critical minds examining the work done here. Some of us may feel a bit offended by his comments, but that will be nothing compared to what will happen once the reactors and documentation will be sent out in the open. We need some serious critics on our methods from some keen minds to stand a chance of surviving the scientific community when this get's reviewed.
We can expect really harsh critics once the reactors are send out, so let's all be open to Doctor Storms very constructive contribution.
Also, big thanks to both the Europe and the US team and all the contributors. You're doing a great job!
RSS feed for comments to this post