Calibration of Stainless Cell and Air Flow Calorimeter
We have completed two initial calibration runs on the insulated stainless cell within the air flow calorimeter (AFC). The goal is to attempt to calibrate and characterize both the stainless cell and the AFC simultaneously. The cell looks like this installed. Here is the link to the Active Wires in the Chambers & Insulated Cell Details blog post.
The whole air flow calorimeter has been set into a cabinet of sorts to try to eliminate the effects of cooler drafts on the exterior of the unit and to isolate it slightly from the temperature variations in the room. The box is open on the top to allow for heat to escape and not build up.
The set up works for drafts, but this room still changes about 1.5 degrees over a day and the cabinet changes with it. We may need to actively control to a higher temp. Below you can see the exterior of the cabinet and the laptop monitoring the experiment from the shelf next to it.
Calibration Cycles
We did 2 calibration cycles so far.
Time constants, settling times
These runs were done with 3 hour steps rising and falling at 4 power levels. It is pretty obvious from the data that the cell has a very long time constant. On a separate, single heating to the same high power level as the calibrations, we had the resulting curve indicating a settling time of approx 5 hours. Interestingly, it also shows the the top interior sensor runs cooler- presumable because of the thinner insulation around the top of the flange and the plumbing and pass through tubes sticking through the insulation.
Resulting curves
Ratios
Are the ratios of the temperature rises consistent? Are they good indicators of constant thermal resistance? The ratio across the insulation dropped at higher temperatures. This is not unusual for most insulation materials. The ratio between the Macor and the inside wall was very constant, which would seem to indicate that the thermal resistance across the hydrogen was consistent.
Degrees/watt
8 to 10 on t_mica! The thermal noise should be minimal, too with the insulation. That means we should have good resolution for future wire tests. If we ever see a wire with 15 to 20 watts output, 2 meters could be self sustaining at 2 watts/meter, 15 to 20 meters could be self sustaining!
Non uniformity of shell
As I noted above, the top interior sensor was cooler than the middle or bottom. Meanwhile, the middle exterior sensor runs the hottest while the bottom exterior runs cooler and it's not entirely clear why. The result is that this form of calorimeter is not gonna be easily approximated based on a pure physics calculation, so we’ll have to be content with an empirical fit. We will also have to watch carefully for mischievous effects caused by slight temperature changes altering the heat flow out through the gas tube and the pass through tubes. That is one reason we are running this cell in the Air Flow Calorimeter, which will at least provide a a nice, stable exterior temperature.
Offset in the middle of run
If you look closely at the data you will see that on both runs at almost the same exact time of day, the exterior temperature slowly rises a small, but noticeable amount. This was caused by the AFC changing air flow rate because the fan was set to right near minimum speed, just barely turning. Our hypothesis is that the fan changes mode slightly and reduced the air flow. We since turned up the fan and just added another filter to slow and even out the air flow back to a speed where it will rise 1 degree C for every 10 watts. Since we made that change, we have not seen another offset like that one. We will have to try to run a more precise calibration again now that the offset seems to be cured.
SS Cell Calibration Data
All in one Zip - including 1 minute raw data, and spreadsheets.
AFC Calibration
I was hoping to include a graph of the AFC output vs SS Cell input power, but the data from the AFC is still not ready. We achieved a rough calibration, but it is proving inaccurate (off by 1.6 watts at 37 W input power) after we changed the fan speed and added an additional filter. When we re-run a more thorough calibration, we will need to zero all the sensors appropriately with offsets while everything is cold and settled. If the calibration doesn't tighten down more after a more careful calibration, at least it makes a nice controlled temperature environment for the SS Cell.
Air Flow Calorimeter Document
We are in the process of making a nice reference document to explain the design and principles of the Air Flow Calorimeter. It is a draft in process, so don't expect perfection, but I wouldn't mind your questions or suggestions to help make it better.
You can find it here: HUG Air Flow Calorimeter
Next steps:
Besides slowly testing this set up, we are working on Version 1.3 of the glass, Celani type cell, while we continue to watch the vertical cells. Pictures and more details soon.
Comments
Your questions are excellent ones. The question of whether there is excess heat during the calibration run gets at a basic problem with this experimental setup, which some researchers have identified in Celani's case -- an important step is to do the calibration with wires that are unlikely to generate excess heat (Ed Storms mentions Pt as a possibility), without the presence of wire (Celani or otherwise) that could end up producing excess heat. But when you open up the cell to change the wire, you mess up the calibration constant.
A catch-22, and hence a flaw in the experimental design. Assuming there is indeed excess heat and the P_xs does not go back to experimental artifact relating to the gradual pressure drop, I think it's possible to get a good sense that it exists. But in this case there will be lingering doubts that make the experiment less than incontrovertibl e.
P_xs is derived from some calculations based on calibration baselines, right? So when the maths shows the output power to be greater than what was established during the calibration then we show a positive P_xs.
However taking a closer look at the data available it appears T_glassout is fairly stable for the duration of the experiment. If the Celani cell is producing excess heat should we not see a higher delta for T_glassout?
It has been a long time since any transitions have been available, but the input from Ascoli got my attention. I ran a dummy transient and let the program calculate the best match for the power input as if it were a calibration.
I used the latest data for the last 12 hours at 30 second intervals. I get a matched power of 105.6 watts with peaks reaching the input that should be available of 106.4 watts.
I used the calibration values obtained during the special run for this experiment. I would consider this a null excess power event. I have been using the outer glass - ambient for the temperature delta. This does not support excess power generation.
1) The pressure sensors are measuring relative to ambient, not absolute pressure. Not possible since the cell curves don't reflect ambient pressure changes.
2) The pressure sensors are degrading identically from long exposure to hydrogen. Possible?
3) Identical ionic membrane diffusion through two different glasses. Possible but very unlikely.
4) Identical bulk absorption into two different cell structures. Also unlikely.
5) Some other unidirectional chemical mechanism binding the hydrogen to some solid in the cell.
6 New science.
I'm sure I missed a few.
In the Extra cell there's no pressure sensor but if we assume the pressure is dropping there too, and there's no active wire in it then the graph of it's temperature rising over the last 5 days might be a clue.
Looks like you're adding just enough H to keep the pressure constant, excellent plan!
The image below provides an update of the power imbalance at cell 1.1 in the period from January 25 to February 5, corrected on the base of the gas density inside the cell. As you can see, the corrected imbalance remains close to zero.
It is to be noted that the formula and the parameters used for the correction were derived from the calibration cycles performed between 4 and 8 January, when the molecular density decreased from 3.1 to 2.7 mbar/K. The present much lower value of density, about 2 mbar/K, may affect this correction.
It would be useful to replay, at the end of this period run at constant power, a few calibration cycles, varying each time the initial (cold) pressure from 0.5 to 2.0 bar (or vice versa), with 0.5 bar steps (4 cycles in total). In this way, one could derive a more precise correction formula to be applied to the results obtained until now.
Image: i.imgur.com/UAHtfNH.jpg
It is important that nothing be done to the system as it sits except for the calibration and step sequence. Regardless of the pressure behavior, it should not be modified since this is the exact conditions under which we suspect excess power of this level.
If you vacuum the gas, or add more, then it leaves open the possibility that your changes have modified the device. And do not change any of the sensors since that also adds questions. We need to test the exact same system that you believe is working well at this time without any changes.
I sincerely believe that my procedures and program have the power to verify the performance of the cells provided that power is not nullified by incidental modifications. I will gladly answer any questions you may have.
The recent observation that the excess power has slowly creeped up over a long period of time is suggestive of a drift in the test set up. Of course, it is also possible that this is due to a delayed Celani process coming up to power. That is why it will be quite informative to see the time domain shape of this transistion if it happens. A skeptic position is that the calibration will be found to be different, the excess power will appear to be null, and the curve will follow a perfect path in time according to the differential equation solution.
If the above things happen and a good quadratic fit exists, then I would doubt the existence of significant excess power.
My procedure to obtain calibration factors has generated a very good quadratic fit to the observed data during that process. Any heat generation mechanism that produces 5 watts at maximum would stand out far above the noise and there is essentially no way that the curve would fit to a reasonable degree.
Any normal calibration drift that is not extreme should be easy to determine by the poor quadratic R^2. I am hoping that calibration parameters will be found that are close to the original values. (Continued)
GET READY!
EeeeeyYiYi.. lol
Go slowly!!! Over hours if possible.
Are we considering that as the internal cell temp goes up, the gas pressure increases over 1 atm, then gas leaks... this lowers the density of the gas and so potentially its ability to conduct power away from the wire to the glass - but the wire is close to the Macor...
What we need is to remove these debates about gas and the effect of the environment...
GET READY!
First we charge a length of pipe with hydrogen, then we use a series of needle valves and quarter-turn plug valves, to control the flow into the reactors. We can usually get a small steps of 0.01 bar or so with this method.
@ 123star
As to how many sensors have the aluminum patch, there is only one. It is on cell 1.1 as the T_GlassOut sensor. We are also using a white thermal paste between the aluminum and glass. The rest of the exterior glass sensors are fixed with Kapton tape.
I propose yet another (simple) test to check if IR absorption matters. I know that all current temperature probes on the glass surface are attached to an aluminum patch (correct?). Aluminum is a good reflector so IR absorption is reduced greatly.
I propose to attach another temperature probe on the glass with ugly, BLACK, electrical tape (so to say) which is presumably opaque at near IR and see if:
-- The absolute reading of the "black tape probe" is higher by a certain offset (I suspect so)
-- The reading of the "black tape probe" follows the T_Macor and internal temperatures, unlike what happens now (as shown by Ecco). This effect, if present, should be imputable to IR absorption characteristics .
Despite this, I am open to the idea of re-upping the pressure back to roughly 1.4 Bar like it was at the peak of the calibrations. My first preference would be to do it warm, though, so we minimally disturb the system.
Your graphs do beg the question, though, if there is excess heat and it isn't leaving through the glass, is it leaving somewhere else? How could it not be warming up the glass exterior?
A visual inspection of the cells just minutes ago shows the cells to be in identical condition to when we started the test. The glass has no visible contaminants. The wires appear to be in the same condition, and so do the supports.
Is there some other sort of "drift" as Dave Roberson mentions that can cause this? I like the idea of testing the warm up time again, but I am not eager to do anything to "extinguish" the small, new fire we may be seeing.
@ David Roberson - I like the premise of your test with the exception that the assumption of no excess power during the warm up step. How do we tell the difference between a change in the thermal performance of the cell and excess heat production during the warm up?
@ Rob B - The pressure drop is almost certainly a leak. We have not been able to "sniff" a leak anywhere, so they must be really small. The Cell pressure appears to be right at atmospheric at the moment. The interesting thing is that we have seen these cells drop below atmospheric before. We'll see what it does going forward. I wonder if tiny amounts of air are getting in and the oxygen is playing a role somehow. Or is it changing the gas composition? Current barometric pressure is 29.88 inches of mercury = 1.0117368 bar.
@ Dieter - The HUGnetLab data acquisition board measures the temperature of the junctions at the terminal strip with an RTD. That temperature is then added to the thermocouple temperature reading as an offset. It is all done in Celsius.
The building sensors are run on an older version of the system. The outside temperatures were sensed with thermistors on the older system. Even that one is collected in C and just displayed in F.
Just trying to understand a bit more about your prediction curves.
Would the calibration factors remain the same with or without excess heat? As of now, we are not quite sure whether there is an exact trigger temp/power/curr ent. There could be excess heat at all power levels that could throw off your predicted equilibrium, correct?
If we performed another calibration as before and we see a different set of factors, would they say that we are not producing excess heat or could it say that we are producing excess heat?
Likewise, if we see the same factors (within tolerance), would that mean that we are not seeing excess heat or would it mean that we are seeing excess heat?
There's lots of good data showing trends over time without changes other than pressure drop. The drop seems well controlled. Is the speculation that there is a leak in the cell?
In this situation I am playing the role of skeptic since it is important that we are correct if we announce that Celani has been replicated.
The proof will be revealed if the cell is allowed to cool down without changing it in any way, and then another calibration performed as before. If all calibration factors are close to those determined the last time, then it is almost certain that excess power is being generated. Without this step, I can not be confident that excess power is generated.
It is important not to change any other factors such as the gas pressure.
RSS feed for comments to this post