The blower fan speed at whatever constant set voltage will be same, for all intents and purposes, so the outlet air velocity will be the same at a constant set outlet configuration.
Wasn't there a problem before that the air speed varied from one part of the orifice to another? I assume the air was not well mixed.
Wasn't there a problem before that the air speed varied from one part of the orifice to another? I assume the air was not well mixed.
The air velocity profile was uneven at the outlet of a plain 20-25 cm cylinder. I don’t think that should be conflated with how well mixed the air is, at least as far as the temperature goes.
The blower fan body outlet is rectangular and has forward (toward the outlet) facing fan blades at one side and the fan blades are turning sideways towards the opposite side, resulting in high pressure air concentrated on one side of the blower fan body outlet, which then continues down the cardboard outlet cylinder with higher pressure air concentrating along one quadrant as it mixes with slower air around it. Variations of 2 m/s from one location to another were common. 20 cm is a very short length in which the uneven air velocity profile can equalize. Several outlet cylinder lengths were tested, with different variations of outlet velocity profiles. None were particularly flat.
Reducing the fan voltage allowed a drop in the air outlet velocity profile unevenness. The B97 fan is rated at 12 V, but Mizuno used about 8.5 V and I am currently using 9.0 V. The reduced voltage reduces fan speed, which reduces the maximum outlet velocity, which reduces the overall velocity range of the blower outlet. It is possible that the blower fan voltage can be tuned to obtain an ideal air flow rate for the inlet and outlet sizes (preventing a noticeable vacuum in the calorimeter box, or too fast air from the outlet, reduce the velocity unevenness as much as possible, etc.) Once determined, this fan setting should be left alone. The ideal fan speed is not exact at this level of precision.
I will also note that most write-ups claim a 5 cm diameter air inlet hole as well as the 5 cm air outlet hole where the blower fan attaches, but photos all seem to show a 10 cm diameter air inlet, integral with a cover blocking a much larger access hole (~ 10 x 20 cm). Tubing and wires enter the air inlet hole. A 5 cm diameter air inlet is restrictive, noticeably changing the fan speed and outlet velocity compared to a larger outlet.
The 5 cm outlet hole does not seem to be restrictive to the roughly 6 cm diameter blower fan inlet, as long as the calorimeter air inlet is not restrictive. Widening the outlet hole to 6 cm made no difference to the fan speed or outlet air velocity, and requires more diligence in sealing the outlet to the blower fan.
I eventually settled on a 75 mm diameter air inlet tube to ensure a constant air inlet geometry, and which helped reduce (somewhat) the internal temperature cycling range and the short-term inlet temperature variations. This also prevents heated air from the calorimeter from affecting the inlet thermocouples. Wiring enters at another point, which is sealed.
The air velocity profile was uneven at the outlet of a plain 20-25 cm cylinder. I don’t think that should be conflated with how well mixed the air is, at least as far as the temperature goes.
Surely an uneven air flow is caused by unmixed air? It means there are streamlines. If it were well mixed, how could there be velocity differences?
Reducing the fan voltage allowed a drop in the air outlet velocity profile unevenness.
That is the opposite of what I would expect. A higher velocity should increase turbulence. Anyway, I can't argue with actual results.
You ended up with this "muffler" design, right? It is different from Mizuno's arrangement. As long as it works, I don't see why that matters.
MR4-3 update
After 16 hours at 40 watts, the cell has been stable at within 1 degree of calibration (106°C). Pressure is also stable with no sign of out-gassing or loading.
Power has now been increased to 80 watts. The live stream continues at https://www.youtube.com/watch?v=6NXCxsrwHOE
Surely an uneven air flow is caused by unmixed air? It means there are streamlines. If it were well mixed, how could there be velocity differences?
That is the opposite of what I would expect. A higher velocity should increase turbulence. Anyway, I can't argue with actual results.
The air is mixed thoroughly when it goes through fan. It is incredibly turbulent as it passes through. That mixed air is then ejected from the fan in an unequal cross section. The fan is quickly brought up to the temperature of the air going through it. If air mixing is a worry, then dealing with how it enters the blower fan should be where to focus some thought. I tried many ways to even out the outlet velocity profile, so that eventually a velocity sensor could be installed that wasn’t incredibly position-sensitive.
You ended up with this "muffler" design, right? It is different from Mizuno's arrangement. As long as it works, I don't see why that matters.
Yes, that design in that photo worked the best at levelling the outlet velocity profile. But it was big and stored a fair bit of heat.
The newer design is much smaller, is made entirely of cardboard and foam, but is not quite as effective at levelling out the velocity profile as the 5 inch diameter metal design. The new design is much better than a straight tube. It is essentially the same design as the one in the photo, just smaller.
I also tried a vane anemometer. The problem is that the K factor (air flow self- blockage factor) of the vane anemometer of about the same diameter as the outlet tube was over 0.5 . That means that the vane anemometer, attached directly to the outlet, resulted in reported velocity of double the true air velocity. The restriction also changes the final airflow, defeating the measurement unless the vane anemometer is made a permanent part of the assembly. Moving the vane anemometer away from the outlet reduces the K factor, but an arbitrary distance from the outlet results in an arbitrary K factor without extensive further research. This means one would need to know the outlet velocity average in order to correctly position the vane anemometer, so it is not an independent measurement.
EDIT: A vane anemometer with a measurement area significantly less than the outlet diameter should work OK.
The easiest way to determine the outlet air velocity is to heavily insulate the box and fan, and calculate the air velocity required to make the measured Delta Temperature, using a moderate to low input power level. Insulated effectively, the steady state losses should be close to zero, and the velocity necessary to balance the HVAC equations should be very close to the true outlet air velocity.
The easiest way to determine the outlet air velocity is to heavily insulate the box and fan, and calculate the air velocity required to make the measured Delta Temperature, using a moderate to low input power level.
That's upside-down! You are supposed to determine the heat from the air velocity.
I guess if it works . . .
Photos of new muffler design and construction.
.
The gap is about 2X the diameter of the smaller tube. The diameter of the larger tube seems to be best at least twice the small tube diameter. (This one isn't quite there, but finding appropriate materials in the right size is a challenge.)
.
Assembled muffler (air velocity polarizer).
.
Fan cover insulation installed. Just drops on, not airtight.
.
That's upside-down! You are supposed to determine the heat from the air velocity.
I guess if it works . . .
It does seem backwards, but if everything else is stable and efficient, it (at the least) sets an air velocity limit that can be compared to.
It only works well if recovery is nearly/effectively 100%.
Okay. For the upcoming lecture, I summarized this test as follows. What do you think? Do you want to change or correct this?
One of the people trying to replicate this experiment was unable to produce a uniform flow of air. His traverse test kept coming up with very different airspeeds at different locations across the face of the orifice – as much as 2 m/s. Yet the blower was the same model, and the tube looked the same to me. I do not know why it did not work.
Some people concluded that because this other calorimeter was not working correctly, and because it looked the same, Mizuno’s calorimeter might not be working. This turns the purpose of a replication upside down. You have to start with the same prosaic, known conditions, then create the anomalous conditions and see whether you observe the same anomalous results. If you do not get the prosaic starting conditions to work, that is not a replication. Not yet.
To be sure, if Mizuno had not carefully checked the air velocity to be sure it was uniform, with both the traverse test and the vane anemometer, this other calorimeter would have called Mizuno’s results into question. In a useful way: it would tell us where to look for an error.
The person doing this experiment ended up devising a completely different kind of exit tube, with a muffler on it. [SLIDE SHOWING NEW MUFFLER] This part of his calorimeter was quite different from Mizuno’s. Does that mean it is not a replication? In my opinion, no. This is not a test of fluid dynamics. As long as both exit tubes work, it does not matter that they are different. I still wonder why his first tube did not work, but this is not of germane to the claim that is being tested, which is that the nickel mesh with palladium produces excess heat.
Edit: I was writing this before Jeds post above... so not related...
The point is, once the air velocity is worked out in a particular configuration it is basically a constant if the airflow circuit is left alone.
You set the blower fan at X voltage = x RPM and it pretty much does the exact same thing each time.
The blower fan voltage must be fairly stable to get good control of the fan RPM and therefore air velocity and volume, so logging it (and current) is a good idea. This electrical part is probably a much greater source of errors than the physical airflow geometry. For example a one volt difference at the fan translates to a huge RPM/airflow difference, which shows up as a commensurate delta T difference. (Lower RPM = larger delta T, and the Fan Laws indicate that generally a 10% drop in RPM = 10% drop in air flow). With use of a dedicated, good quality power supply for the blower there should be no problems.
Display MoreOkay. For the upcoming lecture, I summarized this test as follows. What do you think? Do you want to change or correct this?
One of the people trying to replicate this experiment was unable to produce a uniform flow of air. His traverse test kept coming up with very different airspeeds at different locations across the face of the orifice – as much as 2 m/s. Yet the blower was the same model, and the tube looked the same to me. I do not know why it did not work.
Some people concluded that because this other calorimeter was not working correctly, and because it looked the same, Mizuno’s calorimeter might not be working. This turns the purpose of a replication upside down. You have to start with the same prosaic, known conditions, then create the anomalous conditions and see whether you observe the same anomalous results. If you do not get the prosaic starting conditions to work, that is not a replication. Not yet.
To be sure, if Mizuno had not carefully checked the air velocity to be sure it was uniform, with both the traverse test and the vane anemometer, this other calorimeter would have called Mizuno’s results into question. In a useful way: it would tell us where to look for an error.
The person doing this experiment ended up devising a completely different kind of exit tube, with a muffler on it. [SLIDE SHOWING NEW MUFFLER] This part of his calorimeter was quite different from Mizuno’s. Does that mean it is not a replication? In my opinion, no. This is not a test of fluid dynamics. As long as both exit tubes work, it does not matter that they are different. I still wonder why his first tube did not work, but this is not of germane to the claim that is being tested, which is that the nickel mesh with palladium produces excess heat.
What you might say, might be along the lines of of “Although it might call into question the absolute levels of power and energies reported to some degree, by challenging the absolute calorimeter recovery efficiency reported, it does not affect the strength of the relative power levels claimed, because any possible corrections are a small and constant factor of both the calibration and the active reactor measurements.”
MR4.3 Update
During 24 hours at 80 watts, the cell mostly matched calibration temp within ±1°C. An increase of just under 2°C can be seen from 16:50 to 17:50, perhaps worth a closer look. The pressure declined gradually from 384 to 377 Pa. No unusual neutron or gamma signal was seen.
RGA analysis showed only D2 and DH. The traces of H2O and N2 are the residuals from the analyzer sampling tube and chamber.
The heater will now be increased to 120 watts. The live stream continues at https://www.youtube.com/watch?v=6NXCxsrwHOE
Display MoreMR4.3 Update
After 24 hours at 80 watts, the cell mostly matched calibration temp within ±1°C. An increase of just under 2°C can be seen from 16:50 to 17:50, perhaps worth a closer look. The pressure declined gradually from 384 to 377 Pa. No unusual neutron or gamma signal was seen.
RGA analysis showed only D2 and DH. The traces of H2O and N2 are the residuals from the analyzer sampling tube and chamber.
The heater will now be increased to 120 watts. The live stream continues at https://www.youtube.com/watch?v=6NXCxsrwHOE
Great work!
I hope something Wonderful happens.
MR4.3 Update
During 24 hours at 120 watts, the cell still matched calibration temp within ±1°C. Small fluctuations at around 18:00 and 21:00 were due to changes in ambient temp from open doors etc. The pressure continued to decline slightly during this interval. No unusual neutron or gamma signal was seen.
RGA analysis showed predominantly D2 and DH, with an increasing fraction of mass 2 (H2?). The residual traces of H2O and N2 from the analyzer sampling tube and chamber remain unchanged. However, small signals at mass 32 (O2?) and 44 (CO2) have now appeared, suggesting some out gassing has continued at the higher temperature.
The heater will now be increased to 140, then 150 watts. The live stream continues at https://www.youtube.com/watch?v=6NXCxsrwHOE
Am I mistaken ?
It looks like that after the second step up the neutron counter pattern has changed somewhat
MR4.3 Update
During 24 hours at 150 watts, the cell still matched calibration temp within ±1°C. The pressure increased nearly linearly during this interval, rising from 400 to 450 Pa.
RGA analysis showed the largest fraction to be mass 3, probably DH. The substantial presence of mass 2 (H2), mass 17 (OH+) and mass 19 (HDO) support this conclusion.The signals at mass 28 (N2) and 44 (CO2) have also increased during the interval, confirming that substantial out gassing has continued at 247°C, as was seen during the calibrations.
Regarding the neutron count, an increase of about 50% over background was seen for five minutes or so, at 14:00 on 2 Jan. This is consistent with typical cosmogenic events and probably not related to the cell under test.
The heater will now be increased to 200 watts. The live stream continues at https://www.youtube.com/watch?v=6NXCxsrwHOE
Ha, the same neutron count behavior seen at the start of 150 watts is evident with the current power bump to 200 watts. At about 10 minutes into the heat increase, the count went up to almost twice the background of ~1 cpm for 1 minute. Then after another 10 minutes, it went up again several times, with a 5 minute moving average of 1.5 cpm that continued for several minutes more.. The pattern and the correlation with a recent power increase suggests this may be related to the cell dynamics. Further analysis of the data after the experiment is needed to be sure.
What about the current Geiger counters readings seem a bit high! 13.36 Move upwards!
