-
#340 by ThereIWas3 on 18 Dec, 2015 01:31
-
Another way to eliminate air around the frustum, besides a vacuum, is to fill the space with expanding plastic foam. Put frustrum in a 30cm cubic box. Attach wires. Blow expanding plastic foam like used for insulation into the box. Allow to harden. Scrape off excess. Attach lid. Place on scale apparatus. Apply power. Repeat in all possible orientations.
-
#341 by OnlyMe on 18 Dec, 2015 01:50
-
Another way to eliminate air around the frustum, besides a vacuum, is to fill the space with expanding plastic foam. Put frustrum in a 30cm cubic box. Attach wires. Blow expanding plastic foam like used for insulation into the box. Allow to harden. Scrape off excess. Attach lid. Place on scale apparatus. Apply power. Repeat in all possible orientations.
In the case of rfmwguy's current frustum the magnetron is on top and heats up to between 150 and 200 degrees C. The foam is flamible. -
#342 by rq3 on 18 Dec, 2015 02:08
-
#343 by SeeShells on 18 Dec, 2015 02:11
-
.....
A few weeks ago, I used those Yang plots to determine the efficiency of her EMdrive while the input power increased.
2) The experimental results of Prof. Yang showed strong nonlinearity between the measured force and the input power (see attached chart). Prof. Yang's experiments show diminishing returns (actually slight decrease in measured forces for input power exceeding 300 Watts and general flat response) for increasing input power.
Hence it does not follow that the thermal artifacts will scale linearly in experiments conducted at much higher input power. On the contrary, Prof. Yang's experimental results show strong nonlinearity, with the touted "EM Drive" force dependence on input power effectively dissappearing after about 300 watts, the dependence looks practically flat at input powers greater than 300 W. The nonlinearity of the "force" vs. input power experimental relation of Prof. Yang has not yet been scientifically modeled hence its nature can only be speculated until a verifiable model is demonstrated.
This shows that if anything, experiments conducted at ambient conditions with low power may be highly misleading and NOT linearly scalable to higher powers, just as it would be highly misleading to conflate the flight of an insect with the flight of an airplane (the aerodynamics are completely different at very low Reynolds numbers).(The range of Reynolds number in insect flight is about 10 to 10^4, which lies in between the two limits that are convenient for theories that try to simplify the nonlinearity of Navier Stokes fluid dynamics: inviscid steady flows around an airplane's airfoil and Stokes flow experienced by a swimming bacterium. For this reason, this intermediate Reynolds number range used by insects in their flight is not as well understood as the high Reynolds number regime for airplanes. )
What I did is to bring all generated forces back to 1kW of power to get an idea how her frustum performed while she gradually increased the power...
It didn't show a straight linear relation, but a steep decline in efficiency at first, then leveled out for a long period and at the end showed improvement again, which might hint that a drastic increase in power "might" yield better then expected results.
I find it puzzling to why it stayed almost completely level from around 800W up to 2200W...
The performance degrading from 300W to 700W is probably due to thermal effects?
Nice workup, looking forward to comparing step mode in my tests to see how well I've mitigated thermal issues in keeping the tune for a mode.
Shell
PS: The data and the way they built their frustum does seem to point to a thermal deformation mode resonance shifting until the power slope band width starts to run into more resonance.
This isn't Yang's Spectrum but just one I picked for show and tell.
-
#344 by Flyby on 18 Dec, 2015 08:26
-
Well, what I take out of it is that in order to get a better, clear signal out of the (thermal) noise zone, you either should go for a 200-300W input power or further increase beyond 2500W...
As power is directly related to the thermal effects( buoyancy, material deformation, etc) the ratio of produced force/power needs to be as big as possible to get the best results with the least negative effects.
Too bad we don't have the relation of power/buoyancy (causing additional forces) or power/material deformation (causing frequency shifting, hence less force generation?). That would be interesting to pit those 3 graphs against each other.
Again, although I do play along as if it works, there is certainly no evidence it does. I remain a skeptic until proven otherwise...
Only, one has to engage into it, to investigate the possible effect. So for the sake of investigation and experimenting, one can not do anything else then pretend...
-
#345 by RERT on 18 Dec, 2015 09:45
-
Forgive me for using a single post to respond to multiple others, but hopefully this won't be too confusing.
1. I thought I heard RFMWGUY a few posts back saying indirectly he might give up without perfect isolation of thermal effects. Don't!
You can essentially eliminate thermal effects: as I've suggested before (and I think Space Ghost 1962 was thinking along similar lines) use electric heaters and thermostats to keep the rig at a constant temperature irrespective of whether or not the frustrum is powered. Run tests long enough for the system to generate numbers while at thermal equilibrium. Run control tests without frustrum power. Do both with the frustrum in different up/down/left/right orientations.
Yes, the fact that the whole thing is above ambient temp will create airflow and 'vortex shedding', but this will be the same with and without thrust, if any exists. You don't need to calculate the effects precisely if that is true. With enough runs I'm sure you will find a signal if it's there and is large enough.
If you do this and find an effect, then you can worry about residual errors.
2. Number of repetitions required to identify signal
I think it is profitable to consider every on/off cycle of frustrum power, while at thermal equilibrium, as a separate event. Dynamic response over hundreds of power cycles may reveal more than the gross averages of one or two experiments.
3. OnlyMe at post 196 on dark matter/energy
The courses I took on general relativity and cosmology are now over 34 years ago, so it's fair to say I'm eager to avoid a detailed debate with someone more current!
I acknowledge what you say about dark energy being an extra term in GR equations, not a balancing item for conservation of energy. However, consider this: dark energy implies (so I read) an energy density for the vacuum, and a total energy which changes as the universe expands. If the theory including dark energy has the total energy of the universe conserved, it's very hard to see how the same observed dynamical behaviour is consistent with the conservation of energy in the Universe absent dark energy. In that sense, dark energy is a balancing item for CoE.
I realize the topic might be conceptually slippery, and I may be wrong. I'd be interested in any response.
R.
-
#346 by glennfish on 18 Dec, 2015 13:01
-
...A statistical analysis is unconvincing (and therefore not straightforward) for your tests because the sample population in your tests is way too small to arrive at a conclusion that your sample population is representative of the true statistical population, nor is the sample population large enough to determine what is the appropriate statistical distribution for a parametric test.
I left that determination up to a professional statistician. I did not get that pronouncement from them.
We have a disagreement here. I have no problem with that. ...
I have not seen a statistical argument in these threads disagreeing with the fact that the sample population in your tests is way too small (to arrive at a conclusion that your sample population is representative of the true statistical population). Therefore I have not seen a justification that you can use only statistical tests to arrive at a robust statistical conclusion based on your test results.
PS: I am a member of the American Statistical Association, and use statistical models (as well as information theory) to make a living.
Dr. Rodal,
On the statistical analysis, I'm the target, not RFMWGUY. Any faults in that analysis are mine, not his.
1st. I agree on the sample size issues. My first request after that run were replication runs to characterize things more fully. Unfortunately he had disassembled his setup at that point and was unable to comply.
2nd. The data collection was not designed with a statistical analysis in mind. It required "post-hoc" data mining to pull about 40 mag on / mag off sample sub-sets out of a single run. The data resolution only permitted between 10 and 15 data points per on or off cycle, which in itself is a small sample. Ideally, the data would have been sampled at a much higher rate permitting higher resolution, but it was what it was.
3rd. The analysis that indicated anomalous behavior was based on a suggestion that there would be a difference in the mag-on behavior vs the mag-off behavior, specifically, that the mag on behavior would change during the mag-on cycle. That was not an a-prior assumption. Given that suggestion, it was possible to establish that for this one run, of about 40 cycles, of 10-15 data points per cycle, that the slope behavior for mag-on was statistically significantly different than the mag-off behavior, for the 40 odd sub-sets of that single run. That analysis with raw data was provided in this forum and is referenced in RFMWGUYs report. The algorithms that extracted the data for analysis are included in those uploads and include live VBA code if anyone wants to replicate/critique.
4th. The analysis could not confirm or deny any hypothesis beyond the slope behavior was different given these constraints. As to the cause, the magnitude, the why, post-hoc analysis with a post-hoc experimental design as we both know is the worst possible world. However, the data is indicative of some kind of signal in the noise, which is different from no signal in the noise.
5th. If you review my commentaries in this forum on methods, sampling, experimental design, etc. you see that I'm a fellow traveler on getting this right and up front.
IMHO, RFMWGUY's data shows a signal of some kind in that one run of 40 odd sub-samples. I don't know what that means, and even if it's statistically significant for that one run, which it is, I am in no position to state if it was thrust, delayed thermals, or unicorn dust.
IMHO, the next round of tests requires multiple runs under multiple criteria pre-defined carefully before the power is turned on. I'd be pleased to exchange methodology suggestions with you and if you have specific critiques of the actual data reduction techniques and analysis, I'd love the debate. At that time, the only statistical feedback I got in the forum was "so you have statistically proven that the EMDrive produces thrust?" That was most unhelpful feedback.
Finally, the ideal model would state something like "when I turn the power on I will see x micro-newtons of thrust" and you design the experiment to see if you get x or something else. In this case, since there is no accepted theory, and hence no means of calculating x in advance, the first question to me is simply, can you find x > 0. From my point of view, that pushes the DIY testing into the domain of statistical analysis, and if x is greater than 0, statistically, then maybe we can come up with something that says x = f(stuff). From an engineering and physics point of view, stuff is a lot of possible things that hopefully the DIY folks can gradually reduce from known items like thermals, airflow, Lorenz, etc., until there is either no residual to account for, or there is. -
#347 by Rodal on 18 Dec, 2015 13:06
-
Thank you for your nicely written, objective reply. There is no debate between us. Great, sound suggestions....A statistical analysis is unconvincing (and therefore not straightforward) for your tests because the sample population in your tests is way too small to arrive at a conclusion that your sample population is representative of the true statistical population, nor is the sample population large enough to determine what is the appropriate statistical distribution for a parametric test.
I left that determination up to a professional statistician. I did not get that pronouncement from them.
We have a disagreement here. I have no problem with that. ...
I have not seen a statistical argument in these threads disagreeing with the fact that the sample population in your tests is way too small (to arrive at a conclusion that your sample population is representative of the true statistical population). Therefore I have not seen a justification that you can use only statistical tests to arrive at a robust statistical conclusion based on your test results.
PS: I am a member of the American Statistical Association, and use statistical models (as well as information theory) to make a living.
Dr. Rodal,
On the statistical analysis, I'm the target, not RFMWGUY. Any faults in that analysis are mine, not his.
1st. I agree on the sample size issues. My first request after that run were replication runs to characterize things more fully. Unfortunately he had disassembled his setup at that point and was unable to comply.
2nd. The data collection was not designed with a statistical analysis in mind. It required "post-hoc" data mining to pull about 40 mag on / mag off sample sub-sets out of a single run. The data resolution only permitted between 10 and 15 data points per on or off cycle, which in itself is a small sample. Ideally, the data would have been sampled at a much higher rate permitting higher resolution, but it was what it was.
3rd. The analysis that indicated anomalous behavior was based on a suggestion that there would be a difference in the mag-on behavior vs the mag-off behavior, specifically, that the mag on behavior would change during the mag-on cycle. That was not an a-prior assumption. Given that suggestion, it was possible to establish that for this one run, of about 40 cycles, of 10-15 data points per cycle, that the slope behavior for mag-on was statistically significantly different than the mag-off behavior, for the 40 odd sub-sets of that single run. That analysis with raw data was provided in this forum and is referenced in RFMWGUYs report. The algorithms that extracted the data for analysis are included in those uploads and include live VBA code if anyone wants to replicate/critique.
4th. The analysis could not confirm or deny any hypothesis beyond the slope behavior was different given these constraints. As to the cause, the magnitude, the why, post-hoc analysis with a post-hoc experimental design as we both know is the worst possible world. However, the data is indicative of some kind of signal in the noise, which is different from no signal in the noise.
5th. If you review my commentaries in this forum on methods, sampling, experimental design, etc. you see that I'm a fellow traveler on getting this right and up front.
IMHO, RFMWGUY's data shows a signal of some kind in that one run of 40 odd sub-samples. I don't know what that means, and even if it's statistically significant for that one run, which it is, I am in no position to state if it was thrust, delayed thermals, or unicorn dust.
IMHO, the next round of tests requires multiple runs under multiple criteria pre-defined carefully before the power is turned on. I'd be pleased to exchange methodology suggestions with you and if you have specific critiques of the actual data reduction techniques and analysis, I'd love the debate. At that time, the only statistical feedback I got in the forum was "so you have statistically proven that the EMDrive produces thrust?" That was most unhelpful feedback.
Finally, the ideal model would state something like "when I turn the power on I will see x micro-newtons of thrust" and you design the experiment to see if you get x or something else. In this case, since there is no accepted theory, and hence no means of calculating x in advance, the first question to me is simply, can you find x > 0. From my point of view, that pushes the DIY testing into the domain of statistical analysis, and if x is greater than 0, statistically, then maybe we can come up with something that says x = f(stuff). From an engineering and physics point of view, stuff is a lot of possible things that hopefully the DIY folks can gradually reduce from known items like thermals, airflow, Lorenz, etc., until there is either no residual to account for, or there is.
It is unfortunate that the experimental setup was disassembled at the point of your involvement, as the scientific method is a continuous loop, ongoing process, and it is better not halted:
Other experiments are ongoing and being planned, which will clarify what is going on.
-
#348 by OnlyMe on 18 Dec, 2015 13:49
-
Well, what I take out of it is that in order to get a better, clear signal out of the (thermal) noise zone, you either should go for a 200-300W input power or further increase beyond 2500W...
As power is directly related to the thermal effects( buoyancy, material deformation, etc) the ratio of produced force/power needs to be as big as possible to get the best results with the least negative effects.
.....
Initially I had a similar thought. The problem is Yang never provided enough information about her microwave generation setup aside from mentioning it involved a magnetron. Add.....
Yang provided the following in her 2013 paper,
"... the practical maximum microwave output power is 13 W, 120 W, 85 W, 65 W, 45 W, and 48 W respectively at the nominal output power 200 W, 300 W, 400 W, 500 W, 600 W, and 700 W."
The practical maximums above fit well with the thrust mN in the graph linked below.
http://forum.nasaspaceflight.com/index.php?action=dlattach;topic=39004.0;attach=1087396
Without better specifics from Yang the nominal output - raw power-in figures are of little real value, since it is unknown why there is a better practical output at 300 watts, that falls until at 700 watts it has returned to nearly an identical efficiency as it began at 200. It would be better to use the practical maximums of 13 W, 120 W, 85 W, 65 W, 45 W, and 48.
This might suggest that a cleaner initial microwave source would be better... But even that is guesswork without specific information from Yang and Shawyer.., or a great deal more independent test results... So we wait now for Shell to feel better. -
#349 by glennfish on 18 Dec, 2015 13:55
-
1. I thought I heard RFMWGUY a few posts back saying indirectly he might give up without perfect isolation of thermal effects. Don't!
Stating agreement.
There are many ways to isolate thermal effects. The harder ways and the easier ways. The harder ways are to spend a gazillion hours designing and building so that the physical device has no thermals. IMHO you can never succeed in that approach.
The easy way is to get the thermal behavior to become highly repeatable to the point where you can calculate it exactly with a tiny error confidence interval. When thermals become a calculable function, with a tiny Standard Deviation, you literally subtract out the thermals and start looking for a signal. Lots of ways to do that mathematically & statistically.
IMHO thermals are almost certainly an exact function of total joules per second, frustum temperature & ambient air temperature and ambient air pressure and you can almost certainly come up with an exact equation for a given frustum design that accurately predicts thermal effects. A few trials focused on the thermals only, will quickly produce an equation that can be subtracted from runs where you're trying to see thrust. To get the standard deviation to its minimum will require a modest number of thermal only runs, but once you have that equation for that frustum/mag combination, anything that deviates from that equation is a signal of some kind.
I would invest in a thermocouple or two rather than design a thermally controlled frustum.
-
#350 by Rodal on 18 Dec, 2015 13:57
-
Hey, why not try my design. I think it is easier to understand.
Thank you for your contribution and welcome to the EM Drive thread
Here it is:
http://forum.nasaspaceflight.com/index.php?topic=38996.0
There are many issues for discussion on your report, for example, to pick one:
If a force were to be produced, then work would be done when the electromagnetic engine moves in space (Work = Force * displacement). If the force is produced without energy radiation (or mass outflow), the work is performed without spending energy. Then, the proposed electromagnetic device would be a perpetual-motion machine.
As (I think it was first aptly remarked by Frobnicat in these threads) it is curious that proponents of propellant-less electromagnetic space propulsion that don't rely on external fields for propulsion, like electromagnetic tethers, etc., seem to avoid the consequence of their device being a perpetual motion machine. If the device is a perpetual motion machine, it could be used for energy purposes here on Earth, which prompts the question as to why such discussions (of using their electromagnetic self-accelerators for energy production on Earth instead of for transportation purposes) not addressed. -
#351 by Flyby on 18 Dec, 2015 14:07
-
After refreshing my knowledge about angular momentum and gyroscope behavior and combining that with a detailed study of pictures of Shawyer's "EMdrive on a rotating table", I'm inclined to say that zen-in was (partially) right...
Although I differ from opinion with him that it is not_so_much the sloshing of the water,because much of the directions cancel each other out (tubing in all directions) but the angular momentum of pump motor that is at cause.
I'm kinda perplexed that for an engineer, Shawyer did not think about that.
The more that the problem could have been easily solved simply by re-aligning the pump 90° and let the plane cross the axis of the air-bearing. I'll add some pictures later on.
to illustrate more visually what i mean :
It is not hard to see the very same setup on the EMdrive test rig, except, instead of hanging it on a rope, it floats on an air-bearing spindle, but the rotation of the pump is in exact the same position as the wheel in the video.. and will consequently most likely also make the table turn.
Bottom line is that the credibility of that test entirely depends on the rotation direction of that pump...sad and somewhat disappointing...but true...
I'd say there is a very high probability (50%, depending on the direction the pump motor turns) that the table turns, not due to an EMdrive effect, but simply because of a gyroscopic effect...
not to say...but my confidence in this type of EMtest has been seriously dented. If anything should be replicated it will first, and before anything else, need to solve the torque issue.
added:
From the video Shawyer posted it is impossible to see when the pump was activated. Was the pump already running BEFORE the magnetron was turned on? If not, if it was turned on at the same moment the magnetron was turned on, what direction does it spin?
Does the rig still exists and is it possible to do just a dry run with only the pump running?
Those are legitimate questions that only Shawyer can answer.
If he wants to keep his credibility i think it would be in his best interest to answer these. This is not bullying or hand waving... just some credible engineering questions...
Maybe TT could slip these questions through to him and hope for an answer?
-
#352 by RotoSequence on 18 Dec, 2015 15:09
-
http://i.imgur.com/DUnuPMF.gifv
Always fun to watch wave propagation in other mediums and wonder how this stuff might all be connected.
-
#353 by lmbfan on 18 Dec, 2015 15:57
-
I attempted to do a software analysis of the simple geometry in my last post (based on dimensions and picture provided by rmfwguy), and I ran into a few issues. One is purely software related and I will research that one on my own, however, the following will need to be answered to a reasonable degree of certainty before any accurate analysis is done:
1) Please verify for the galvanized steel housings, specific heat of 490 J/(kg K)
2) Please verify thermal conductivity - engineeringtoolbox.com specifies a range of 6-26 W/(m K)
3) Wattage of the magnetron that is converted to heat (i.e. how many Watts is the heat source?)
4) Please verify density of housing is 7.85 g/cm^3
5) Convection coefficient for convection (I found something that said it was ~10.45 kg/(sec^3 C) for air at 0 m/s)
I may have more questions as I delve further into this. I will, of course, do my best to answer any issues myself, however, if someone can verify with a minimum of effort and save me hours of research time, assistance would be greatly appreciated. -
#354 by rfmwguy on 18 Dec, 2015 16:36
-
1) Please verify for the galvanized steel housings, specific heat of 490 J/(kg K)
a) It is galvanized steel.
2) Please verify thermal conductivity - engineeringtoolbox.com specifies a range of 6-26 W/(m K)
a) Cannot help here
3) Wattage of the magnetron that is converted to heat (i.e. how many Watts is the heat source?)
a) 900 watts
4) Please verify density of housing is 7.85 g/cm^3
a) cannot help here
5) Convection coefficient for convection (I found something that said it was ~10.45 kg/(sec^3 C) for air at 0 m/s)
a) cannot help here
-
#355 by zen-in on 18 Dec, 2015 17:08
-
After refreshing my knowledge about angular momentum and gyroscope behavior and combining that with a detailed study of pictures of Shawyer's "EMdrive on a rotating table", I'm inclined to say that zen-in was (partially) right...
Although I differ from opinion with him that it is not_so_much the sloshing of the water,because much of the directions cancel each other out (tubing in all directions) but the angular momentum of pump motor that is at cause.
I'm kinda perplexed that for an engineer, Shawyer did not think about that.
The more that the problem could have been easily solved simply by re-aligning the pump 90° and let the plane cross the axis of the air-bearing. I'll add some pictures later on.
U-toob was here
It is not hard to see the very same setup on the EMdrive test rig, except, instead of hanging it on a rope, it floats on an air-bearing spindle, but the rotation of the pump is in exact the same position as the wheel in the video.. and will consequently most likely also make the table turn.
Bottom line is that the credibility of that test entirely depends on the rotation direction of that pump...sad and somewhat disappointing...but true...
I'd say there is a very high probability (50%, depending on the direction the pump motor turns) that the table turns, not due to an EMdrive effect, but simply because of a gyroscopic effect...
not to say...but my confidence in this type of EMtest has been seriously dented. If anything should be replicated it will first, and before anything else, need to solve the torque issue.
added:
From the video Shawyer posted it is impossible to see when the pump was activated. Was the pump already running BEFORE the magnetron was turned on? If not, if it was turned on at the same moment the magnetron was turned on, what direction does it spin?
Does the rig still exists and is it possible to do just a dry run with only the pump running?
Those are legitimate questions that only Shawyer can answer.
If he wants to keep his credibility i think it would be in his best interest to answer these. This is not bullying or hand waving... just some credible engineering questions...
Maybe TT could slip these questions through to him and hope for an answer?
I don't believe gyroscopic precession can be used to explain Shawyer's video. The pump is not spinning fast enough for gyroscopic effects. But just having a motor mounted with its shaft parallel to the rotational axis of the apparatus is enough to start it rotating. Several years ago I mounted a motor on a rotating table. When the motor was spun up the table rotated. When the motor was switched off the table rotated back to its original position. We don't now enough about Shawyers apparatus to be able to analyze what is happening. That's why the focus has been on DIY experiments. -
#356 by SteveD on 18 Dec, 2015 17:09
-
Unless someone demonstrates or points out a potential error source that approaches a triple-digit micronewton level, I will not chase those phantoms. Lorentz force in the horizontal axis falls into this category as does thermal plume turbulence (fluid analysis), imho. Show me a non-jet down force test result or example and I may take that off of my phantom list. There were no jets identified on the top plate emanating from the mag per my thermal scans, nor was there uneven heating of any of the outer surfaces; a clear sign of air jets.
So, statistics are most straight-forward way to analyze beam displacement variances between mag on and mag off cycles. No other error source hypothesis approaches the force level needed to interrupt the thermal lift track; Lorentz, air jets, plume turbulence, thermal expansion, etc., etc.
If anyone has other ideas...I'm all ears and will test for them if funding allows next year.
Well I had thought that the transformer hum might have vibrated the hot air trapped in the heat sink chamber, causing more of it to spill out, reducing lift. Would think that this would have shown up clearly on the thermal camera though. Best was I can think of to test for this would be simply proceeding as planned. If you remove the magnetron from the top of the thing and it goes down more with higher Q . . . well that's interesting.
I think the big problem with mangetrons are that their outputs can take a random walk. Sometimes you get very little in resonance and sometimes you get more. Statistics would seem to be an excellent way of dealing with an rf source given to random walks.
-
#357 by zen-in on 18 Dec, 2015 17:19
-
...
Doc, while it was not consistent each time, its easy to see from the video of the mag spectrum that its own rf stability is more likely the cause rather than a transient thermal natural convection effect, which I am trying to understand what you mean here.
Remember, the displacement was downward (against lift), not upwards during mag-on at a higher occurance over the 2000 or so data points. Your suggestion seems to imply vertical (upward) jetting which I cannot visualize in my test. The rising plume of thermals should (if I understand basic fluid dynamics) impose an upwards lift, not a downward force.
Is there something you can point me to help me understand an opposite effect of what I am thinking? i.e. a rising, thermal plume causing a downwards force?
You used a wire mesh construction in the hopes of eliminating bouyancy and jetting effects. However I believe the parts of your apparatus that do promote thermal effects are the Copper tray shapes above and below the wire mesh fustrum. It doesn't matter if the fustrum is pointed up or down, there will always be one tray that is positioned to trap hot air and supply lift. -
#358 by rfmwguy on 18 Dec, 2015 17:31
-
Except therm vids show little heating below mag
-
#359 by Rodal on 18 Dec, 2015 17:47
-
It is explained by thermal convection on a plate that is heated above the plate (by the magnetron). Convection currents are generated whereby the hotter air above the plate experiences a buoyancy force going upwards. As a result of this the plate is virtually "sucked up" experiencing a lift force and the colder air above the hot air goes downward to occupy the space previously occupied by the hotter air. This convection is time-dependent, the fluid flow is not laminar, but it involves vortices above and below the plate. Vortex shedding takes place intermittently. The lift experienced by the plate and by the hot magnetron partially open cavity (due to their own buoyancy) is a chaotic function of time (due to the nonlinear nature, time dependent nature of the Navier-Stokes equation, particularly in this low Reynolds number regime). Due to the low Reynolds number regime being above Stokes flow, in an intermediate region of the Navier-Stokes flow which is difficult to model (unless done numerically) the time dependence of the convection and vortex shedding is complicated, involving chaotic fluctuations with time. The turning ON and OFF the magnetron further complicates the time-dependent nature of these fluctuations.
The chaotic time dependence of the lift force, and its direction is not subject to intuition any more than you can intuit the time dependence and direction of lift of a wing in the region of stall, beyond the critical angle of attack.
