-
#760 by Rodal on 26 Dec, 2015 03:15
-
Doc, you're poynting again!Everyone recalls from your microwave oven Owner's Manual (you read that, right?
I know it's only 2D but can give you a feel for microwaves...
) that you should not operate the oven with it empty. If the RF energy is not absorbed in heating the food, it reflects back to the magnetron, which can then overheat.
An EmDrive looks to me a lot like an empty microwave oven. Where is all the energy going? Clearly some goes into heating the walls of the frustrum, and we are going to meausre that. If there is a force generated, some energy has to go into that, and exactly how much that is will be interesting.
But some is being reflected back. A measurement of VSWR could let us calculate that, so can figure how much net energy goes into the frustrum to do something there.
http://www.met.reading.ac.uk/clouds/maxwell/
I love the evanescent waves (no energy ha) heating the potato.
http://www.met.reading.ac.uk/clouds/maxwell/microwave_oven.html
Shell
It is fascinating how the Poynting vector (the energy flux: the rate of energy transfer per unit area) forms two vortices on the sides of the potato. Electromagnetic vortices !
This bugs me because the potato gets cooked in what 6 minutes? By what? Evanescent waves. In the simulation it looks as if very little energy is getting into the potato compared to the modes of traveling energy outside in the microwave cavity.
Even in the poynting plot nothing is shown but we know and can even do the numbers how much energy goes into that potato. I've read very good articles on the extraordinary spin and momentum of evanescent waves but time and time again I run into there isn't anything there as it's simply a decaying wave function. It's like the 800 pound gorilla in the room.
Enough silly ranting...
Shell
Your question is an excellent one. The problem is that the University of Reading simulation does not give a complete, good picture of what is going on in the microwave cooking process, as it only shows the electromagnetic Poynting vector with the potato probably modeled as a dielectric insert at the bottom of the oven. To model the microwave cooking process you need to have an analysis that couples the results of the electromagnetic solution to a heat transfer study, since what cooks the potato are the tiny vibrations in the potato (mainly the vibration of water molecules in the potato. "Vibration resonance" at the atomic level that results in heat generation. The University of Reading modeling ignores this interaction that resonates the water molecules dipoles). This webpage from Fujitsu does a better job, since it shows both the electromagnetic field solution as well as the heat transfer solution:
http://www.fujitsu.com/jp/solutions/business-technology/tc/fields/cae/poynting/microwave-oven.html
(if you don't read Japanese, make sure to use a browser like Google Chrome that automatically asks you whether to translate from the Japanese to English, and then answer yes)
Fujitsu uses their software named Poynting (see http://www.fujitsu.com/global/documents/about/resources/publications/fstj/archives/vol44-4/paper10.pdf) which, just like Meep, uses a Finite Difference Time Domain solution scheme.
Look at the animations, for example
This case is an excellent example why MEEP cannot give anywhere close to the full picture of what is going on in the EM Drive, since Meep only looks at the solution of Maxwell's equations.
It would be necessary to couple the solution to a thermal analysis. NASA tried to do that with COMSOL, although I think that their analysis was conducted at given eigenfrequencies (obtained from an finite element eigenvalue analysis): I think that NASA's COMSOL analysis is not a transient analysis in time, unlike the Meep solutions we have examined.
That's why I was telling RFMWGUY and GLENFISH that one must conduct a thermal analysis of RFMWGUY's experiment to understand what is going, one cannot do this based on intuition (in RFMWGUY's experiment is even more complicated because there is natural convection from the magnetron at low Reynolds number (involving perhaps chaotic, and certainly time dependent vortex shedding), that would mean coupling the analysis to a computer fluid mechanics code would be necessary to understand what is going with lift and drag produced by natural convection with a superimposition of turning the magnetron on and off.
From an experimental viewpoint, Shell has gone the right way: remove the magnetron from the EM Drive, which removes the big heat source. Eventually and ideally it would be best to isolate the EM Drive from the outside environment (to eliminate the influence of natural convection) and/or to also conduct experiments heating the EM Drive to the same extent but without RF excitation of the cavity, and compare results. -
#761 by TheTraveller on 26 Dec, 2015 03:19
-
A resonant cavity with a loaded Q of 50k (unloaded Q of 100k), at 2.45 GHz has a Time Constant (TC) of approx 6.5 usec being Q unloaded / (2 Pi Freq). Cavity fill time is then 5 x TC seconds or in this case approx 32.5 usec.
Whatever is going to happen will happen by the time 5 x TC seconds has passed.
If MEEP is doing the right stuff, should see reflected power maxed at the start and then drop to min at TC = 5. Which means almost no power inside the cavity to start and cavity stored energy follows a normal 5 TC charge curve.
Would be interesting to see a MEEP graph of cavity stored energy versus time, to see if it matches cavity fill time reality. -
#762 by aero on 26 Dec, 2015 03:24
-
A 30 times increase is speed would prove very useful to propagate meep to RF steady state. Not thermal steady state though. Perhaps a combined meep with variable resolution (step size) running on GPu's would begin to address the problem because 1000 years divided by 30 is still 30 years per second of simulated real time.
I do know that commercial FDTD codes with variable resolution are also available, so that also can be done.
Interestingly, related to the speed test results from Stanford. my computer runs on an AMD Phenon(TM) II 840T quad core engine. One thing about this engine is that it incorporates AMD's vector math built-in. That feature is not common and as I understand it, no longer available from AMD, but I think it helps speed meep calculations.
And doing the calculation 14 days per microsecond actually equals 38,356 years per second. That is the problem.
Yes, I agree. 30X isn't a lot when you are talking about something that grows in cubed fashion. I think the dynamic density of voxels on GPU is the only obvious solution to get an accurate answer. I'm going to poke around a bit more, but I doubt a decent open source one exists yet in any form.
Good. We need more eyes looking at the heart of the problem, and possible solutions.
Also, related to Dr. White's conjecture on the QV origin of the EM drive effect, I did find a paper detailing a time domain model of the QV. Incorporating the QV as a material option into the meep model would offer the hope of determining the validity of Dr. White's conjecture. The math was beyond me except to observe that it is time domain. Perhaps an E&M expert could evaluate the potential, but again, the voxel density would likely increase by orders of magnitude uniformly throughout the lattice. Although maybe that problem could be addressed by modelling only small volumes of QV within the cavity? -
#763 by aero on 26 Dec, 2015 03:30
-
A resonant cavity with a loaded Q of 50k (unloaded Q of 100k), at 2.45 GHz has a Time Constant (TC) of approx 6.5 usec being Q unloaded / (2 Pi Freq). Cavity fill time is then 5 x TC seconds or in this case approx 32.5 usec.
Whatever is going to happen will happen by the time 5 x TC seconds has passed.
If MEEP is doing the right stuff, should see reflected power maxed at the start and then drop to min at TC = 5. Which means almost no power inside the cavity to start and cavity stored energy follows a normal 5 TC charge curve.
Would be interesting to see a MEEP graph of cavity stored energy versus time, to see if it matches cavity fill time reality.
Yes, it would be most interesting. Thirty-five microseconds is only about a year and a half run time, too.
-
#764 by DnA915 on 26 Dec, 2015 03:39
-
Yes, it would be most interesting. Thirty-five microseconds is only about a year and a half run time, too.
Hey, those are the times I like hearing! Divide that by 30 with the GPU software and find some other optimizations and that has the potentials of just taking days. -
#765 by TheTraveller on 26 Dec, 2015 03:42
-
A resonant cavity with a loaded Q of 50k (unloaded Q of 100k), at 2.45 GHz has a Time Constant (TC) of approx 6.5 usec being Q unloaded / (2 Pi Freq). Cavity fill time is then 5 x TC seconds or in this case approx 32.5 usec.
Whatever is going to happen will happen by the time 5 x TC seconds has passed.
If MEEP is doing the right stuff, should see reflected power maxed at the start and then drop to min at TC = 5. Which means almost no power inside the cavity to start and cavity stored energy follows a normal 5 TC charge curve.
Would be interesting to see a MEEP graph of cavity stored energy versus time, to see if it matches cavity fill time reality.
Yes, it would be most interesting. Thirty-five microseconds is only about a year and a half run time, too.
Roger has shown a thrust curve during Rf power applied for 20% of 1 TC.
Note the thrust curve starts at zero, peaks at the end of the Rf pulse and then slowly drops off. So it would appear thrust is generated as Rf energy enters the cavity and thrust increases as the cavity energy fill progresses.
As far as I know his cyro cavities do not use an internal antenna but are feed Rf via a side wall mounted slit.
-
#766 by VAXHeadroom on 26 Dec, 2015 03:49
-
Meep runs up to now have been extremely unrepresentative of actual running times for the EM Drive experiments, so the main goal of massive amounts of processing power would be to run the finite difference Meep solution to times approaching real EM Drive runs (at least in the order of seconds).
This would enable one to understand:
...
5) Meep plots up to now have only shown the 3D electromagnetic field components at a plane. What is needed is a full 3-D solid electromagnetic field plotting processing capability to show the 3-D electromagnetic field distribution throughout the 3D volume instead of just an arbitrary plane.
...
I have already created videos with 5 slices, and have gained sufficient understanding of generating data from a given meep control file that I can create a video with as many slices as you'd like - including up to 'all of them'. The problem is figuring out a reasonable way to be able to look at data that dense. I've repeatedly asked for suggestions and none have been forthcoming. I know people are watching my videos because I can see the counts on YouTube - some up in the 500 range. This forum is the only place those videos are linked. If you have examples of what you'd like to see I'd be game to try to create it with this data.
Additionally, I can do any arbitrary math on any/all of the data meep can generate and output that as graphics as well. For instance showing ExH as a 3D vector (yes I know meep can do that and output it - just an example).
The meep CSV files are inconvenient for me to use (but that's all I have at the moment) and I've been trying to get the HDF5 source code distribution to compile into a library but have not yet been successful (it's a lot more complicated than it needs to be IHMO plus I'm rebuilding a bathroom and Christmas), but once I can do that I will try to link the ability to import/view the data into a game engine (Unity) where we can let individuals control the speed and viewpoint and data sets of the 3D display in real time. That will be generally useful to a wider community than this so it's a worthwhile project. -
#767 by Rodal on 26 Dec, 2015 12:23
-
Suggestions where given. For example, see here: http://forum.nasaspaceflight.com/index.php?topic=39004.msg1464599#msg1464599 most recently.Meep runs up to now have been extremely unrepresentative of actual running times for the EM Drive experiments, so the main goal of massive amounts of processing power would be to run the finite difference Meep solution to times approaching real EM Drive runs (at least in the order of seconds).
This would enable one to understand:
...
5) Meep plots up to now have only shown the 3D electromagnetic field components at a plane. What is needed is a full 3-D solid electromagnetic field plotting processing capability to show the 3-D electromagnetic field distribution throughout the 3D volume instead of just an arbitrary plane.
...
I have already created videos with 5 slices, and have gained sufficient understanding of generating data from a given meep control file that I can create a video with as many slices as you'd like - including up to 'all of them'. The problem is figuring out a reasonable way to be able to look at data that dense. I've repeatedly asked for suggestions and none have been forthcoming. I know people are watching my videos because I can see the counts on YouTube - some up in the 500 range. This forum is the only place those videos are linked. If you have examples of what you'd like to see I'd be game to try to create it with this data.
Additionally, I can do any arbitrary math on any/all of the data meep can generate and output that as graphics as well. For instance showing ExH as a 3D vector (yes I know meep can do that and output it - just an example).
The meep CSV files are inconvenient for me to use (but that's all I have at the moment) and I've been trying to get the HDF5 source code distribution to compile into a library but have not yet been successful (it's a lot more complicated than it needs to be IHMO plus I'm rebuilding a bathroom and Christmas), but once I can do that I will try to link the ability to import/view the data into a game engine (Unity) where we can let individuals control the speed and viewpoint and data sets of the 3D display in real time. That will be generally useful to a wider community than this so it's a worthwhile project.
Show the external surface distribution of the magnetic field with a 3-D surface contour plot to compare with NASA's
Is your Meep model a realistic model of reality?
Have you verified your Meep model vs. experiments or vs. other solutions?
Can you use Meep's graphical output to interpret what is going on?
Let's verify it and then use it to make a prediction
1) Verify (using Meep) the thermal camera measurements of NASA for their geometry, giving TM212 (assuming that the thermal profile is due to induction heating). Take a look at NASA's report on how the COMSOL analysis of the magnetic field looks like. Compare NASA's magnetic field prediction (and therefore, NASA's thermal measurements of induction heating) with that of your model using Meep.
********That's the first step. Any numerical model must be verified with experimental data, and with other solutions, before attempting to use the numerical model to make any predictions. You have to verify the accuracy and usefulness of your Meep model.That's suggestion number one********
2) Run Shell's geometry to predict (showing the magnetic field intensity) what the thermal camera should measure when viewing her experiment. (assuming that the thermal profile is due to induction heating)
For example: does your Meep model predict mode TE013 ? Then show us what the magnetic field intensity (assumed to result in the thermal profile) should look like, according to the Meep analysis of Shell's experiment (assuming that the thermal profile is due to induction heating)
Additional suggestions:
* You state <<I have already created videos with 5 slices, and have gained sufficient understanding of generating data from a given meep control file that I can create a video with as many slices as you'd like - including up to 'all of them'.>>. What is needed is to show the fields on the curved lateral conical surfaces of the frustum of a cone. Can you show the fields on the curved lateral conical surface ?
**Whenever you show a field distribution with a color scheme showing different magnitudes of the field, include (as in the example by NASA shown above) a numeric color ribbon for people to be able to tell what is the magnitude of the fields shown
*** Meep's standard output is in Cartesian global coordinates but the natural intrinsic coordinates of the problem are spherical. It is not that useful to look at Cartesian components, for the field components that are azimuthal and the components simultaneously perpendicular to the axis of axisymmetry and perpendicular to the azimuthal direction. Therefore it would be useful to convert your output to a spherical coordinate system and show the output therefore in the natural coordinates of the problem, which satisfies spherical symmetry (completely so for semi-spherical, rather than flat ends). -
#768 by SeeShells on 26 Dec, 2015 13:20
-
I've liked Paul Krugman's thoughts and insights into the the things that we surround ourselves with that make us human. And I believe he is right. We are seeing a change in (us, humanity) addressing such things as global warming and trying to take the first steps to the stars.
While I don't know if the EMDrive, Q-Thruster, ERD (call it what you want) will give us a world of flying cars or the fast lane to the planets and maybe beyond. I do know we seem to have pulled out heads out of the sand and are sincerely trying on many fronts. Here and NASA and several other private institutions we are sincerely trying to make a drive that seems to defy known physics. Poo poo all you want but, you know if it does, we take a step forward. If it doesn't then we still have the path we walked on to look back and say we have learned.
My hat's off to the cutting edgers, the crazy eddies, the dreamers and even the poo pooers (we need balance) for a better new year.
Shell
http://www.nytimes.com/2015/12/25/opinion/things-to-celebrate-like-dreams-of-flying-cars.html?smid=fb-share&_r=0 -
#769 by SeeShells on 26 Dec, 2015 13:53
-
A resonant cavity with a loaded Q of 50k (unloaded Q of 100k), at 2.45 GHz has a Time Constant (TC) of approx 6.5 usec being Q unloaded / (2 Pi Freq). Cavity fill time is then 5 x TC seconds or in this case approx 32.5 usec.
Whatever is going to happen will happen by the time 5 x TC seconds has passed.
If MEEP is doing the right stuff, should see reflected power maxed at the start and then drop to min at TC = 5. Which means almost no power inside the cavity to start and cavity stored energy follows a normal 5 TC charge curve.
Would be interesting to see a MEEP graph of cavity stored energy versus time, to see if it matches cavity fill time reality.
Yes, it would be most interesting. Thirty-five microseconds is only about a year and a half run time, too.
Don't you feel we're the blind men describing the elephant? Only seeing a little slice of what's happening?
Can someone explain to me in the SETI program where they have thousands of users logged in to give a just a little of their computer time to solve the heavy equations in their search for signals burred in the noise. Could we do something like that? Could someone with a better knowledge of distributed processing help here or contact SETI to see how they did it?
Would it even help in this quest?
Shell -
#770 by Rodal on 26 Dec, 2015 14:17
-
Well, actually I would like to use your example of SETI as an example how assumptions can end up with no results, because of incorrect initial assumptions. Initially (1940's and 1950's) people thought that intelligent, sentient beings would communicate wirelessly the same way we did initially: first analog radio transmissions, followed by analog TV transmissions and so on and on. The idea was that a signal from a planetary system on a distant star could be deciphered as being regular, with ORDER and therefore be distinguished from random noise.A resonant cavity with a loaded Q of 50k (unloaded Q of 100k), at 2.45 GHz has a Time Constant (TC) of approx 6.5 usec being Q unloaded / (2 Pi Freq). Cavity fill time is then 5 x TC seconds or in this case approx 32.5 usec.
Whatever is going to happen will happen by the time 5 x TC seconds has passed.
If MEEP is doing the right stuff, should see reflected power maxed at the start and then drop to min at TC = 5. Which means almost no power inside the cavity to start and cavity stored energy follows a normal 5 TC charge curve.
Would be interesting to see a MEEP graph of cavity stored energy versus time, to see if it matches cavity fill time reality.
Yes, it would be most interesting. Thirty-five microseconds is only about a year and a half run time, too.
Don't you feel we're the blind men describing the elephant? Only seeing a little slice of what's happening?
Can someone explain to me in the SETI program where they have thousands of users logged in to give a just a little of their computer time to solve the heavy equations in their search for signals burred in the noise. Could we do something like that? Could someone with a better knowledge of distributed processing help here or contact SETI to see how they did it?
Would it even help in this quest?
ShellQuoteIn 1967, a radio signal was detected using the Interplanetary Scintillation Array of the UK Mullard Radio Astronomy Observatory by Jocelyn Bell and Antony Hewish. The signal had a 1.337302088331 second period and 0.04 second pulsewidth. It originated at celestial coordinates 19h 19m right ascension, +21° declination. It was detected by individual observation of miles of graphical data traces. Due to its almost perfect regularity, it was at first assumed to be spurious noise, but this hypothesis was promptly discarded. After that, the discoverers proposed an alternative explanation that the signal might be a beacon or a communication from an intelligent extraterrestrial civilization and named it Little green men 1 (LGM-1).
Before the nature of the signal was determined, the researchers, Bell and her Ph.D supervisor Antony Hewish, somewhat seriously considered the possibility of extraterrestrial life:
We did not really believe that we had picked up signals from another civilization, but obviously the idea had crossed our minds and we had no proof that it was an entirely natural radio emission. It is an interesting problem - if one thinks one may have detected life elsewhere in the universe how does one announce the results responsibly? Who does one tell first?
However, as we have progressed, this idea has been turned on its head. You see, the further we progress in human wireless communication, the more we compress the information. Actually, nowadays we even encrypt the information in order to protect it. The more we compress the information, the less ordered it becomes, and the more difficult it becomes to distinguish from randomness (unless one has the key).
It is now accepted that any advanced civilization would actually communicate by means that would make their communications appear to be random sequences because the information would be highly compressed and perhaps even encrypted. Therefore the original SETI program could only work to look at communications from civilizations in their very early stage of communications, encompassing perhaps a few dozen years (which when compared to the lifespan of a civilization is a miniscule amount of time). Therefore this may explain why SETI has not been successful up to now and why such an approach is doomed.
So, I use the SETI example (your point about distributed processing is well taken and my comment does not address it) to show that approaches to find something can only benefit from discussion that encompasses all viewpoints, most prominently skeptical viewpoints, in order to make sense of our Universe...
The initial assumption that advanced civilizations would communicate with very regular patterns has now been shown (by ourselves, in just a few years of progress) to have been an incorrect assumption.
As to what is going on with the EM Drive: we have to find out experimentally, by scientifically analyzing the experiments and not by relying on assumptions. -
#771 by OnlyMe on 26 Dec, 2015 14:24
-
SETI uses a different kind of distributed processing. They gather data and break it into chunks that can be handled given time by many desktop computers.A resonant cavity with a loaded Q of 50k (unloaded Q of 100k), at 2.45 GHz has a Time Constant (TC) of approx 6.5 usec being Q unloaded / (2 Pi Freq). Cavity fill time is then 5 x TC seconds or in this case approx 32.5 usec.
Whatever is going to happen will happen by the time 5 x TC seconds has passed.
If MEEP is doing the right stuff, should see reflected power maxed at the start and then drop to min at TC = 5. Which means almost no power inside the cavity to start and cavity stored energy follows a normal 5 TC charge curve.
Would be interesting to see a MEEP graph of cavity stored energy versus time, to see if it matches cavity fill time reality.
Yes, it would be most interesting. Thirty-five microseconds is only about a year and a half run time, too.
Don't you feel we're the blind men describing the elephant? Only seeing a little slice of what's happening?
Can someone explain to me in the SETI program where they have thousands of users logged in to give a just a little of their computer time to solve the heavy equations in their search for signals burred in the noise. Could we do something like that? Could someone with a better knowledge of distributed processing help here or contact SETI to see how they did it?
Would it even help in this quest?
Shell
MEEP in this case is what generates the data, so what would be needed is both to create an installation package for meep that can run on different OSs and then some way to distribute individual chunks of the simulation process... The problem with that is that there is no way to know where second two will begin until you have simulated second number one.
If experiment were producing a volume of data that required distributed processing, it probably could be handled.., with existing software. The thing is we (not meaning me) are not generating that kind of information/data depth from experiment.., yet! (He says hopefully.)
--- I have come to rely on spell checkers so heavily, that I have come to despise them! --- -
#772 by DnA915 on 26 Dec, 2015 14:25
-
Don't you feel we're the blind men describing the elephant? Only seeing a little slice of what's happening?
Can someone explain to me in the SETI program where they have thousands of users logged in to give a just a little of their computer time to solve the heavy equations in their search for signals burred in the noise. Could we do something like that? Could someone with a better knowledge of distributed processing help here or contact SETI to see how they did it?
Would it even help in this quest?
Shell
Hey Shell,
I am fairly knowledgeable on distributed computing. One larger one I have participated in was the Folding at Home protein folding program. I believe at least at a time it was considered one of the worlds largest super computers. One thing that a massively disconnected super computer like this requires is that the jobs do not require much communication between them. On FAH, they would assign a single protein combination to you and your individual computer would check it. A bit back in this thread I offered to build this. We could make a host program that a user could run where some of the admins could assign individual MEEP jobs without a huge amount of effort. From what I can hear from those with EM simulation here (I'm a CFD guy myself), is that the current time period and grid resolution is not sufficient. When trying to share the same simulation across multiple computers, you actually have to have an overlap in the data and then cross share between each step of all the grid connection boundaries. For this to actually be worth it, I imagine it would take a minimum of a Gigabit connection, so this would not be worthwhile at this time in my opinion.
- David -
#773 by SeeShells on 26 Dec, 2015 14:34
-
Well, actually I would like to use your example of SETI as an example how such an enterprise can end up with no results, because of incorrect initial assumptions. Initially (1940's and 1950's) people thought that intelligent, sentient beings would communicate wirelessly the same way we did initially: first analog radio transmissions, followed by analog TV transmissions and so on and on. The idea was that a signal from a planetary system on a distant star could be deciphered as being regular, with ORDER and therefore be distinguished from random noise.A resonant cavity with a loaded Q of 50k (unloaded Q of 100k), at 2.45 GHz has a Time Constant (TC) of approx 6.5 usec being Q unloaded / (2 Pi Freq). Cavity fill time is then 5 x TC seconds or in this case approx 32.5 usec.
Whatever is going to happen will happen by the time 5 x TC seconds has passed.
If MEEP is doing the right stuff, should see reflected power maxed at the start and then drop to min at TC = 5. Which means almost no power inside the cavity to start and cavity stored energy follows a normal 5 TC charge curve.
Would be interesting to see a MEEP graph of cavity stored energy versus time, to see if it matches cavity fill time reality.
Yes, it would be most interesting. Thirty-five microseconds is only about a year and a half run time, too.
Don't you feel we're the blind men describing the elephant? Only seeing a little slice of what's happening?
Can someone explain to me in the SETI program where they have thousands of users logged in to give a just a little of their computer time to solve the heavy equations in their search for signals burred in the noise. Could we do something like that? Could someone with a better knowledge of distributed processing help here or contact SETI to see how they did it?
Would it even help in this quest?
Shell
However, as we have progressed, this idea has been turned on its head. You see, the further we progress in human wireless communication, the more we compress the information. Actually, nowadays we even encrypt the information in order to protect it. The more we compress the information, the less ordered it becomes, and the more difficult it becomes to distinguish from randomness (unless one has the key).
It is now accepted that any advanced civilization would actually communicate by means that would make their communications appear to be random sequences because the information would be highly compressed and perhaps even encrypted. Therefore the original SETI program could only work to look at communications from civilizations in their very early stage of communications, encompassing perhaps a few dozen years (which when compared to the lifespan of a civilization is a miniscule amount of time). Therefore this may explain why SETI has not been successful up to now and why such an approach is doomed.
So, I use the SETI example to show that approaches to find something can only benefit from discussion that encompasses all viewpoints, most prominently skeptical viewpoints, in order to make sense of our Universe...
The initial assumption that advanced civilizations would communicate with very regular patterns has now been shown (by ourselves, in just a few years of progress) to have been an incorrect assumption.
I will agree with you wholeheartedly in why SETI hasn't decoded anything worth while. I've thought along the same lines for years. If they have cracked and unbreakable quantum communications scheme or another technology a level above that we would be like a caveman trying to hear microwave cell phone communication. It is a very small slice of time I fear that communications like were using will be used in any evolving ETI.
That said, that I agree with you. I still wonder what we couldn't use the same scheme that SETI uses for distributed internet processing and maybe they would just share or help us to see if we could use some of it.
My mainframe days are as antiquated as ferrite Iron Core memories and transistors I'm afraid.
Shell -
#774 by Rodal on 26 Dec, 2015 14:35
-
...From what I can hear from those with EM simulation here (I'm a CFD guy myself), is that the current time period ...
That's correct. The amount of time modeled in the Meep runs does not even come close to the dozens of microseconds necessary to model the initial response, or close to show steady state in the actual experiments (encompassing seconds of response, and in cases like rfmwguy's experiment even the magnetron is turned on and off for seconds at a time).
- David...(I'm a CFD guy myself)...
- David
Concerning your experience with Computational Fluid Dynamics, there is a need to model some of these experiments natural convection problem. It involves low Reynolds number flow, but above the Reynolds number for the Stokes flow approximation, so a full Navier-Stokes model may be needed.
For example, rfmwguy has an experiment where he placed a naked magnetron on top of an EM Drive. The magnetron got very hot, which resulted in natural convection currents on the flat plate the magnetron was resting on, with unknown time dependence of the vortex shedding that resulted from it. The problem is more complicated because, rfmwguy turned the magnetron on and off during his experiment which interacted with the natural convection initially set by turning the magnetron on. This resulted in a complicated, low Reynolds number, transient response. The response is NOT unique: sometimes turning the magnetron ON results in a force down during his experiment and sometimes it does not. Glenfish made a valiant attempt at analyzing the response statistically, but he admitted that the statistical population of the sample population is not large enough to ascertain whether it represents the true statistical population of the (unknown transient response details of the) physics behind this experiment.
So, a computational fluid dynamics model of rfmwguy's experiment natural convection interaction with his turning the magnetron on and off would be fun to see.
What CFD computer program do you have experience with, and are you willing to take your time to model the natural convection in these experiments?
Since this is read by a wide audience, this question is really addressed at large: it would be useful for somebody to model the low Reynolds number, natural convection in these experiments with a CFD model -
#775 by DnA915 on 26 Dec, 2015 15:31
-
Concerning your experience with Computational Fluid Dynamics, there is a need to model some of these experiments natural convection problem. It involves low Reynolds number flow, but above the Reynolds number for the Stokes flow approximation, so a full Navier-Stokes model may be needed.
For example, rfmwguy has an experiment where he placed a naked magnetron on top of an EM Drive. The magnetron got very hot, which resulted in natural convection currents on the flat plate the magnetron was resting on, with unknown time dependence of the vortex shedding that resulted from it. The problem is more complicated because, rfmwguy turned the magnetron on and off during his experiment which interacted with the natural convection initially set by turning the magnetron on. This resulted in a complicated, low Reynolds number, transient response. The response is NOT unique: sometimes turning the magnetron ON results in a force down during his experiment and sometimes it does not. Glenfish made a valiant attempt at analyzing the response statistically, but he admitted that the statistical population of the sample population is not large enough to ascertain whether it represents the true statistical population of the (unknown transient response details of the) physics behind this experiment.
So, a computational fluid dynamics model of rfmwguy's experiment natural convection interaction with his turning the magnetron on and off would be fun to see.
What CFD computer program do you have experience with, and are you willing to take your time to model the natural convection in these experiments?
Since this is read by a wide audience, this question is really addressed at large: it would be useful for somebody to model the low Reynolds number, natural convection in these experiments with a CFD model
So fair disclosure up front: I write software for a living currently and only have a start on my M.S. in Aerospace, although I have taken a CFD class already that concentrates on both implementation and theory. I have some experience in both Ansys Fluent and OpenFoam. I would be much more comfortable doing 3D simulations in Fluent. Unfortunately, I don't currently have access to Fluent as I am sure you already know, it cost an arm and a leg. Its been over a year since I have used OpenFoam and I don't think with the current level of free time I have that I would be able to mesh the problem in 3D. If anyone wants to help model it, I'd be happy to leave it running on my computer for a week or two though as I have a fairly high end machine that would be ideal for this type of work.
Also, just on a side note, wouldn't this be rather hellish to model? It seems like having a fine mesh like rfmwguy's EMDrive would require an ultra fine grid to get near accurate results of the air interacting with the heated mesh, not to mention the difficulty of creating the model. I am sure there are tools that I am unaware of for modeling this kind of surface though.
-
#776 by Rodal on 26 Dec, 2015 15:43
-
You are correct that it would be hellish to model, a low Reynolds number, above the R# for Stokes flow, CFD model needing 3-D full Navier Stokes implementation: it would make the Meep model look like children's play (although for rfmwguy's experiment there is some axisymmetric symmetry that could be exploited to reduce modeling resources, it has four-fold symmetry about 90 degree increments, since I think he used square plates). That's why I pointed that out before and why it is uncertain to use intuition to really say what went on with lift and drag in rfmwguy's experiment.
Concerning your experience with Computational Fluid Dynamics, there is a need to model some of these experiments natural convection problem. It involves low Reynolds number flow, but above the Reynolds number for the Stokes flow approximation, so a full Navier-Stokes model may be needed.
For example, rfmwguy has an experiment where he placed a naked magnetron on top of an EM Drive. The magnetron got very hot, which resulted in natural convection currents on the flat plate the magnetron was resting on, with unknown time dependence of the vortex shedding that resulted from it. The problem is more complicated because, rfmwguy turned the magnetron on and off during his experiment which interacted with the natural convection initially set by turning the magnetron on. This resulted in a complicated, low Reynolds number, transient response. The response is NOT unique: sometimes turning the magnetron ON results in a force down during his experiment and sometimes it does not. Glenfish made a valiant attempt at analyzing the response statistically, but he admitted that the statistical population of the sample population is not large enough to ascertain whether it represents the true statistical population of the (unknown transient response details of the) physics behind this experiment.
So, a computational fluid dynamics model of rfmwguy's experiment natural convection interaction with his turning the magnetron on and off would be fun to see.
What CFD computer program do you have experience with, and are you willing to take your time to model the natural convection in these experiments?
Since this is read by a wide audience, this question is really addressed at large: it would be useful for somebody to model the low Reynolds number, natural convection in these experiments with a CFD model
So fair disclosure up front: I write software for a living currently and only have a start on my M.S. in Aerospace, although I have taken a CFD class already that concentrates on both implementation and theory. I have some experience in both Ansys Fluent and OpenFoam. I would be much more comfortable doing 3D simulations in Fluent. Unfortunately, I don't currently have access to Fluent as I am sure you already know, it cost an arm and a leg. Its been over a year since I have used OpenFoam and I don't think with the current level of free time I have that I would be able to mesh the problem in 3D. If anyone wants to help model it, I'd be happy to leave it running on my computer for a week or two though as I have a fairly high end machine that would be ideal for this type of work.
Also, just on a side note, wouldn't this be rather hellish to model? It seems like having a fine mesh like rfmwguy's EMDrive would require an ultra fine grid to get near accurate results of the air interacting with the heated mesh, not to mention the difficulty of creating the model. I am sure there are tools that I am unaware of for modeling this kind of surface though.
So, unless somebody else appears in the thread willing and able to perform such an analysis, let's just agree that the way forward in experimentation is to get the magnetron out of the way as done by Shell, and to structure the experiment program to address heat convection as discussed previously: 1) by insulating the EM Drive all around, and/or 2) also heating the EM Drive without introducing RF into the cavity so as not to excite electromagnetic fields inside the cavity, and/or 3) symmetric arrangements of twin EM Drive's at each end of the balanced beam (one heated without RF injection) -
#777 by VAXHeadroom on 26 Dec, 2015 17:14
-
Suggestions where given. For example, see here: http://forum.nasaspaceflight.com/index.php?topic=39004.msg1464599#msg1464599 most recently.
Show the external surface distribution of the magnetic field with a 3-D surface contour plot to compare with NASA's
Trying to wrap my head around what would be needed. Something like
summation of ABS( cosine(Angle(hvec[row][col], <Plane Normal>) ) ) * VectorLength( hvec[row][col] ) across all frames? That's purely a WAG for the equation...
Is your Meep model a realistic model of reality?
Have you verified your Meep model vs. experiments or vs. other solutions?
Can you use Meep's graphical output to interpret what is going on?
I'm using aero's model(control file) of SeeShells' device, so hopefully I'm working with a verified model.
Yes I can use H5View to look at the data as well as spreadsheet programs to look at the subsequently generated CSV files.
Let's verify it and then use it to make a prediction
1) Verify (using Meep) the thermal camera measurements of NASA for their geometry, giving TM212 (assuming that the thermal profile is due to induction heating). Take a look at NASA's report on how the COMSOL analysis of the magnetic field looks like. Compare NASA's magnetic field prediction (and therefore, NASA's thermal measurements of induction heating) with that of your model using Meep.
********That's the first step. Any numerical model must be verified with experimental data, and with other solutions, before attempting to use the numerical model to make any predictions. You have to verify the accuracy and usefulness of your Meep model.That's suggestion number one********
2) Run Shell's geometry to predict (showing the magnetic field intensity) what the thermal camera should measure when viewing her experiment. (assuming that the thermal profile is due to induction heating)
For example: does your Meep model predict mode TE013 ? Then show us what the magnetic field intensity (assumed to result in the thermal profile) should look like, according to the Meep analysis of Shell's experiment (assuming that the thermal profile is due to induction heating)
Additional suggestions:
* You state <<I have already created videos with 5 slices, and have gained sufficient understanding of generating data from a given meep control file that I can create a video with as many slices as you'd like - including up to 'all of them'.>>. What is needed is to show the fields on the curved lateral conical surfaces of the frustum of a cone. Can you show the fields on the curved lateral conical surface ?
Suggestion 2) above is - I think - the same as the graphical output suggestion earlier. I'll need help with the required equations...
I think I can show the surfaces. My thought on this is to look for where the field intensity (H field vector length) has an adjacent cell that drops below 1e-6 and just plot/compute those. A cursory look of the data inside the frustum shows minimum field intensities of >1e-3 so a few orders of magnitude should be a good margin (have to verify). Additionally, since meep is smoothing data across cell boundaries I may need to sum the last two adjacent to a cell <1e-6.
**Whenever you show a field distribution with a color scheme showing different magnitudes of the field, include (as in the example by NASA shown above) a numeric color ribbon for people to be able to tell what is the magnitude of the fields shown
I've been thinking on how to do that for about a week and think I have it. I can possibly do it directly in POV-Ray, but it may require some outside processing to build the graphic to put in a video.*** Meep's standard output is in Cartesian global coordinates but the natural intrinsic coordinates of the problem are spherical. It is not that useful to look at Cartesian components, for the field components that are azimuthal and the components simultaneously perpendicular to the axis of axisymmetry and perpendicular to the azimuthal direction. Therefore it would be useful to convert your output to a spherical coordinate system and show the output therefore in the natural coordinates of the problem, which satisfies spherical symmetry (completely so for semi-spherical, rather than flat ends).
I'm not going to be much help here I think...although...hmm let me think on at least a display-oriented solution...
-
#778 by Rodal on 26 Dec, 2015 17:58
-
Quote
I'm using aero's model(control file) of SeeShells' device, so hopefully I'm working with a verified model.
Shell's device cannot yet be used as confirmation of a Meep model: Shell's experiment has no thermal data provided yet. When she does provide it, there are still the issues of ceramic plates of different thickness on each surface. And it has no independent (for example COMSOL Finite Element) analysis for comparison. One needs to independently compare the Meep solution with other solutions and with experimental results to verify the Meep model. I proposed NASA's experimental and model data for TM212 for verification, because it is one of the very few comparisons we have with both experiments and numerical analysis by other means.
The need for verification is not due to Meep as a code but it has to do with the well known principle that Garbage in - Garbage Out even with the best code. As an example, it was only recently that the issue with the copper conductivity was addressed. There may be other issues with the model as well (*).
It is only by verification with other solutions and experimental data that one can gain confidence in a numerical model.
_______
(*) For example, one issue extensively discussed is that the Meep model encompasses a much shorter time frame (fraction of a microsecond) than necessary, and hence the (yet malformed) electromagnetic fields (not having achieved steady state) may not model well what will be shown by the thermal model measurement during time frames of seconds -
#779 by VAXHeadroom on 26 Dec, 2015 18:14
-
Quote
I'm using aero's model(control file) of SeeShells' device, so hopefully I'm working with a verified model.
Shell's device cannot yet be used as confirmation of a Meep model: Shell's experiment has no thermal data provided yet. When she does provide it, there are still the issues of ceramic plates of different thickness on each surface. And it has no independent (for example COMSOL Finite Element) analysis for comparison. One needs to independently compare the Meep solution with other solutions and with experimental results to verify the Meep model. I proposed NASA's experimental and model data for TM212 for verification, because it is one of the very few comparisons we have with both experiments and numerical analysis by other means.
The need for verification is not due to Meep as a code but it has to do with the well known principle that Garbage in - Garbage Out even with the best code. As an example, it was only recently that the issue with the copper conductivity was addressed. There may be other issues with the model as well (*).
It is only by verification with other solutions and experimental data that one can gain confidence in a numerical model.
_______
(*) For example, one issue extensively discussed is that the Meep model encompasses a much shorter time frame (fraction of a microsecond) than necessary, and hence the electromagnetic fields may not model well what will be shown by the thermal model measurement during time frames of seconds
OK, well maybe 'scrutinized' is a more accurate term than 'verified' in this case
I really only meant to say I'm not using a model file I made up - several of us are working with the same model.
Divide and conquer would seem to be a good strategy - if several people can work on the same input data and each work to verify one aspect of the model we might make some headway.
I will work to understand the thermal profile for the data I have. Once I have a toolpath established I will publish all the code so it can be peer reviewed.
) that you should not operate the oven with it empty. If the RF energy is not absorbed in heating the food, it reflects back to the magnetron, which can then overheat.
