Author Topic: EM Drive Developments - related to space flight applications - Thread 2  (Read 2165813 times)

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5895
  • USA
  • Liked: 6045
  • Likes Given: 5325
...
Now, in this case, the entire chamber is in the near field of these 15 cm waves. Near fields can be pretty non-intuitive (for example, not weakening as 1/R^2). I would be especially wary of what's going on in the reactive near field (within 2 cm or so of the drive), which is likely to be non-intuitive even for near-fields.

Do we know if MEEP adequately models the reactive near field? I don't see signs of it in the plots, but I do not trust my intituition, so things may be fine, but I thought I would ask.
Good observation.  I do not see any evidence of the reactive near field at the boundaries  of the upper picture (attached below) either.
Perhaps it is due to the low fidelity of the color contour plots that only display color intensity and do not display contour lines, it would be better to also plot contour lines as contour lines would better show the reactive near field than just color intensity.

Perhaps it is due to how the boundary conditions were set up. I do not know how the boundary conditions for the Finite Difference solution were set up. 

I would be interested in both the near field on the EM Drive as well as the field lines next to the vacuum chamber.  Again, it may be due to the lack of contour lines, but what I see are waves going through the vacuum chamber as if there would be no boundaries in the upper picture below.  I agree with @aero that the bottom picture below shows near fields (particularly near the vacuum chamber corners) that make some sense to me.

To be rigorous, MEEP does not integrate Maxwell's equations, MEEP uses a finite difference solver to solve Maxwell's equations by finite differences both in space and time.

The answer to <<Do we know if MEEP adequately models the reactive near field>> is that since MEEP is a Finite Difference solver, its adequateness is completely dependent on the mesh discretization.  Just because a MEEP 3-D solution takes 20 hours for example, it does not mean that is near convergence to the correct solution.







If this is a 3-D solution, it would help to also see the circular cross-section of the vacuum chamber, perpendicular to the longitudinal axis of the vacuum chamber, to also better understand what is going on.  It is very difficult to tell what is going on in a 3-D problem when one only can see a rectangular cross-section (is it the rectangular cross-section that crosses the center of the vacuum chamber ?)

...
Also, please don't forget that static fields can cause forces. "It would be nice" to see a breakdown of force by cause here.

Another excellent point.  That would also help to better understand what is going on.

Good effort, please carry on  It is laudable to have someone at this forum dedicate so much time and effort towards a classical physics explanation of the EM Drive.
« Last Edit: 02/16/2015 01:55 PM by Rodal »

Offline aero

  • Senior Member
  • *****
  • Posts: 2785
  • 92129
  • Liked: 724
  • Likes Given: 249
Including the vacuum chamber requires a lattice size of 26x31 x no-size, so all runs will be 2D with the vacuum chamber included. I tried to run a 3D but ran out of memory before resolution got as large as 10. That is not enough resolution to detect small features of the thruster cavity model and so is not useful for our purposes.

After I finish my current set of runs I can remove the vacuum chamber from the model and shrink the lattice to perhaps get some useful 3D data of just the thruster cavity. Even then in 3D the resolution will be low.

As for whether or not meep integrates the equations, I think it is a matter of terminology. When I went to school we called what meep does "numerical integration." If you want to change my terminology and re-educate me so be it, but a finite difference solver is new terminology to me, and is very similar if not exactly what we called numerical integration in the early days of digital computers.
Retired, working interesting problems

Offline Reactionless

  • Member
  • Posts: 7
  • Liked: 1
  • Likes Given: 1
While this single run detected only 1/3 of the experimental measured value, I will brazenly write that I think we can forget about axions and dark matter. We can forget about exotic theory and consider that the thrust force likely results from a little understood characteristic of evanescent waves and the forces they generate.
Well, partly perhaps, but lets wait until I run the same model without the vacuum chamber. I have ran enough cases already that I feel confident that there will be forces in that case to.

Pardon the ignorance of this non-scientist, but with these results are you suggesting that the drive's thrust is only an interaction with the testing apparatus? In the second quote you say that you are confident that forces will be generated without a vacuum chamber. Does this mean that the drive would produce thrust in free space?

Offline TMEubanks

In these particular runs (which I believe is just the Drive and the chamber) the reactive near field is probably irrelevant, but for modeling the Drive + pendulum suspension it won't be.

Does MEEP support variable mesh sizes? It may be necessary to decrease the mesh size near structures.

And, for that matter, what is the mesh size in these plots? I think the rule of thumb is < lambda / 40, which is a little less than 4 mm.

(Note, by the way, that evanescent, or exponentially decaying, waves are just one part of near field behavior and not all near field waves have to rapidly decay, particularly if you have something like a waveguide set up.)

Offline aero

  • Senior Member
  • *****
  • Posts: 2785
  • 92129
  • Liked: 724
  • Likes Given: 249
While this single run detected only 1/3 of the experimental measured value, I will brazenly write that I think we can forget about axions and dark matter. We can forget about exotic theory and consider that the thrust force likely results from a little understood characteristic of evanescent waves and the forces they generate.
Well, partly perhaps, but lets wait until I run the same model without the vacuum chamber. I have ran enough cases already that I feel confident that there will be forces in that case to.

Pardon the ignorance of this non-scientist, but with these results are you suggesting that the drive's thrust is only an interaction with the testing apparatus? In the second quote you say that you are confident that forces will be generated without a vacuum chamber. Does this mean that the drive would produce thrust in free space?

Lets wait a few days. Then I should be able to answer that with data.
Retired, working interesting problems

Offline aero

  • Senior Member
  • *****
  • Posts: 2785
  • 92129
  • Liked: 724
  • Likes Given: 249
In these particular runs (which I believe is just the Drive and the chamber) the reactive near field is probably irrelevant, but for modeling the Drive + pendulum suspension it won't be.

Does MEEP support variable mesh sizes? It may be necessary to decrease the mesh size near structures.

And, for that matter, what is the mesh size in these plots? I think the rule of thumb is < lambda / 40, which is a little less than 4 mm.

(Note, by the way, that evanescent, or exponentially decaying, waves are just one part of near field behavior and not all near field waves have to rapidly decay, particularly if you have something like a waveguide set up.)

No - meep does not support variable mesh size. I suggested that it be added and was pointed to the source code and invited to program it in, or alternatively to commercial software that does support it.

Needless to say, I'm running meep with a uniform mesh. In this vacuum chamber/EM thruster model that characteristic can be viewed as a limitation because most of the lattice volume is empty so would need only a relatively few points to fully characterize the fields. But meep is free ...
Retired, working interesting problems

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5895
  • USA
  • Liked: 6045
  • Likes Given: 5325
...

As for whether or not meep integrates the equations, I think it is a matter of terminology...
Numerical Integration involves approximating definite integrals by  summing discretized areas.  This is not what the Finite Difference method does. 

Sorry about being perhaps overly rigorous in the use of the word "integrate". I point this (finite? pun-intended  :) ) difference because it is important to understand the convergence problems with Finite Difference methods (as opposed to integration methods like the Boundary Element method, for example, or methods based on variational principles like the Finite Element Method).
What the Finite Difference method does is instead to approximate solutions to differential equations using finite differences to approximate derivatives. 
The idea of a finite difference method is the transformation of a continuity domain to a discrete set of points, the mesh. In every grid point the given differential operator is approximated by a difference-operator.

The issue is that numerically, numerical differentiation is always a much trickier problem than numerical integration (from a convergence viewpoint).

The Finite Difference method is a very old method (references going back to the 19th century) but great progress was made using it during and after World War II, due to the development of the digital computer, due to Von Neuman and Friederichs, mainly due to the Manhattan Project.

At MIT's ASRL very complex Finite Difference codes were developed, for example the PETROS code:
http://bit.ly/1AJ5Vgt in addition to Finite Element and other types of numerical analyses.
« Last Edit: 02/16/2015 04:15 PM by Rodal »

Offline TMEubanks

Dear reactionless

Quote
Pardon the ignorance of this non-scientist, but with these results are you suggesting that the drive's thrust is only an interaction with the testing apparatus? In the second quote you say that you are confident that forces will be generated without a vacuum chamber. Does this mean that the drive would produce thrust in free space?

1.) "are you suggesting that the drive's thrust is only an interaction with the testing apparatus?" I personally feel that that is the way to bet. In any case, such interactions must be exhaustively ruled out before we can consider this to be a "real" (i.e., new physics) effect. This is a small effect (10 micronewtons is the force a 1 mm chunk of ice exerts sitting on your hand), and this will not be trivial.

2.) "In the second quote you say that you are confident that forces will be generated without a vacuum chamber. Does this mean that the drive would produce thrust in free space?" Not necessarily (that is, after all, what we are trying to find out). I did not say this, but I believe that previous MEEP runs did estimate a force for a Drive only run (which was  an artifact, not a real free space force), and that was what was being referred to.

This is not trivial, and fully solving it may be beyond the resources that the amateurs on this forum can reasonably be expected to bring to bear, but it doesn't mean we can't have some fun, and maybe bring some useful things to light, in the mean time. 

Offline Notsosureofit

  • Full Member
  • ****
  • Posts: 656
  • Liked: 704
  • Likes Given: 1367
...

As for whether or not meep integrates the equations, I think it is a matter of terminology...
Numerical Integration involves approximating definite integrals by  summing discretized areas.  This is not what the Finite Difference method does. 
I point the difference because it is important to understand the convergence problems with Finite Difference methods (as opposed to integration methods like the Boundary Element method, for example, or methods based on variational principles like the Finite Element Method).
What the Finite Difference method does is instead to approximate solutions to differential equations using finite differences to approximate derivatives. 
The idea of a finite difference method is the transformation of a continuity domain to a discrete set of points, the mesh. In every grid point the given differential operator is approximated by a difference-operator.
The Finite Difference method is a very old method (references going back to the 19th century) but great progress was made using it during and after World War II, due to the development of the digital computer, due to Von Neuman and Friederichs.

At MIT's ASRL very complex Finite Difference codes were developed, for example the PETROS code:
http://bit.ly/1AJ5Vgt in addition to Finite Element and other types of numerical analyses.

Yes, I remember having a Digital Differential Analyzer (w/ a magnetic drum memory) but replaced it w/ a PDP-4 (Fortran 4K memory) before getting it running.

Offline Reactionless

  • Member
  • Posts: 7
  • Liked: 1
  • Likes Given: 1
1.) "are you suggesting that the drive's thrust is only an interaction with the testing apparatus?" I personally feel that that is the way to bet. In any case, such interactions must be exhaustively ruled out before we can consider this to be a "real" (i.e., new physics) effect. This is a small effect (10 micronewtons is the force a 1 mm chunk of ice exerts sitting on your hand), and this will not be trivial.

Is there any way for this to be controlled for at this point in the experimentation, or does it require a larger or differently configured vacuum chamber (does the GRC have one that would control for this)?

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5895
  • USA
  • Liked: 6045
  • Likes Given: 5325
Including the vacuum chamber requires a lattice size of 26x31 x no-size, so all runs will be 2D with the vacuum chamber included. I tried to run a 3D but ran out of memory before resolution got as large as 10. That is not enough resolution to detect small features of the thruster cavity model and so is not useful for our purposes.

....
Does 2-D here mean a

1) a flat plane geometry (as in a Euclidean x,y coordinate plane), or are you solving

2) the  problem in the circumferential direction by assuming axisymmetry, using Fourier analysis in the circumferential space coordinate?

==> My guess is that the answer is 1) (a flat plane) and that's why it is not possible to produce circular cross-sections of the results

Does 3-D mean

3)a solution in a Euclidean "brick" x , y , z ? or does it mean

4) a solution assuming axysymmetry?

Thanks
« Last Edit: 02/16/2015 03:17 PM by Rodal »

Online WBY1984

  • Full Member
  • *
  • Posts: 176
  • Liked: 117
  • Likes Given: 105
I've been following this thread as a lurker for a little while, but my knowledge is serverely lacking and have a lot of difficulty understanding what is being said. Is it the consensus at this point that it isn't a real effect, but merely experimental artefact?

Offline TMEubanks

Well, you can calculate it, or you can try and devise better experiments (e.g., the same test in a fiberglass chamber). In the end, I suspect it will take a test in space to really be sure, but that's expensive, so it is entirely proper to make it jump through all kinds of hoops here on the ground first. (And, note, it is quite possible that it will either be rejected or just fade away in the process.)

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5895
  • USA
  • Liked: 6045
  • Likes Given: 5325
...
No - meep does not support variable mesh size. ..
That's useful information, as a uniform Finite Difference mesh size implies extremely long computer running times to get convergence in regions of steep gradient of the electromagnetic field.  Otherwise steep gradient regions will not be well modeled.

Offline TMEubanks

Quote
That's useful information, as a uniform Finite Difference mesh size implies extremely long computer running times to get convergence in regions of steep gradient of the electromagnetic field.  Otherwise steep gradient regions will not be well modeled.

This raises the question - do we know if NASA is doing this kind of modeling? Maybe we could get computational resources from NASA.

Offline RotoSequence

  • Full Member
  • ****
  • Posts: 777
  • Liked: 572
  • Likes Given: 780
I've been following this thread as a lurker for a little while, but my knowledge is serverely lacking and have a lot of difficulty understanding what is being said. Is it the consensus at this point that it isn't a real effect, but merely experimental artefact?

It's generally safe to assume that anomalous outcomes are the product of an artifact in the experiment, rather than hitherto undiscovered or unrecognized physical effect(s). Right now, the hope is to produce conclusive proof that the positive thrust reports are either the product of interactions with the testing apparatus, or are the real product of new physics. We don't have a clear answer yet, but we're making headway.  :)
« Last Edit: 02/16/2015 03:33 PM by RotoSequence »

Offline Notsosureofit

  • Full Member
  • ****
  • Posts: 656
  • Liked: 704
  • Likes Given: 1367
... or a new product of old physics

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5895
  • USA
  • Liked: 6045
  • Likes Given: 5325
Well, you can calculate it, or you can try and devise better experiments (e.g., the same test in a fiberglass chamber). In the end, I suspect it will take a test in space to really be sure, but that's expensive, so it is entirely proper to make it jump through all kinds of hoops here on the ground first. (And, note, it is quite possible that it will either be rejected or just fade away in the process.)

Marshall, how about these methods proposed by Mulletron to test whether the EM Drive thrust is due to evanescent wave interaction:

I can think of 3 ways to test the evanescent wave theory.

1) Is the measured thrust the same with the chamber door open and closed?

2) Is the thrust still there when the test article is rolled out of the chamber. Not sure if 2 is possible.....

3) Change the conditions near the resonant cavity; like wrap the thing in thick foam and then wrap all that with foil, see what the thrust does.

...
« Last Edit: 02/16/2015 03:41 PM by Rodal »

Offline TMEubanks

Quote
Marshall, how about these methods proposed by Mulletron to test whether the EM Drive thrust is due to evanescent wave interaction:

Jose, these are all sensible, but it is not clear if they will be conclusive. E.g., if you open the door you get air currents, and thus a different anomalous thrust.

I worry about the mount and pendulum. Can those be insulated? Made of plastic?

Have any tests been done reversing the drive?

Offline TMEubanks

Quote
Change the conditions near the resonant cavity; like wrap the thing in thick foam and then wrap all that with foil, see what the thrust does.

I would definitely recommend something like this be done.

Tags: