-
#2080 by Rodal on 21 Jan, 2016 16:55
-
More on excitation of TE012: it was so difficult to excite TE012 for NASA, even with COMSOL finite element modeling capabilities, that although NASA got the highest force/power for the TE012 experiment they switched to TM212.
When we asked Paul March in previous threads why did they stop using TE012 he essentially answered that they had lots of trouble being able to excite TE012 on a consistent basis. They were using a loop antenna.
Yes, they did not have an adjustable end, maybe that can help. One wonders how Shawyer and Yang can be sure of what mode they excited since they did not report using any thermal camera measurement...
One of the things meep got right was when we used a single antenna placed like EagleWorks to excite the cavity was a Betty Crocker Blender of rotating modes and poynting vectors. Symmetry is the key to locking in a mode along with the correct injection ports and location.
It puts doubt that the quoted modes of excitement were excited without any verifying data from thermal imaging. Thermal imaging is critical to assure you are indeed exciting the correct modes and backed up by computer modeling.
Shell
Didn't VaxHeadRoom postprocess the Meep results and made movies of your fustrum fed with waveguides?
If I recall correctly the VaxHeadRoom videos confirm the IslandPlaya model of the waveguide-fed fustrum: no TE012 mode shape was shown in the VaxHeadRoom videos.
https://forum.nasaspaceflight.com/index.php?topic=39004.msg1468747#msg1468747Interesting !
So it looks like Shell's experiment should not give a TE013 mode, according to this Meep model. It looks like m=1 rather m=0 as you previously stated...
So Meep results agree with Island Playa model of SeeSheel frustum: m=1 instead of m=0
It looks like the waveguides excite the dominant mode shape with m=1.
QUESTION: Are there Meep output files that I can post-process with Wolfram Mathematica to double-check what mode shape Meep actually predicted for SeeShell frustum with waveguides?[/quote]
-
#2081 by aero on 21 Jan, 2016 18:15
-
[/quote]More on excitation of TE012: it was so difficult to excite TE012 for NASA, even with COMSOL finite element modeling capabilities, that although NASA got the highest force/power for the TE012 experiment they switched to TM212.
When we asked Paul March in previous threads why did they stop using TE012 he essentially answered that they had lots of trouble being able to excite TE012 on a consistent basis. They were using a loop antenna.
Yes, they did not have an adjustable end, maybe that can help. One wonders how Shawyer and Yang can be sure of what mode they excited since they did not report using any thermal camera measurement...
One of the things meep got right was when we used a single antenna placed like EagleWorks to excite the cavity was a Betty Crocker Blender of rotating modes and poynting vectors. Symmetry is the key to locking in a mode along with the correct injection ports and location.
It puts doubt that the quoted modes of excitement were excited without any verifying data from thermal imaging. Thermal imaging is critical to assure you are indeed exciting the correct modes and backed up by computer modeling.
Shell
Didn't VaxHeadRoom postprocess the Meep results and made movies of your fustrum fed with waveguides?
If I recall correctly the VaxHeadRoom videos confirm the IslandPlaya model of the waveguide-fed fustrum: no TE012 mode shape was shown in the VaxHeadRoom videos.
https://forum.nasaspaceflight.com/index.php?topic=39004.msg1468747#msg1468747Interesting !
So it looks like Shell's experiment should not give a TE013 mode, according to this Meep model. It looks like m=1 rather m=0 as you previously stated...
So Meep results agree with Island Playa model of SeeSheel frustum: m=1 instead of m=0
It looks like the waveguides excite the dominant mode shape with m=1.
QUESTION: Are there Meep output files that I can post-process with Wolfram Mathematica to double-check what mode shape Meep actually predicted for SeeShell frustum with waveguides?
I saved the raw data on my new backup drive. Generating the csv files can be quick if the run can be identified. Usually via the log file posted on Google drive but if the .png's have been separated from the log file identifying the run becomes problematic. -
#2082 by Rodal on 21 Jan, 2016 18:20
-
...
I saved the raw data on my new backup drive. Generating the csv files can be quick if the run can be identified. Usually via the log file posted on Google drive but if the .png's have been separated from the log file identifying the run becomes problematic.
Please let me know if you are able to do it. If you can get the csv output files for SeeShells with both waveguides. Then I could write a Mathematic algorithm to output the results the same way that FEKO is outputting (notice that FEKO outputs only positive values so it is just outputting the norm, which is easy to calculate). ( https://en.wikipedia.org/wiki/Norm_(mathematics) )
I bet that Meep results are correct and that this is just an issue of interpretation of the mode shapes -
#2083 by aero on 21 Jan, 2016 18:30
-
...
I saved the raw data on my new backup drive. Generating the csv files can be quick if the run can be identified. Usually via the log file posted on Google drive but if the .png's have been separated from the log file identifying the run becomes problematic.
Please let me know if you are able to do it. If you can get the csv output files for SeeShells with both waveguides. Then I could write a Mathematic algorithm to output the results the same way that FEKO is outputting (notice that FEKO outputs only positive values so it is just outputting the norm, which is easy to calculate). ( https://en.wikipedia.org/wiki/Norm_(mathematics) )
I bet that Meep results are correct and that this is just an issue of interpretation of the mode shapes
Well, first of all, SeeShells, can you identify for all to see, my Googe drive data set that we are considering? -
#2084 by Tellmeagain on 21 Jan, 2016 18:34
-
Whatever you DIYer's do to model the forces on the cavity walls, I think it is important to know before hand that you can not use any simulation to calculate a small "net thrust" by subtraction two big forces, one on the forward direction and the other on the reverse direction, because of numeric errors. When you subtract two similar big numbers each with numeric errors, the net result is often just numeric error. An easy example is to measure your weight twice. The first time you measure your weight when standing on the scale upright. The second time you measure your weight standing on your hands. Now subtract the two numbers. What is the net weight of 10 or 100 micro Newton or even 0.1 pounds you get? The Yang simulation had this mistake.
-
#2085 by SeeShells on 21 Jan, 2016 18:39
-
...
I saved the raw data on my new backup drive. Generating the csv files can be quick if the run can be identified. Usually via the log file posted on Google drive but if the .png's have been separated from the log file identifying the run becomes problematic.
Please let me know if you are able to do it. If you can get the csv output files for SeeShells with both waveguides. Then I could write a Mathematic algorithm to output the results the same way that FEKO is outputting (notice that FEKO outputs only positive values so it is just outputting the norm, which is easy to calculate). ( https://en.wikipedia.org/wiki/Norm_(mathematics) )
I bet that Meep results are correct and that this is just an issue of interpretation of the mode shapes
Well, first of all, SeeShells, can you identify for all to see, my Googe drive data set that we are considering?
Sorry I've been busy.
I believe it was this one we ran.
https://drive.google.com/drive/folders/0B1XizxEfB23tUWlaQjJRRkVSOWs -
#2086 by Rodal on 21 Jan, 2016 18:52
-
Whatever you DIYer's do to model the forces on the cavity walls, I think it is important to know before hand that you can not use any simulation to calculate a small "net thrust" by subtraction two big forces, one on the forward direction and the other on the reverse direction, because of numeric errors. When you subtract two similar big numbers each with numeric errors, the net result is often just numeric error. An easy example is to measure your weight twice. The first time you measure your weight when standing on the scale upright. The second time you measure your weight standing on your hands. Now subtract the two numbers. What is the net weight of 10 or 100 micro Newton or even 0.1 pounds you get? The Yang simulation had this mistake.
1) Concerning:Quoteyou can not use any simulation to calculate a small "net thrust" by subtraction two big forces, one on the forward direction and the other on the reverse direction, because of numeric errors... What is the net weight of 10 or 100 micro Newton or even 0.1 pounds you get? The Yang simulation had this mistake.
Yang calculated all the forces, including the side forces on the conical walls (see attachment below).
One should not conclude that Yang calculated a force based on only the forces at each end, since she did include the effect of the forces on the side walls. (*)
That is not "the mistake" that Yang made.
2) While it is correct to state " When you subtract two similar big numbers each with numeric errors" if you care to calculate the forces due to the Maxwell stresses at both ends you will see that the forces are different to within accurate precision, and this necessarily follows from conservation of momentum, since they are balanced by the forces on the side walls and the time rate of change of the Poynting vector, as well as any currents on the metal walls.
The classical conservation of momentum and conservation of energy equations for the EM Drive transient solution give unequal forces at the ends and no-self acceleration if correctly applied:
______________
(*) I think that confusion stems from the fact that in early papers by Yang she always started by summarizing Shawyer's theory where Shawyer only takes into account the forces at the ends. The fact that Yang summarizes Shawyer's theory does not mean that she used or endorsed Shawyer's theory, as examination of her equations (see attachment below) shows that Yang did not use Shawyer's formulation
-
#2087 by Tellmeagain on 21 Jan, 2016 19:00
-
Yang calculated all the forces, including the side forces on the conical walls (see attachment below).
It is incorrect to conclude that Yang calculated a force based on only the forces at each end, since she did include the effect of the forces on the side walls.
That is not "the mistake" that Yang made.
I did not say she only calculated forces on each end plate or end wall. What I said was "each direction". Her mistake was in subtracting two similar big numbers each with numeric error. -
#2088 by Rodal on 21 Jan, 2016 19:03
-
You may think that that is her mistake, but there is no evidence of that. Instead there is evidence of her only calculating and showing the Maxwell stress terms instead of calculating the time-dependent terms as also being forces.Yang calculated all the forces, including the side forces on the conical walls (see attachment below).
It is incorrect to conclude that Yang calculated a force based on only the forces at each end, since she did include the effect of the forces on the side walls.
That is not "the mistake" that Yang made.
I did not say she only calculated forces on each end plate or end wall. What I said was "each direction". Her mistake was in subtracting two similar big numbers each with numeric error.
If you calculate the forces you will see that the forces at the ends (derived from Maxwell's stress) must be different, and that this can be calculated with precision. Her mistake instead is in not taking into account the contribution of the other terms, like the time dependent terms in the conservation of momentum equation, as giving a net force.
Notice:
Yang only calculates force from Maxwell stress terms.
Fetta only calculates the Poynting vector
Instead of calculating the total force, which is zero.
This is the balance, as per conservation of momentum:
Observe that, in a transient problem where the time derivative of the Poynting vector with respect to time is not zero, the Maxwell stress gradient must be non-zero, in order to balance the momentum
Hence in a transient problem you can have a non-zero time rate of Poynting vector, and hence a non-zero stress gradient, and yet zero-self-acceleration, because they balance each other.
Applying the divergence theorem, the surface integral of the time derivative of the Poynting vector also behaves like a force, which balances the force from the Maxwell stress tensor contribution.
or:
Yang only calculates the right hand side of the above equation. The left side also acts like a force, and it balances the right side
When you account for both the left and the right side, you get a perfect zero
As long as the EM Drive is considered as a closed system that does not interact with other fields. -
#2089 by VAXHeadroom on 21 Jan, 2016 19:59
-
QUESTION: Are there Meep output files that I can post-process with Wolfram Mathematica to double-check what mode shape Meep actually predicted for SeeShell frustum with waveguides?
[/quote]
I have the data files I generated, but they're 96GB just for the H files...
If you want a subset of the CSV files (slices) from that data I can gen whatever you need.
-
#2090 by Rodal on 21 Jan, 2016 20:03
-
QUESTION: Are there Meep output files that I can post-process with Wolfram Mathematica to double-check what mode shape Meep actually predicted for SeeShell frustum with waveguides?
I have the data files I generated, but they're 96GB just for the H files...
If you want a subset of the CSV files (slices) from that data I can gen whatever you need.
[/quote]
Too big for me
How about if we start by please giving us the links to your movies SeeShell's frustum fed with waveguides ?
I'm not sure I got the right ones/the latest ones/ the best ones you generated
They may solve the interpretation problem just by looking at them again
Thanks
-
#2091 by aero on 21 Jan, 2016 20:08
-
QUESTION: Are there Meep output files that I can post-process with Wolfram Mathematica to double-check what mode shape Meep actually predicted for SeeShell frustum with waveguides?
I have the data files I generated, but they're 96GB just for the H files...
If you want a subset of the CSV files (slices) from that data I can gen whatever you need.
I found the data on my system but that run was made in November. That is important because (as best I can tell) the copper model was still the old erroneous one. The mode shapes may be unaffected but I wouldn't trust that myself.
I know that the copper model was updated before I supplied a model to VAXHeadroooom, so his data is correct for the run he made. Only question is, is that the configuration you want to see? And will you need both E and H files?
And the 96GB is the size of the raw data files, that is all possible slices in all directions for every saved time-step. csv file slices would be small enough upload. Of course you need to identify again the slices of data you need to see. -
#2092 by VAXHeadroom on 21 Jan, 2016 20:18
-
QUESTION: Are there Meep output files that I can post-process with Wolfram Mathematica to double-check what mode shape Meep actually predicted for SeeShell frustum with waveguides?
I have the data files I generated, but they're 96GB just for the H files...
If you want a subset of the CSV files (slices) from that data I can gen whatever you need.
Too big for me
How about if we start by please giving us the links to your movies SeeShell's frustum fed with waveguides ?
I'm not sure I got the right ones/the latest ones/ the best ones you generated
They may solve the interpretation problem just by looking at them again
Thanks
[/quote]
This is the one with the highest spatial resolution AND the highest time resolution.
The full volume of the data is 390x327x291 and the frustum occupies about the middle 75% (using meep's set_res to 275)
The time is 112 frames across one wave, or about 3.2 degrees of phase.
This is only two horizontal slices. I have all the data, but figuring out a way to be able to show more than this - well - it got too messy to look at. If there is interest, I could gen an animation from this same dataset but with the two end plates instead of horizontal slices...
-
#2093 by Rodal on 21 Jan, 2016 20:22
-
QUESTION: Are there Meep output files that I can post-process with Wolfram Mathematica to double-check what mode shape Meep actually predicted for SeeShell frustum with waveguides?
I have the data files I generated, but they're 96GB just for the H files...
If you want a subset of the CSV files (slices) from that data I can gen whatever you need.
I found the data on my system but that run was made in November. That is important because (as best I can tell) the copper model was still the old erroneous one. The mode shapes may be unaffected but I wouldn't trust that myself.
I know that the copper model was updated before I supplied a model to VAXHeadroooom, so his data is correct for the run he made. Only question is, is that the configuration you want to see? And will you need both E and H files?
And the 96GB is the size of the raw data files, that is all possible slices in all directions for every saved time-step. csv file slices would be small enough upload. Of course you need to identify again the slices of data you need to see.
OK I had forgotten all this stuff we have to go through
I need
* E field in all 3 directions
* H field in all 3 directions
* definitely the small end (assuming that the waveguide is entering near the big base, and we use the small base to tell what whether it is TE01p or TM11p)
That could be enough, if (and only if) I don't need to check that a) the the small end is at the correct location, b) what is "p" (whether p=1 or 2 or 3)
Otherwise I will also need
* a longitudinal plane, in order to verify that the small end is at the correct location and
* to verify "p", that p=2 and we have TM112 instead of TM113 for example -
#2094 by Rodal on 21 Jan, 2016 20:24
-
QUESTION: Are there Meep output files that I can post-process with Wolfram Mathematica to double-check what mode shape Meep actually predicted for SeeShell frustum with waveguides?
I have the data files I generated, but they're 96GB just for the H files...
If you want a subset of the CSV files (slices) from that data I can gen whatever you need.
Too big for me
How about if we start by please giving us the links to your movies SeeShell's frustum fed with waveguides ?
I'm not sure I got the right ones/the latest ones/ the best ones you generated
They may solve the interpretation problem just by looking at them again
Thanks
This is the one with the highest spatial resolution AND the highest time resolution.
The full volume of the data is 390x327x291 and the frustum occupies about the middle 75% (using meep's set_res to 275)
The time is 112 frames across one wave, or about 3.2 degrees of phase.
This is only two horizontal slices. I have all the data, but figuring out a way to be able to show more than this - well - it got too messy to look at. If there is interest, I could gen an animation from this same dataset but with the two end plates instead of horizontal slices...
[/quote]
Thank so much for going through this trouble
That is a movie for the longitudinal plane
Looks like "p" =2
Do you have a movie showing the surface of the small end, looking perpendicular to the surface, so that all you can see is a circular disk ? -
#2095 by glennfish on 21 Jan, 2016 21:08
-
I agree Doc, the Fockers
have a lot to show once a responsible person starts learning the ins and outs. This will take more than a few days to understand and generate useful data. Agreed, don't think there are any models for stress forces directly...whodathunkit? Should the emdrive enigma be resolved, new factors would have to be added to standard electromechanical models. I nominate you to lead that effort (should this whole project turn out to be authentic). 
There is also notsureofit's hypothesis...I'd love to see that coded in an online software package we can tweak as we move forward. I got the domain http://rfdriven.com for Glenn to play with on his own servers, he might be more than happy to add a page or two for this on-line calculator.
WarpTech can hopefully return and add his own mix to the emdrive software. TT has his own spreadsheet...perhaps that's a decent place to start.
I'd also welcome DeltaMassRev2 back to the forum as I always enjoyed his "anti-emdrive" critiques and chat about floobie sticks.
Java & PHP are freely available
Pearl is available, but I have to install it
Python is not supported by the domain. SSL issues.
spreadsheets & such are easily uploadable.
Within these constraints, ask and thee shall receive. -
#2096 by aero on 21 Jan, 2016 21:18
-
Here is a .gif from my first successful run of Tajmar's cavity. Successful? Well, at least the geometry is correct. The time domain images indicate that I need something more so I did not check resonance, nor did I run the copper model.
This cavity slant height is 68.6 mm, big and small diameters 108.2 and 77 mm. There are a number of things I could try but the debug runs I've made tell me it is something other than increasing the length. Perhaps my modelled coupler and cavity feed dimensions are wrong. Wave guide is a WR 340 so I know its dimensions. I've attached a profile view of the cavity but note that because the picture of the real cavity includes an adjustable small end plate internal to the frustum cone, the height of the cone looks much shorter in my model than in the photograph. To see that the height is probably modelled correctly, you can compare the height of the cone to the height of the WR 340 waveguide which is 86.36 mm.
Has Tajmar hinted to anyone that he used a z-choke or iris inside his feed, and if so, at what point?
-
#2097 by Rodal on 21 Jan, 2016 22:31
-
Here is a .gif from my first successful run of Tajmar's cavity. Successful? Well, at least the geometry is correct. The time domain images indicate that I need something more so I did not check resonance, nor did I run the copper model.
1) Thank you so much for going through all this trouble
This cavity slant height is 68.6 mm, big and small diameters 108.2 and 77 mm. There are a number of things I could try but the debug runs I've made tell me it is something other than increasing the length. Perhaps my modelled coupler and cavity feed dimensions are wrong. Wave guide is a WR 340 so I know its dimensions. I've attached a profile view of the cavity but note that because the picture of the real cavity includes an adjustable small end plate internal to the frustum cone, the height of the cone looks much shorter in my model than in the photograph. To see that the height is probably modelled correctly, you can compare the height of the cone to the height of the WR 340 waveguide which is 86.36 mm.
Has Tajmar hinted to anyone that he used a z-choke or iris inside his feed, and if so, at what point?
2) Since Tajmar did not provide any other information and unless somebody from the Tajmar team comes to NSF to give the needed information, perhaps SeeShells can provide the choke size she considered to use for her waveguide feed and you could use the smallest size she found. Why the smallest size? Because Tajmar's cavity is so tiny, and we know that in order to feed a cavity the entry hole dimension has to be much smaller than the length of the side wall being entered. -
#2098 by ScottL on 21 Jan, 2016 23:09
-
Java & PHP are freely available
Pearl is available, but I have to install it
Python is not supported by the domain. SSL issues.
spreadsheets & such are easily uploadable.
Within these constraints, ask and thee shall receive.
From a simple web app point of view, PHP will work, however; it is an abomination of a language and I loathed to recommend it in professional work. Java has a large overhead and the JVM can be kinda clunky, but it'll do the job. That being said if you were going full fledge web app development, I'd pick something like Play (Java without the overhead) or go straight to Microsoft's ASP.NET MVC 5/6, or either one's API foundation with responsive UI via angularjs (assuming professional work). If it's not that serious though, by all means stick with PHP. -
#2099 by rfmwguy on 21 Jan, 2016 23:28
-
This is definitely a NEW (1/21/16) Russian video about EMDrive, nothing other than a chat. Perhaps a Russian language reader can summarize this for us. Part of the video description translates to:
Shestopalov Anatoly Ivanov and Mikhail Y. after the workshop ( discussions and talks on the sidelines )
Report Ivanov MJ , Mamayev VK, Serov YL, Yanovsky LS " Combustion and detonation
