tag:blogger.com,1999:blog-560846452637252304.post1462887636465605362..comments2019-02-13T23:47:41.665-08:00Comments on The Polywell Blog: Taking A Stab At SimulationDr. Matthew Moynihannoreply@blogger.comBlogger5125tag:blogger.com,1999:blog-560846452637252304.post-4902124699670112072016-09-19T00:48:19.838-07:002016-09-19T00:48:19.838-07:00This comment has been removed by a blog administrator.Cn Strong Powerhttps://www.blogger.com/profile/13900530635547984206noreply@blogger.comtag:blogger.com,1999:blog-560846452637252304.post-87577478723260497692013-07-21T20:45:37.047-07:002013-07-21T20:45:37.047-07:00Also, I am not an expert on finite difference mode...Also, I am not an expert on finite difference models. Many of the Polywell simulations done by Joel Rogers, Joe Khackan and the Bussard/Krall team were all Particle-in-Cell models. Bussard described a "1.5 dimensional Vlasov-Darwin model" in the early 90's. So, I have heard the term before, but I do not know much about the model, maybe you can explain it to me. Part of what I am so proud of about "The Polywell Blog" is that it explains things in clear language. The easier the posts are for you to read, the longer I took in writing them.<br /><br /><br />Back in March, I was trying to model particle motion. I was using newtons equations of motion to map how the particle moves. Indrek Mare had already looked at this problem and determined modeling it like this would not work. The reason had to do with vectors that were not aligned. Overtime this led to an error between the modeled and real motions which would grow unreasonable. The solution was to use the Runge-Kutta method. I encoded this into Excel and MATLAB, but ultimately had to put the problem on the shelf. I may come back to it. I gave some of the details in this Talk Polywell post.<br /><br /><br />I think there is enough theory work to go around. This includes tackling Riders works. I explained Rider's 1995 paper here. It took me 8 months to write that review and it is not perfect. I have not had time to work on his thesis. I think the polywell is so complex that it is hard to believe that any one person could fully describe it. I include myself in that. I am humbled by the amount of knowledge needed to understand this machine. So, when a theortican claims he has it figured out, it is hard to believe he didn’t miss something. We should just build it and find out.<br /><br /><br />Lastly - This stuff is really hard. Nobody knows if it is going to work or fail. It may fail, we do not know. Because it is so complex, my stance has always been to be open about everything. The goal is to make the idea – its’ strengths and weaknesses available to anyone. Hopefully we can build enough interest to get research really moving.John Smithhttps://www.blogger.com/profile/00648753231923496817noreply@blogger.comtag:blogger.com,1999:blog-560846452637252304.post-26224533091968884662013-07-21T20:40:22.845-07:002013-07-21T20:40:22.845-07:00Jim!
You are the first person to comment in a cou...Jim!<br /><br />You are the first person to comment in a couple of months. I have a couple of comments and questions.<br /><br /><br />First - elliptic integrals. The model in "taking a stab at simulation" relied on the first elliptic integral. I solved this in MATLAB with a simple command - but to do an excel model I needed to use an approximation. I got the approximation from this paper. The integral went into a nasty equation which predicted the x, y and z magnetic field inside the Polywell. That set of equations was pulled from Khachan's 2011 paper. <br /><br /><br />Magnetostatics is a great place to start. If you can do three dimensions you will be ahead of both me and the khackan team. Their latest paper, only did a two dimension simulation. I am working on a review of that paper, right now. Personally, I do not think we need to go three dimensions yet, we can still learn allot more from a two dimensional model. I would love to see a simple Java applet which lets people fuddle with the Polywell parameters to see what kind of behavior pops out. I have described this on Talk-Polywell, but the idea needs development.<br /><br /><br />Ultimately, I think what we need to do is pair simulations with dimensional analysis. There are many dimensionless numbers in plasma physics and the Polywell has got to have several which affect it. It is also likely that the machine has modes of operation. A good starting point is the fusor, which we know has at least 3 modes of operation - Halo mode, Star Mode and Converged Core mode. Simulations allow us the flexibility to model Polywells that we could never build, and eliminate vast chunks of operating space before we ever build a machine. I would not be surprised if the Navy team has gone down this path already - but unfortunately they are not publishing.<br /><br /><br />In doing any code, the trickiest part is getting realistic results. You have to get the code to duplicate a real world test. I have wasted a year or so, working with code which I thought was meaningful, but was actually flawed. That is why in "taking a stab" I used four methods to check my results: Bussard's 2006 paper, simple Biot-Savart math, Excel and finally, MATLAB. I am confident in my code - and I encourage anyone to rip it apart. I encourage you to use to benchmark your stuff.<br /><br /><br />You are using a finite difference model, right? You have the model encoded in some software, right? What is the program that you are running your code in? Certainly excel would be limited - I doubt it would be able to tell you that there is a non-symmetric cloud in the center. I use excel to test simple calculations, to make sure that MATLAB is running correctly. Whenever there is a discrepancy, it leads me to some mistake that needs correcting. It took me about 8 months (off and on) to get excel, Matlab and simple equations to agree.John Smithhttps://www.blogger.com/profile/00648753231923496817noreply@blogger.comtag:blogger.com,1999:blog-560846452637252304.post-77120761840172548922013-07-19T18:20:18.799-07:002013-07-19T18:20:18.799-07:00Hi, thanks for the Polywell Blog forcus on simulat...Hi, thanks for the Polywell Blog forcus on simulations. I have researched several plasma fusion models and have some items that might be of use to you. Additionally, I would be willing to integrate these with your models or just deliver them. <br /><br />1. Fast and accurate Elliptic Integrals (all types, in Excel/VBA, Object Pascal and C., tested! References: BC Carlson: http://dlmf.nist.gov/19#PT3, Asymptopic approximations for symmetric elliptic integrals" http://127.0.0.1:4664/redir?url=http%3A%2F%2Farxiv%2Eorg%2Fabs%2Fmath%2F9310223v1&src=1&schema=2&s=3qJ3xCtXEfF1iiWkH3H5Ts_q7bI.http://apps.nrbook.com/empanel/index.html#pg=309 (but it has an coefficient error which I fixed)," <br /><br />3. I designed these to to calculate the 3d magnetostatic (or electrostatic) ring configuration fields using your parameters. <br /><br />2. I have A slow and crude Excel 'perfect' particle confinement inside a 'porous holey' sphere. <br />I saw non-symmetric compressed cloud circulating about the center.<br />Also observed the critical dependence of confinement on the energy of the particles and the 'ramping' of confinement voltage.<br />I Plan to rewrite to C++/Pascal to check. <br /><br />3. I used a simple 3d FDTD model to calc both static and dynamic EM configuration. I lost the original working model when my computer fried, and I am now rewriting and testing. Missing Planned features: complex PML, spherical modelling and symmetry, geometry specification.<br /><br />4. I am looking for a decent Vlasov-Darwin plasma model to try with PIC, but I can't figure the math out yet. Am focussing on 'cold' plasmas (initial room temp) to evaluate various confinement geometries as plasma temp and density increase.<br /><br />5. I purchased a basic high vacuum, two stage system and HV power supply, but have not assembled it. To test the results of the models, since I could not locate any extensive data sets from the various 'fusor' (IEC,etc) groups.<br /> <br />6. I Also have Wolfram Mathematica available as another tool and am working on a FDTD model to simplify the programming required to evaluate the mathematical models.<br /><br />To my knowledge Rider's critiques have not been addressed by the IEC/Polywell community. <br /><br />Jim Hargis<br />8093 S Oneida Ct<br />Centennial, CO 80112<br />303-220-0253<br />jimhargis@comcast.net<br /><br />jimhargis@comcast.netnoreply@blogger.comtag:blogger.com,1999:blog-560846452637252304.post-24664177400416662832013-02-27T12:18:17.686-08:002013-02-27T12:18:17.686-08:00Hello World!
Want your OWN copy of the MATLAB ...Hello World!<br /><br /> Want your OWN copy of the MATLAB code that was used in this post? Download free copies at:<br /><br />https://github.com/ThePolywellGuy<br /><br />Polywell Posts, Sketch up models, Matlab code! All available for free download! <br />John Smithhttps://www.blogger.com/profile/00648753231923496817noreply@blogger.com