User:Brian P. Josey/Notebook/2010/05/20
|Project name||Main project page|
Previous entry Next entry
I had some trouble today running my simulations in FEMM. The issue is in how I set up and executed my models. I created three new models where I took four strings of data on two, and two stings of data on the other. Eager to go ahead and calculate away at the field, I drew a box around the areas that I was interested in. The problem was that while the box was small, 0.03 inches square, I designated a minuscule grid size at 5.0 * 10-5. This grid size, I found, is necessary to get smooth data in MATLAB. After a few hours, a couple of crashes, and several load screens, I finally reached the point where FEMM literally told me that it "couldn't solve the problem." With nearly 900,000 nodes, I finally asked too much from FEMM, and I had to redesign it. In place of one large box, I've made two smaller boxes for each of the points that I was taking data at, and cut the number of nodes down to about 250,000. It still takes a while, but the computer can handle it.
The first model I made was a simple iron rod, like a nail wrapped in copper wire with a current passing through it. From this simple electromagnet, I took only two strings of data: one going out from the tip of the nail like all the others ('5-20-10 electro 1') and one going from the perpendicular to the tip just inside of the flow cell ('5-20-10 electro 2'). The strings in parenthesis are my file names so that I don't lose them. I am using this one to serve as the basis by which I will judge the other magnets today.
From this one, I created a second model that included a neodymium magnet with opposite magnetization along side the nail. From this one, I took four stings of data: off the tip of the iron core ('5-20-10 iron 1'), inside the flow cell near the cone ('5-20-10 iron 2'), away from the edge of the permanent magnet ('5-20-10 iron 3') and inside the flow cell at the same point ('5-20-10 iron 4'). Here is a picture of the model:
This one features the neodymium magnet, but when I simulate the iron electromagnet by itself, I simply change the material for the right box into air.
The second model that I created is a little more complicated than the first one. While the first model was a purely theoretical model, this one actually represents a magnet set up that I created in the lab itself. Here is a picture of its real life counterpart:
To create this one, I found a piece of magnetic metal in one of drawers and place the cone magnet in the center, flanked by two of the smaller cylinders that we use to hold up papers. Because I make this models in axisymmetric, the model doesn't actually represent the exact real world counterpart. I set the simulation up so that the magnets are sitting in the middle of the piece of stainless steel across the shortest distance. When this is simulated, it rotates this around the center of the cone, and cuts out the farthest pieces of the stainless steel. This is still an acceptable simulation, as the stainless steel does not contribute anything to the magnetization, and symmetry evens out all of the other issues with the magnitization. Here is a picture of the model in FEMM:
Like I did with the iron and neodymium combination above, I took four data points. For reference the files are: