From: Paul
Date: Sun, 27 May 2001 23:07:32 +0100
Subject: Re: [TSSP] Toroid field stress
Terrell W. Fritz wrote: > I am not the worlds greatest programmer Ah, no one ever is until they get stuck in. You're getting the results, so you're doing OK :) > ...However, another program could setup the input.txt file, run > E-Tesla6, and go get the voltage (with the peak loop added to ET6). > Then it could change the input text file value again and repeat over > and over to scan the range of variables you wish. Yep. Something to generate a heap of input files, according to some scheme designed to cover the planned range of coils and loads. Then a loop to run ET over the heap to produce another heap of output files. Then something which trawls the heap, pulls together the pairs of in and out files, and spits out CSV records into a summary file. Then bring out your favourite plotting package and look at the results from a few angles. Pick out the trends visualy, apply small one-dimensional fits to take out each of the configuration axes, one at a time. > I imagine a simple basic program could do that and run ET6 externally with > the "run" command? I am not sure how to have one program run another > executable externally... However, I am sure it can be done probably really > easily. You'll find it very easy I think, but if you want to fix up ET6 accordingly, I'll script up something to run it over the heap - it can run on my 14 P500 cluster which is quiet for the forseeable, and I'll send you the results. Will need some consideration of the range of input parameters - a few trial runs to establish how sensitive the answers are to things like width and height of coil, etc, might help. > Let me know if I this sounds promising and I'll get ET6 fixed for it. Go for it. > Would probably want to do some testing to find a reasonable grid size too. > Them 600 grids take forever to run! Any improvement you can do now is multiplied significantly when you're faced with running a thousand or so coils. An hour or two's thought can save a week or more of running time. Eg, maybe the answers are relatively insensitive to the presence of the coil and you can just model a floating topload? Never a dull moment around here, eh? Cheers, -- Paul Nicholson, Manchester, UK. --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.