From: Paul
Date: Fri, 22 Dec 2000 08:10:57 +0000
Subject: Re: [TSSP] Final solution...great chance for it
boris petkovic wrote: > > A mere 50 parts, that would be quite quick - at > > least once > > the Cext and Cint matrices are calculated - that > > takes > > a few hours on the cluster. Once Cint and Cext are > > obtained, with only 50 steps the solution for each > > omega is just a few seconds. > ---- > You're talking about computing times for CW state or > complited transients solved for every case in time > domain? > ---- Yes, that's the time taken to solve the network for AC steady state conditions at a single frequency. > ----- > You know,once I thought that the only way except > Maxwel field theory,for simulating the coils was the > method we are employing today (LC connected network > simulations). > After seeing Bodlovic's work,and heard aboout "dream" > softwares he told me about I changed my mind. > ----- The two approaches are equivalent when the free space wavelength of the frequencies involved are large compared with the dimensions of the secondary. Eg, with Maxwells equ, integral( H.ds) = integral( J.dA) + d/dt integral(D.dA) the approximation allows us to drop the term d/dt integral(D.dA) since we can safely ignore the magnetic induction caused by displacement currents. The resulting integral( H.ds) = integral( J.dA) when combined with another Maxwell equ, integral( E.ds) = - d/dt integral( B.dA) gives the familiar V = L dI/dt. This is the 'quasi-static' approach and it seems very efficient to me to deal with a single coefficient of inductance, rather than the line and surface integrals of the fields themselves. Ultimately, both our and Bodlovic's methods will boil down to a finite element network solution involving a matrix of coefficients which relate the V and I of the elements. The difference I think will lie with how the coefficients are obtained. We use brute force laplace or boundary element method to obtain the coefficents C and table lookup for L. Sounds like Bodlovic has something smarter and more elegant to obtain these. I'll refrain from further speculation until I've seen his paper. > ---- > He gave me also one written in English that > illustrates > basical lines of his approach. > He wasn't too happy when publisher insisted on > transforming his field model into "flash and bones" of > distributed circuit where electrical and magnetical > parts are treated separately. Sounds like that amounts to dropping terms like the d/dt integral(D.dA) mentioned above - ie removing the direct E to B field coupling so that the two fields are independent and only coupled through the action of charge movement on the winding - see intro to pn2511. > Such representation is incomplete,and some results > derived from EM field approach are lost. Yes, eg far field radiation of EM energy - a few tens of mW for a 1kW coil. We can afford to ignore that. We also ignore the slight spiral 'twist' to the B field occuring due to the longitudinal component of current flow. If Bodlovic handles these, his program will be welcome to those constructing electromagnetic containers for plasma experiments! > However ,even the simplification was novelity,and msms > and model were in a good agreement. although his experimental comparisons must have been for particularly simple configurations in view of his limited accounting of external environment. > Give me your snail mail address. Paul Nicholson, Middle Horsewood Cottage, Lumbutts, Todmorden, OL14 6HT England. Thanks Boris, -- Paul Nicholson, Manchester, UK. --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.