From: Paul
Date: Wed, 23 May 2001 17:45:45 +0100
Subject: Re: [TSSP] Genetic optimisation
Hi Bert, All, Bert Hickman wrote: > I only wish I had more time to devote... I wish you had too - things always seem a lot clearer after you've put them into words. Looks like 26kV/cm is the figure to work with at normal TC frequencies, - comes as some relief that the value is not severely frequency dependent, things would be tricky if so! If I've understood things right, the leader formation begins and continues as long as this gradient can be maintained just ahead of the leader. I guess the significant threshold involves meeting this value at the surface of the smooth toroid and once a leader begins to form, its sharp point will then ensure that the leader forms rapidly for quite some distance - even though the 'background' field from the topload would, by itself, fall below the 26kV/cm threshold only a little way from the surface. Subject to the toroid having enough charge available to support that formation. What stops the leader formation? I guess either it hits earth or it runs out of charge - the toroid is depleted and the 26kV/cm cannot be maintained at the tip? So a big toroid would be reluctant to break out (modest surface gradient), but it would throw a long streamer as soon as it did (lots of charge available)? I'm afraid I've got some more questions! > A long spark (>6 cm) is characterized by multiple avalanches in an > evolutionary sequence: streamer flash(es) --> leader propagation > (fed by groups of streamers) --> spark (if leader > bridges the gap). Can we assume that this whole sequence takes place in a timescale short compared with an RF cycle - I suppose thats so because if not there would be a bigger frequency dependence? When the HT falls away, do things recombine and settle down sufficiently that on the next half cycle there is no 'memory' of the previous half cycle? Ultimately what I'm fishing around for is some confidence that some acceptable and realistic account can be taken of the breakout thresholds, otherwise attempts at non-linear time domain modeling will founder on that point. I feel as though we are on top of the technical matter of computing the response and now, quite suddenly it seems, we are up against this more difficult problem of finding a load conductance function which provides an acceptable summary description of the breakout dynamics. If I've got things right, then Terry should be able to calculate quite easily the top voltage at which streamers should suddenly start to form, and we might also be able to calculate an estimate of streamer length too (as a function of topvolts). Given those two separate figures (or functions) we would then have the choice of optimising for max topvolts or max streamer length, using the same genetic software but with two different merit functions. Cheers, -- Paul Nicholson, Manchester, UK. --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.