Hey cp,
hmmm, that sounds great and truly extraordinary...
I once had the honour to accomplish some pretty fine welding- and brazing trials for the Wendelstein 7X project, which is a thermonuclear reactor experiment, see also:
http://www.ipp.mpg.de/eng/for/projekte/w7x/for_proj_w7x.htmland I can remember well, they had the highest levels of joint-quality I have ever experienced.
O.K. this is an important fact you have described and thus I suppose you have to get approximately the same chemical composition in weld deposit as in the base material.
Interesting task...
As far as I assume you won't - with reservation - find a standardized filler for the base metal you have to process. On the other hand I don't know the geometrical dimensions of the part you are talking about.
Under the presumption the part is a large dimensioned one you would need a welding process having a higher arc power density, e.g. Gas Shielded Metal Arc Welding (GMAW) which however needs a specific coiled wire electrode, having the necessary (standardized) composition.
Gas Tungsten Arc Welding (GTAW) would be an alternative but having - compared with the GMAW - an only lower density of arc power.
Thus I would think that Plasma-Welding may be a possible to use welding process. In combination with a flux - perhaps standardized available(?) - which is improving the wetting-properties and a piece of your original base material it could lead to success. At least I would suppose to attain the most similar chemical composition and thus the most similar properties compared with the untreated base material.
Or you can try to find a brass-filler having a higher zinc-content for compensating the zinc-vaporization which probably will occur, to attain a nearly similar composition. Honestly I have never tried it by myself but from the physics side I guess it could work.
Perhpas has another one of the appreciated fellows in the forum personal experiences with PAW of brass and is thus able to point us in the right direction.
Really interesting matter, cp!
Regards,
Stephan