Apologies for the boring technicality of what follows. Skip by all means.
Been doing a bit of research into cartridge brass and it's properties, especially Young's modulus. I think I know know what is happening and why. See the curves below (courtesy Physics Stack Exchange, not for brass, but illustrates the point). When you fire the forces are so great that the brass will go instantaneously through it's elastic limit and passes straight into plastic deformation, that is the curved top bit. That is why you can fire-form cartridges. If it is on the rising part its ok, if it passes over then its all downhill from there until the break point at the end of the curve. Note you have far less latitude once the material has been work hardened, see the second steeper curve. Even though the extension begins later it ends earlier and the curve part falls off the cliff sooner and more dramatically. In a nutshell, reloaded brass needs to be quite closely contained in a tight space or you will eventually push it over the edge of the cliff. The greater the head space, the sooner this may happen, even after one reload in my case.
So what is the solution?
1. Bring the head space back into tolerance if you can.
2. If this is impossible use only new brass so you are always on the pre-worked curve. Then discard the brass.
3. There is the option theoretically to anneal the fired brass to restore it's ductility and get back to something like the first curve before reloading. However this is dicey with a headspace issue because plastic deformation will certainly have occured in the first firing anyway and all of the next deformation event will happen at the thinned point created first time around. Young's modulus is about extension per unit length, and if the thinned band is all of say a millimeter or so, then the same few thou of movement due to head space will be happening at that 1mm - result being a split case.
One more thing I have come to realise is that forward pivoted doubles will have some flex, and of course the top barrel in an O/U sees more of it just taking the geometry into account. Now add the extra mechanical advantage of the top barrel firing event trying to swivel about the pin and you see it exacerbates the problem. Cure? A Greener bolt or some other form of top lock up, rising bite, doll's head, etc. Only use modest calibres in O/U especially small framed ones obviously, and never load a hot charge.
View attachment 383480