Extrapolation assumes things which may not be true and
this can only be found by experiment. Somebody else's
preferably as such things cost time and money.
In the case at hand I'd wonder / worry about competing
failure mechanisms, #2 passing #1 as pulsed heat changes
from averaged and distributed across the close-in thermal
mass to localized adiabatic as some secondary effect like
runaway rolls on, etc. Linear, in the face of so much square
law, exponential, feedback loop type stuff when you push
material / physics limits, seems like a bad bet. So "No"
I'd call "No" but "Yes" I'd say "maybe" to any analysis.
Besides, blowing stuff up is more fun than your average
multiphysics FEA not to mention cheaper and quicker.