I calculate kerma in air for photons in a given region (like a ionization chamber) and it takes cca 18
hours to achieve an uncertainty of 1% (maybe it is not such a long
time). I tried the LAM-BIAS card with sdum blank (I found it like a
technique to reduce the interaction length), but this did not have any
effect on the result with respect to uncertainty and also the value of
kerma. There is a question whether the lengths of this calculation can
be reduced. It looks like not.
Indeed the LAM-BIAS card has no role at all here, since, as far as photons/electrons are concerned, it acts only on their photonuclear/electronuclear interactions, which are totally irrelevant in your context.
In absolute terms, 18h of CPU is not a long time, albeit for your low energy problem it may appear a lot. But it depends on the required accuracy and your geometry, as well as the electron transport threshold and its relevance. One could consider some geometry related biasing (e.g. region importance biasing), however the related effort may not be worthwhile, considering that you get anyway your result within one day.
Thank you very much.