there are three methods that work well, all require a lot of work. But, I haven't done astro photography. There's another discussion here somewhere on extending ISO range for astrophotography with some good links.
Following are mostly my interpretive terms; for this exercise, exposures = layers.
1. Average-down blending, probably most film response like.
bottom (first) layer at 100% opacity, all successive layers' opacity at 100/(1+N layers remaining); i.e. each layer is contributing 25% of remaining exposure value.
therefore, a 10 layer opacity % stack is bottom layer-100 50 33 25 20 17 14 12 11 10-top layer
It might not be logical until either you try it or try it with film, essentially it replicates how film grain layers become correctly converted with successive, planned, exposures. Except with digital layers, you can't start with a transparency.
2. True film multiple exposure [log 2x] either adjusting opacity or exposure values of layer stacks. Column 1 is exposure #'s <--> column 2 is exposure compensation (f stops):
1 and 2 work OK up to about 10 layers, then it becomes a lot of work for additional layers' minimal contributing results.
Supposedly all dslr manufacturers with in camera multiple exposure mode use same proprietary software reportedly using one of above techniques, probably a bit more elegant in its checking and self-adjustment.
3. Even opacity %age black layer blending: the only method to stack high multiples, 20 - 100 layers, simulating multiple exposures.
Essentially, each layers' opacity % is (100/N layers).
But, you start off with a transparency, so the first or bottom layer must be opaque black: nothing else works, I've tried. A levels adjustment layer is required to finish as the histogram is often compressed to the left, but it's deceiving as the high multiple layers have a lot of dynamic range that jumps out with startlingly flexible adjustability.
I know there are several applications that will automate this. I tried one quite some time ago and it was terrible compared to the 100% adjustability of every non-destructive step in all methods I describe above. Maybe they're better now?
The learning curve is a bit steep, but it's a lot better than some I've seen.
I've been making a library of dark, flat, and bias files for use in processing. One thing to note is that you'll need to review these files for good quality before using them in a batch.
Also, don't be afraid to save the files and come back to them later for better processing techniques.
Lastly, the output files from DSS are 16 or 32 bit TIFF. These can be a bit difficult to handle. I've recently been turned on to Pix Insight LE as a PP tool. This seems to work really well for me (once I learned how to use it). It's a free tool even though it's a bit old. Since you have Lightroom, you might be able to handle the deep bit files already.
I've been rather pleased with my K10D for taking images of the night sky, especially with the summer eye candy. Good luck with your own efforts!