Think I've got this figured out. Process should run as follows:
1. Buy minerals.
2. Haul minerals to compression station.
3. Work out how many units of 425mm railgun 1 you could build with the tritanium.
4. You will be short some other minerals. In a spreadsheet test, 16.8 billion isk of minerals for building a set of capital ships was short 19.2 million pyerite, 7.8 million isogen, and 850 megacyte. These missing minerals are supplied from a stock in the compression station. Record the quantity of minerals you took from that stock.
5. Build the 425mm railgun I. In my spreadsheet test, it took 3.5 days to build 8350 units using 10 manufacturing slots.
6. Jump the railguns (209,000 m^3) and remaining uncompressed minerals (35,000 m^3) to lowsec.
7. Reprocess railguns.
8. Remember the minerals you took out of the stock in the compression station to build the railguns, and recorded the quantities of? Set those same quantities aside to take back to the compression station.
9. Move the rest of the minerals from the reprocessing station to the building station.
10. Move the minerals you set aside back to the compression station to be re-used.
If it works like my spreadsheet says it will, this will cut my jump freighter jumps on a 10-ship manufacturing run from 30 to 1, saving me around 2 hours of effort per run and substantially increasing my profit-per-hour-at-the-keyboard. The railgun blueprints are in research now, and I anticipate starting mineral compression in about 2-1/2 months.
By the way, the waste when compressing 855 million units of tritanium, at railgun ME 525, is 159k units of tritanium -- something like one-fortieth of one percent.