Look into allowing each refinery its own input file to reduce length of refineries.json. Structure could be something like a file that hold names of all biorefinery json file names and loaded individually from there.