I'd encourage you to use rules of thumb and back-of-the-envelope. Here are some ideas that could be useful:
The problem with any kind of detailed calculations is the number of unknowns: