While we have models from geologists and geo-physicists, I think it’s very difficult to work out the best course of action in each case without a lot of prior e

While we have models from geologists and geo-physicists, I think it’s very difficult to work out the best course of action in each case without a lot of prior examples (typically trial and error as you point out). In modelling oil wells, oil industry engineers typically have lots of variables but far fewer data points (well head pressures, flow rates and hopefully some downhole pressures/temperatures) which makes building a good physics based mathematical model tough. The structure and permeability of the rock matrix is difficult to characterise both before the fracking and after. All situations are different and once you start operations which are expensive, you can’t go back and start again if you make mistakes. There have been considerable improvements in model theory, sensors/instrumentation and computer power in the past decades. Hopefully this can be harnessed to provide cheaper geothermal power for the future as well as managing oil fields more effectively.