Easy oil is gone: Scientific challenges of high-risk petroleum exploration

Konstantin Osypov & Aime Fournier,1 Schlumberger Ltd2

After you earn a PhD in some advanced science or technology, what do you do with it? One option is to work on problems that are both fundamental and lucrative. For example, in economic geology, we just can't see where the treasure –hydrocarbon reservoirs, say– is buried. And, drilling an exploration well is expensive –unless it pays off. However, can't you probably tell with your eyes shut, just listening to your voice, whether a room is empty or occupied? And can't you probably get a sense of a box's contents, just by handling it? Similarly, just by remote sensing, we can infer the likely location of reservoirs as follows.3

Perform tomographic experiments (create enormous, explosive acoustic energy on the surface).
Collect and analyze data (refracted, reflected and/or diffracted energy).
Build a mathematical/geophysical/statistical model (spatial functions for attributes like density or stiffness tensor).
Then more problems arise. Even very plentiful data are too few to determine a perfectly accurate, unique, noise-insensitive model: the problem is ill-posed. Even linearized around small model updates and data increments, it is still ill-posed. Model preconditioning is a remedy that constrains update size, smoothness and/or geological plausibility.