Robust pre-drill pore pressure models are typically obtained using offset well data and seismic velocities. Recent feedback from the industry seems to suggest that there is no general consensus on the impact of uncertainty on well design in terms of risks and costs, and the methods by which this uncertainty may be quantified. We present here a review talk aimed to highlight the understanding of, and the reduction of, this uncertainty during well planning.
The first task is to recognise where uncertainty can occur, that is, which inputs are likely to have the most significant effect if associated uncertainties are propagated though the model. The model inputs can be a specific data type i.e. density or an interpretation of data i.e. a fluid gradient or overburden, each with an intrinsic uncertainty of differing magnitude.
A secondary challenge is that the choice of pore pressure algorithm itself can impact or constrain the final pore pressure model simply due to the mathematical nature of the relationship used. This is particularly true of shales, where “human” influence such as the definition of an NCT can introduce additional bias.
Finally, what is the applicability of standard statistical techniques to the uncertainty process, for instance the Monte Carlo simulation? Can we accurately compare different models?
The discussion will include a review of scenario modelling, standard deviation, and basin modelling approaches, and this talk will conclude by introducing a potential way forward, one where we could consider both data-driven uncertainty in conjunction with a more process-driven approach.
The ultimate aim is to help us make operational decisions based on uncertainty analysis.