Pre-drill pore pressure models are typically obtained using offset well data and seismic velocities. The workflows used to generate these pressure models are, actually, relatively straightforward and the algorithms used, not complex. However, this is deceptive as there are very many other factors, not just simple-to-understand or visualise issues such as data quality etc, that can mean that the results produced are inaccurate, and at worst, potentially dangerous. One of keys questions therefore is how to reduce the overall uncertainty such that the final pore pressure range is as small as possible. Perhaps the ultimate achievement would be to say what the odds were of the highest modelled pressures “coming in”, rather than saying it’s “possible”.

This paper therefore will aim to explain and define meaningful uncertainty, whether this is purely in data itself, the choice of algorithms we choose and investigate how or if statistics can play an important part. An important conclusion however, is that if statistics are to be used, they need to be sensibly applied, and then this approach needs to be run in parallel with a geological approach. The two together can be adding confidence that “you have done the best possible job”.