--> All Models Are Wrong — Some Are Useful

AAPG Annual Convention and Exhibition

Datapages, Inc.Print this page

All Models Are Wrong — Some Are Useful

Abstract

Over the last 20 years, the evolution of basin modeling/petroleum system analysis has been primarily focused on increasing the number of processes and the complexity of the simulations. I will present examples to show that this approach generally have not helped reduce exploration risk, and propose new approaches that do help reduce risk. A common application of petroleum systems models is to create a simulation in which a prospect is charged with the “desired” type and volume of hydrocarbons; however the vast majority of these prospects eventually result in dry holes. In other cases, modeling results may have caused some companies to miss out on the biggest discoveries in the past decade. Due to the significant uncertainties in the required input of these models, there is never an unique answer to the important questions. But due to the emphasis on scientific/technological aspects of modeling such as custom and multi-component kinetics, 3D and 4D heat/fluid flow analysis, “new and improved” gridding algorithms, etc, the finished model gives the illusion of accuracy but not the reality. Indeed, the time and resources need to build complex models encourages the human tendency to psychologically “anchor” on a single scenario. Therefore too few scenarios are tested and some important risks are not considered. The eye-catching graphical output of such models also makes it hard to see that the outcome of the model is usually simply the direct consequence of user choices of the unknowns – fault seal behavior, source facies in the undrilled kitchen, etc. The typical modeling approach centers on the modeling of source rock maturity and timing of expulsion. I will present arguments and examples to show these aspects rarely have much significance in understanding charge risk, and relying on the conventional wisdom may actually lead to incorrect conclusions and decisions. The most important parameters that do directly influence probability of charge and fluid type (gas vs oil) have been largely neglected. These include seal capacities versus carrier bed relief, and UEP (ultimate expellable potential) of source rock versus migration losses. I will demonstrate how these directly influence charge risk and present simple and practical methods to risk these parameters to help improve success rate.