A Framework for Uncertainty Quantification and Technical-to-Business Integration for Improved Decision-Making

Abstract

Abstract To support the E&P investment decision-making process we use computer models extensively. This paper discusses in a conceptual way the potential merits of moving much more than what is currently being practised into the direction of "fully probabilistic" and "fully holistic" modeling, initially at the cost of model precision. In our opinion, the currently prevailing paradigm of maximum model precision (i.e. more physics, more grid blocks, more detail) may severely limit the optimization of the E&P decision-making process. Evidence for this statement is obtained when analyzing why the average of production and cashflow forecasts generally fails to coincide with the truth, as revealed in time: high-precision models on average result in biased forecasts and, hence, often lead to sub-optimal decisions or missed opportunities. To explain our approach, we introduce two postulates, discuss three modeling dimensions and link this to the business process of decision-making. The first postulate, when modeling the E&P value chain for investment decision-making, is that all uncertainties having a potential material effect on the model outcomes must be quantified within a comprehensive, internally consistent framework, and be taken into account when making E&P investment decisions. The second postulate is that, when trading-off the degree of "model precision" vs. the degree of uncertainty modeling and/or the degree of holistic value chain modeling, the latter two are more important than the former, especially when uncertainties are large. Initially therefore, reduced-physics or "approximate" models may be used, enabling a more comprehensive decision analysis. After having thus optimized the decisions, the sensivity (or "robustness") of the optimal decision to model precision should be tested. Multi-tiered decision-making, real options and an extended definition of "value" are also discussed within the framework's context. Introduction The current debate on uncertainty modeling and decision-making, in our opinion, lacks a clear framework and, hence, lacks the conditions to make substantial progress in the area of improved forecasts and decision-making. A multitude of professional meetings have been held on this theme by a multitude of organizations. The meetings however generally result in limited guidance on how to proceed conceptually and often divert into rather detailed building blocks that fail to demonstrate their relevance within the overall framework of decision-making. While acknowledging the many excellent papers by a variety of authors, we would like to build on their work and present a framework aimed at putting the building blocks together. Since the final objective of our modeling activities is to support the decision-making process, the role of each detailed modeling activity should be understood within this context. The value loop The first step is to understand how "value" is generated. Value is not generated by an E&P company's physical assets only (i.e. hydrocarbon reservoirs). In order to capture the potential value of an asset, data on that asset must be acquired, the data must be processed and interpreted, mathematical models of the assets must be constructed in order to assess the benefits of certain actions (e.g. investments), decision options must be generated, decision criteria must be applied under various constraints, the optimal decision is then implemented, which again results in the intrinsic value being realized and new data being acquired. In this "value loop", the role of models is essential. Models allow decisions to be made and activities to be implemented, thereby generating new data and closing the loop. Let us therefore look more in detail into how we can conceptualise our modeling work.