Factors Contributing to Uncertain Production Forecasts: 3D Reservoir Modelling Perspectives

Abstract

Abstract Finding and developing oil and gas assets has always been a risky business. The industry has a history of technological advances that have helped to reduce the risk. However, risk has not yet been fully reduced due to inherent uncertainties in the workflows used to generate production forecasts of the oil and gas fields using 3D reservoir models. Since, the reservoir properties vary spatially due to reservoir heterogeneities (occur at all scales, from pore scale to major reservoir units), to obtain reasonable production forecasts, an adequate understanding of the limitations imposed by the data, associated uncertainty, or the underlying geostatistical algorithms or approaches and their input requirements for the 3D reservoir models are absolutely necessary. Based on the lessons learned from 3D reservoir modelling studies performed in-house in different projects, available public domain literature, authors' and industry experiences, some of the identified key factors affecting production forecasts are: sparse and non-representative data, biased estimates of Original Hydrocarbon In-Place, non-representative inputs distribution in the reservoir models due to lack of conceptual geologic model, inadequate static and dynamic models, poor use of seismic data, use of improper analogs, non-unique history matching calibration processes for brownfields and inappropriate use of uncertainty workflows and tools. To demonstrate and quantify the impact of different key factors under uncertainty which affect Hydrocarbon-In-Place, recoverable resources and production forecasts, using real field data for a clastic reservoir, a 3D static reservoir model was built using appropriate geo-statistical techniques and closed feedback loop between 3D static and dynamic models. Finally, the results are discussed which indicate that the evolution of modelling process will continue as new techniques/technologies are developed and implemented. This will enhance our ability to capture the physical realities of the real subsurface world, generate better production forecasts to reduce the risk associated with field developments. 1. Introduction The Oil and Gas demand is pushing the oil industry to move to higher risk areas (ultradeep water) to find and develop new assets to support the world energy supply. To a certain extent, the advances in technology have been quite helpful in mitigating some of the key risks, but the risks are not yet fully reduced due to our inability to formulate a complete and precise description, and characterization of the reservoir leading to the uncertainty in our understating of reservoir description (Singh et al, 2009 and 2013). Even though, a reservoir was deposited geologically and evolved into a unique hydrocarbon-bearing entity, it cannot be deterministically defined or completely determined because of subsurface complexity and limited data. The industry is heaviliy using the geostatistical techniques to model the subsurface heterogeneities including the uncertainty of geologic and petrophysical variables to assess their impact in the Hydrocarbon Initially In-Place (HIIP) and recoverable resources (Orellana et al., 2014). There are highly visible efforts in the industry to improve 3D reservoir models which are mainly driven by the E&P business needs, exponential growth in advances of computing, from teraflops (1012 floating point operations per second) to petaflops (1015) since the early 90's and advances in software. Improved parallel networking algorithms have significantly decreased the Central Processing Unit (CPU) run time. These advances have led to an exponential increase in the number of cells of the 3D reservoir models from a few thousands cells in 1990's to billions of cells in recent years and a significant decrease in their cell size from 300 - 600m in 1990's to 5 - 10m in recent years (Singh et al., 2013). The reduced CPU run time for dynamic simulation has significantly reduced or eliminated up-scaling of large size 3D static reservoir models. Higher numerical solution accuracy and flexibility to handle fully integrated Giga-Cell 3D reservoir models, in-turn, has allowed better capture of geological heterogeneities and has improved production forecasts predictability under uncertainty.