Rapid earth modelling for appraisal and development studies of deep-water clastic reservoirs and the concept of ‘procycling’

Abstract

Creating earth models for deep-water appraisal and development studies is perhaps the most challenging task confronting the petroleum geologist today. Data are limited (few wells, limited core, untested seismic quality), time is limited and drilling, testing and facilities costs are very high. Uncertainty in geological characterization of the reservoir can have the greatest potential impact on project value. How can a thorough characterization of reservoir uncertainty be made based on limited data and in a timely fashion? A workflow for creation of suites of models for appraisal and development studies of deep-water reservoirs is described. The goal of the workflow is to rapidly construct suites of earth models based on limited data that capture the full range of uncertainty in reservoir characteristics and properties. After characterizing possible distributions for individual parameters, suites of earth models are built in a single step using an experimental design framework, aided by a powerful workflow manager which automates earth model construction. Earth models created using the experimental design framework are seamlessly linked to flow simulation software. Plackett–Burman, folded Plackett–Burman and Full Factorial experimental designs were used in different appraisal and development cases. Multiple experimental designs were produced by adding and modifying uncertainty parameters as additional data arrived, and ideas about the possible character of the reservoir evolved. In the appraisal case described here, 6 experimental designs were made, 388 earth models were created and studied, and 79 of those models were dynamically simulated. The process of quickly building and re-building suites of earth models using experimental designs to address changing perceptions and concerns about uncertainty in reservoir character is termed here ‘procycling’. Procycling is complementary to experimental design studies, in that multiple experimental designs are employed over time: procycling focuses on changes in predictions made by individual experimental design studies. The results of procycling are not necessarily to change the perception of uncertainty (for example the range of possible outcomes), but to anchor what the limits of uncertainty are and what the most important uncertainties are with the given data.