Sampling and Data Collection Strategies for Minimizing Inaccuracy in Predictive Pre-Drill Reservoir Quality Model Simulations

Tobin, Rick¹ (1) BP America, Inc, Houston, TX

Predictive diagenetic models of sandstone reservoirs are increasingly being relied on to help estimate total reserves and fluid deliverability (flow rate) ahead of the drill bit. However, model simulations of key rock properties (porosity and permeability) may ultimately fail to accurately predict observed rock properties in the wellbore because of unrepresentative sample selection and/or incomplete or inaccurate data collection used to build the model. The good news is that highly accurate pre-drill model predictions can be achieved given a logical sampling strategy coupled with generation of a complete and accurate data set.

Sampling strategy should be fit for purpose, and must ensure that petrophysically defined rock facies being analyzed and used in the model statistically represent the net thickness (“net h”) of the defined pay being used in resource calculations (i.e., reserves and flow rates. Sample selection should also avoid the inclusion of petrofacies that are not being used in the net h calculations, or petrofacies that are beyond the current capability of the modeling software being used.

Petrographic and core analysis data collected for modeling input must be as complete and accurate as possible. Data collection protocol should be designed to avoid the following potential pitfalls that can lead to poor model performance: sample handling and damage avoidance, sample preparation technique, sample preparation and analysis under the correct net confining stress, quality of equipment being used, quality of petrographic description and diagenetic interpretations, and thin-section point count analysis accuracy and technique. Detailed examples of strategy issues and subsequent model performance are compared.