Using Measurement Uncertainty to Calculate Reservoir Volumes and Reduce Risk in Prospects

Abstract

This paper's objective is to explore how to better quantify the effect of uncertainties in reservoir interpretation and overcome conventional interpretation limitations. These include generating just one model despite data supporting different interpretations; ambiguities in the seismic data; and difficulties in quantifying uncertainties in static reservoir properties, such as spatial description and volumes. As part of the new method, uncertainty information is collected and paired with an interpreted geologic feature, such as a horizon, fault or contact, thereby more accurately representing the data's limitations and the interpreter's vision for the geologic structure. Through measurement of these uncertainties, seismic interpreters can generate a suite of horizon and fault configurations to calculate probability distributions for static reservoir properties as well as provide more complete constraints on seismic amplitude modelling and inversion. In the initial interpretation stage of the workflow, uncertainties are represented by an uncertainty envelope that changes size based on the interpreter's estimate of uncertainties and where the best-estimate interpretation of a geologic feature and an associated uncertainty are generated. Follow-up stages include applying structural modelling algorithms to construct a geologically consistent model and applying fault uncertainty where a method (Røe et al, 2013) is applied to simulate the position and geometry of faults in the model. The final stage computes static bulk volumes of the reservoir and uses these volumes to compute a posterior probability distribution to reduce risk in the prospect. The new workflow was applied on seismic data from the Norwegian Continental Shelf where uncertainty envelopes were developed and structural modelling algorithms applied. A grid model was then built for each realization from which the interpreter used the uncertainty in the faults to perturb the model and generate volumetric statistics for each realization. One hundred realizations of the structural model were generated and the distribution of the computed bulk volumes plotted and used to directly indicate the P10, P50 or P90 volumes. The paper concludes that this new workflow can successfully quantify the effect of uncertainties in reservoir interpretation and generate volumetric statistics for each realization, thereby providing crucial input into the calculating of reservoir volumes and reducing the risk of prospects.

AAPG Datapages/Search and Discovery Article #90189 © 2014 AAPG Annual Convention and Exhibition, Houston, Texas, USA, April 6–9, 2014