Geostatistical characterization of a fluvial reservoir system: An example from SouthEast Asia

Abstract

Abstract Significant gas reserves have been discovered in the fluvial reservoirs of Southeast Asia. Economical exploitation of these reserves is complicated because the majority of reserves are very thin and isolated vertically. Therefore a well must be completed in, and produce from, several distinct sands to recover economical volumes of gas. Three dimensional seismic surveys have traditionally been used to locate the most favorable areas for the location of drilling platforms based on the seismic amplitude response of the shallower gas sands. A pilot study was recently performed on one of these fields to determine if geostatistical techniques could further improve the placement of development wells and production platforms. The design of the pilot study recognized specific limitations in both the acoustic properties of the reservoir sands and the quality of the seismic data. Shallow, relatively massive sands are easily recognizable by a characteristic amplitude "bright spot" and often both the tops and bases of the reservoir sands are interpretable on the seismic data. Cokriging and conditional simulation were performed on one of these sands to enhance the estimation of the reservoir volume and to determine the optimum placement of infill wells. Deeper in the stratigraphic section the sands are typically much thinner and not clearly defined in the seismic data. Sophisticated seismic attribute extraction methods were applied to obtain volume measurements that could be compared to net sand over an interval of more than 200 meters. Several maps of net sand were generated using conditional simulation guided by the volume averaged seismic attributes. These realizations were analyzed to produce a map of the probability of sand occurrence above a threshold value. In the deepest portion of the stratigraphic section the reflectivity related to gas bearing sand reservoir becomes indistinguishable from amplitudes associated with shale filled channel systems. Analysis of the data becomes more difficult due to lower frequencies and higher noise levels in the seismic data. The primary aim of the pilot study was to classify amplitude anomalies .into either productive sand reservoirs, or non-productive shale channels. Multivariate discriminant analysis was employed with a training set derived from available well locations to classify anomalies to the most probable lithology. The pilot study was successful and proved that geostatistical techniques added significant information that was not available through conventional interpretation and mapping techniques. The success is largely due to a thorough understanding of the geology, petrophysics and seismic responses of the sand reservoirs. Detailed knowledge of these factors was fundamentally important in being able to design a study that was focused towards important economic objectives that were achievable within the limitations of the seismic data. Introduction This paper documents a geostatistical reservoir characterization pilot project that was applied to a major gas field in SE Asia. Drilling results had shown that the reservoir quality was laterally heterogeneous and traditional mapping methods did not provide accurate estimates of reservoir parameters between wells. The objective of this study was the geostatistical mapping of reservoir parameters in the gas field by integrating petrophysical data and seismic attribute maps from a 3D survey. The study area is roughly 8.5 × 12km and contained more than 33 wells from 3 development platforms. An interpreted 3D seismic survey covered a large portion of the field.