A Novel Approach to Developing and Conditioning Part Field Models for a Complex Fluvio-Deltaic Reservoir: An Example from the Wara Formation in the Greater Burgan Field

Abstract

The Wara reservoir is one of the main producing formations of the giant Greater Burgan field. It has been on production under natural depletion for many years. A massive water-flood of this formation has recently commenced. This was preceded by a large-scale pilot water flood the aims of which included enhancing reservoir understanding. This paper describes how historical data, including data from the large-scale pilot, were used to construct representative part filed models. The area of the pilot water flood has significant volumes of data, including core and log data and dynamic data such as pressure transient data, interference tests, tracer tests and cased-hole logs. These provide valuable information for reservoir characterization. The Wara formation was deposited in a tidally influenced fluvio-deltaic environment where sand continuity is complex. There was a desire to develop realistic geological and simulation models that accounted for our understanding of Wara geology and were consistent with the large volumes of surveillance data. A major challenge was the choice of an appropriate area for the part field model. This was chosen so as to allow water influx into the area of interest over the life of the field to be accounted for and to allow relatively simple boundary conditions to be applied. Geological models were constructed using object based techniques. These models used reservoir rock types that were developed to broadly match permeability-height estimates from pressure transient data. The geological models were not guaranteed to account for the sand connectivity inferred form the surveillance data. A streamline based screening technique was used to exclude models that did not broadly capture the interpreted connectivity. Dynamic simulation models were then developed and conditioned to data using conventional assisted history matching techniques. At this stage, some sensitivities related to boundary conditions were explored. Sand connectivity was not varied at this stage. Some examples are given as to how the resulting conditioned models have been used to address questions about expected future reservoir performance. Specifically questions related to proposed well spacing and pattern type are discussed. This paper describes a novel approach to developing models that are geologically realistic and are consistent with the interpretation of reservoir connectivity from a range of surveillance data. This involves using a streamline based screening tool before using assisted history matching techniques. Such an approach can be applied to both part and full field models. The challenges of using such an approach with part field models are described. Some guidance is given to know when it would be appropriate to try to develop and condition part field models.