Abstract
The model maturation method proposed by Joosten et al (Joosten et al, 2011) was applied to a large, complex and mature Niger Delta reservoir system aimed at obtaining a range of acceptable history-matches as basis for evaluating the feasibility of water injection. The method involved forcing gridblock level changes in reservoir properties to obtain model matches for observed field performance. These changes were then used as feedback to improve/update the geo-thinking (understanding of the subsurface).
Of the various model maturation components/types in use, a combination of Experimental Design (ED) with the Adjoint methods of assisted history matching was employed. The first stage involved the assessment of the initial uncertainty space using ED techniques. This resulted in the identification of the most significant uncertainty parameters with respect to key history-match criteria and led to a reduction in the size of the uncertainty parameter space. This reduced (more manageable) space formed the input into the Adjoint process which essentially involved the (automated) identification of the location and extent of changes in gridblock permeability required to improve the model response vis-à-vis production performance.
Despite previous examples of the successful application of this approach to integrated reservoir modelling (Joosten et al, 2011), results in this case showed that it was not entirely possible to fully account for all the forced permeability changes from both a subsurface and surface perspective. A number of reasons were identified for this outcome. Chief amongst these is the lack of a full understanding of the dynamic behaviour of the (complex) faulting in the field. In the end, a compromise position was adopted where the extent and location of changes in gridblock permeability was restricted based on observed well log information. This position was adjudged to still give acceptable history-match results that could reasonably be used to underpin the water flood project.
