A method for generating geomodels conditioned to well data with high net:gross ratios but low connectivity
Deirdre A. Walsh, T. Manzocchi
Abstract
The sand connectivity in object- and pixel-based models is inevitably controlled by the proportion of sand present, and these methods seem unable to generate models that reproduce systems with low connectivity at high net:gross ratios. A new workflow is described which addresses this limitation and permits poorly connected facies models conditioned to well data to be built. The approach combines the compression algorithm with multiple-point statistics (MPS) modelling. Geometrically transformed wells and appropriately scaled training images provide the inputs to the MPS modelling. The inverse transformation is applied to the resultant MPS model, leading to the creation of reservoir geomodels with realistic, user-defined connectivity while also honouring well data. The approach is described and validated using a range of models. Considerations of other potential workflows using different types of training image suggest that application of the compression algorithm may be necessary in general to achieve models with realistic connectivity using the simplest and most widely-available pixel-based MPS method (the SNESIM algorithm).