Geology realism control in automated history matching

This paper proposes the method of an automatic history matching (HM) with the ability to preserve geological realism and an example of its application in one of the fields in Western Siberia. The method assumes takes into account all identified uncertainties at the early stages of geological model construction and their synchronized variation within realistic limits during HM. There are interrelations between petrophysical and geological uncertainties, which significantly effect on the reservoir dynamics. Hence, changes in one of the parameters during HM should influence the synchronous change of the others in order to preserve the geological consistency of the simulation model within the given geological concept. For example, changes in parameters for porosity calculations leads to changes in parameters for permeability calculations, other model characteristics that related to permeability and so on. Schematically the process of calculating one iteration of petrophysical parameters on the example of dependency of porosity and the SP-log can be described in the following way. After identifying the uncertainty limits based on hard data, the regression line is selected. Natural noise is added by the triangular distribution. As the result, the porosity log is built with all the values within realistic limits. However, there is not single probable dependence on the SP. Dependency type is determined by the changing values of the controlling variables (boundary points of the selected regression line) for each of the identified dependencies in the process of automated HM. Iterative algorithm evolution strategy is used for automatic HM. During HM objective function based on mismatch of calculated and historical data for 50 well of sector model of real field is minimized. The distinctive feature of proposed method is the rejection of single deterministic relationships between petrophysical and geological parameters in favor of variations of these parameters within identified uncertainty range. This method allows speeding up HM and providing control of geological realism of simulation models. As a result, confidence in forecasting based on a set of adapted and geologically realistic models is increased.