Interpretation of well logs and stratigraphic correlation between wells or outcrops are intertwined and several interpretation-correlation loops may be necessary. However, an increasing number of interpreters may generate a larger set of correlations from only one data set. To reduce human induced interpretation bias due to the interpreter, we propose a computer-assisted method to generate stochastic multi-well correlations, used in uncertainty studies. The purpose of this work is to present an algorithmic translation of sedimentary concepts, into cost functions related to each possible correlation. These functions compute correlation costs from reservoir properties measured along well path. Then all these correlation costs are used to populate a cost matrix in order to apply the Dynamic Time Warping algorithm and compute the n-best correlation sets, having the n-least cumulative costs, based on core-sample and well-log interpretations in all wells with respect to their depositional conditions in a chronostratigraphic framework and the chosen rules, e.g. facies, dip, etc. Well markers are points along a well path corresponding to sedimentary facies. These are described by two parameters: (1) the sedimentary facies representing the depositional environment and (2) the relative distality of the well computed from its position along the sediment transport direction. The main principle of correlation, that we use in this method is that a well marker (described by facies and distality) cannot be correlated with another well marker described by a deeper facies at a more proximal position, or a shallower facies at a more distal position.

Look at the video here.