Identifying reservoir depletion patterns from production-induced deformations with applications to seismic imaging

Introduction

Our objective is to study the application of geological and geomechanical constraints to the regularization of seismic inversion problems. Well tie-ins and use of a priori information on sediment geometry are examples of such constraints that have found robust quantitative application in regularized seismic inversion. Although some success has been achieved in attempts to connect subsurface velocities and impedance to poroelastic rock properties (Mavko et al., 2009), the application of such relations to constraining velocity inversion runs into difficulties stemming from uncertainties both in the subsurface parameters and in the choice of a semi-empirical rock model to use. However, displacement and stress measurements are the source of independent information about the subsurface, and in principle may be used for the regularization of seismic inversion problems in areas of poor illumination.

More specifically, in the framework of time-lapse seismic imaging, we are interested in answering the following two questions:

-

Can we robustly estimate production-induced changes in seismic velocities and impedance (with respect to a base-line model) from subsidence measurements?

-

Can we subsequently use the updated model for the regularization of time-lapse inversion where repeat surveys have illumination gaps or are contaminated with noise?

While we discuss the second question in the penultimate section, this paper is mostly concerned with the first question. We begin with an overview of the existing theory of production-induced deformation, then proceed to the formulation of our method for modeling displacements from known pressure drop in a ``slowly varying'' poroelastic medium. A technique for inverting the pore pressure change from measured displacements is described next. The proposed modeling and inversion techniques are demonstrated on synthetic examples based on real-world production-induced subsidence cases. The section on spatial heterogeneity of elastic earth models discusses the impact of heterogeneity on modeling and inversion results and puts forward an extension of the proposed technique for tackling vertically-layered and arbitrarily heterogeneous media. The paper concludes with a discussion of velocity and impedance change estimation from strain, with an outline of a process for picking the appropriate semi-empirical rock model.

Identifying reservoir depletion patterns from production-induced deformations with applications to seismic imaging