Estimation and Re-Parameterization of Saturation and Pressure Changes From Time-lapse Seismic Data Based on Statistical Rock Physics Modeling
Abstract
Abstract
Time-lapse seismic data can be used to monitor the fluid displacement and pressure variations during reservoir
production. Production, injection and depletion alter rock and fluid properties, such as saturation and pressure, causing a change in the elastic response of surface seismic waves. These changes can provide valuable information to better understand flow
mechanisms
within the
reservoir
and detect areas where hydrocarbon accumulates. The goal of this work is to present a probabilistic method to quantitatively interpret time-lapse seismic data and estimate changes in
reservoir
properties. First, a rock physics model calibrated at the well location is defined to link the changes in pressure and saturation to their geophysical response. Changes in saturation, at the seismic scale, can be described using Biot-Gassmann's relations, whereas an empirical relation must be introduced to describe the pressure effect. The joint saturation-pressure model requires a set of well logs and laboratory measurements to determine the properties of the
reservoir
rock and fluid components and calibrate the empirical parameters of the model. A statistical approach is then introduced to include measurement model uncertainties. We then propose a Bayesian approach to solve the inverse rock physics problem and estimate changes in saturation and pressure from time-lapse seismic measurements. The mathematical method integrates Bayesian inversion, seismic forward modeling, and the statistical rock physics model. In the proposed method, we combine the likelihood function of saturation and pressure changes given elastic property changes with the probability distributions of elastic changes estimated from time-lapse seismic data. The result of this methodology is a 3D model of the posterior distributions of
reservoir
property changes conditioned by time-lapse seismic data. We finally include a model order reduction method in the inversion workflow. In particular, we apply the proper orthogonal decomposition algorithm to the inverted model of
reservoir
properties and re-parameterize the time-lapse inverted results in order to determine the location of the fluid-front during
reservoir
production. This step allows reducing the dimension of the geophysical dataset from millions of measurements to hundreds of spatial coordinates. The workflow is applied to synthetic data for validation and to a case history in the North Sea.
AAPG Datapages/Search and Discovery Article #90260 © 2016 AAPG/SEG International Conference & Exhibition, Cancun, Mexico, September 6-9, 2016