4D Petrophysical Seismic Inversion - Case Studies

Four‐dimensional imaging using geophysical data is of increasing interest in the oil and gas industries. While travel‐time and amplitude variations are commonly used to monitor reservoir properties at depth, their interpretation can suffer from a lack of information to decipher the parts played by different parameters. In this context, this study focuses on the slowness and azimuth angle measured at the surface using source and receiver arrays as complementary observables. In the first step, array processing techniques are used to extract both azimuth and incidence angles at the source side (departure angles) and at the receiver side (arrival angles). In the second step, the slowness and angle variations are monitored in a laboratory environment. These new observables are compared with traditional arrival‐time variations when the propagation medium is subject to temperature fluctuations. Finally, field data from a heavy‐oil permanent reservoir monitoring system installed onshore and facing steam injection and temperature variations are investigated. The slowness variations are computed over a period of 152 days. In agreement with Fermat's principle, strong correlations between the slowness and arrival‐time variations are highlighted, as well as good consistency with other techniques and field pressure measurements. Although the temporal variations of slowness and arrival time show the same features, there are still differences that can be considered for further characterization of the physical changes at depth.

Using slowness and azimuth fluctuations as new observables for four‐dimensional reservoir seismic monitoring

Cacqueray

Roux

Campillo

2015

Deep neural network application for 4D seismic inversion to changes in pressure and saturation: Optimizing the use of synthetic training datasets

Côrte

Dramsch

Amini

et al. 2020

In this work, we tackle the challenge of quantitative estimation of reservoir dynamic property variations during a period of production, directly from four‐dimensional seismic data in the amplitude domain. We employ a deep neural network to invert four‐dimensional seismic amplitude maps to the simultaneous changes in pressure, water and gas saturations. The method is applied to a real field data case, where, as is common in such applications, the data measured at the wells are insufficient for properly training deep neural networks, thus, the network is trained on synthetic data. Training on synthetic data offers much freedom in designing a training dataset, therefore, it is important to understand the impact of the data distribution on the inversion results. To define the best way to construct a synthetic training dataset, we perform a study on four different approaches to populating the training set making remarks on data sizes, network generality and the impact of physics‐based constraints. Using the results of a reservoir simulation model to populate our training datasets, we demonstrate the benefits of restricting training samples to fluid flow consistent combinations in the dynamic reservoir property domain. With this the network learns the physical correlations present in the training set, incorporating this information into the inference process, which allows it to make inferences on properties to which the seismic data are most uncertain. Additionally, we demonstrate the importance of applying regularization techniques such as adding noise to the synthetic data for training and show a possibility of estimating uncertainties in the inversion results by training multiple networks.

Bayesian inversion of 4D seismic data to pressure and saturation changes: Application to a west of Shetlands field

Côrte

Amini

MacBeth

2022

A Bayesian inversion methodology is proposed that inverts angle-stacked 4D seismic maps to changes in pressure, water saturation and gas saturation. The inversion method is applied to data from a siliciclastic reservoir in the west of Shetlands, UK continental shelf. We present inversion results for three seismic monitor surveys and demonstrate the added value of pressure-saturation inversion by providing insights into reservoir connectivity and fluid dynamics across 14 years of reservoir production. In these surveys, 4D seismic signals related to waterflood, pressure increase and depletion, and gas exsolution are evident and overlap each other in many regions. To regularize this ill-posed inversion problem, we propose a Bayesian formulation that incorporates spatially variant prior information derived from a history-matched reservoir simulation model and well pressure measurements. The benefit of incorporating these multi-disciplinary data as prior information is demonstrated by comparing to inversion results using a spatially invariant prior. We show that the method takes advantage of the multi-disciplinary prior information to make more precise inferences where the seismic data are most uncertain. This leads to more realistic spatial distributions for the pressure and water-saturation inversion results. The non-uniqueness in this non-linear inversion is studied by analysing uncertainty estimates produced by stochastic sampling of the Bayesian posterior distribution. Posterior standard deviations are observed to be related to the sensitivity of the seismic amplitudes to the changes in each dynamic property as well as the degree of overlap between changes in different dynamic properties. Estimated pressure increases have a posterior standard deviation of approximately 1 MPa, whereas posterior standard deviations for pressure decrease are on average 4 MPa, with higher values applicable to regions of gas exsolution. The posterior standard deviation for estimates of gas saturation change is 0.07 for moderate, visible saturation signals, but 0.005 for low-to-zero gas saturation change. The posterior standard deviation for estimated water saturation change is mainly influenced by overlapping changes in pressure and gas saturation. When water changes dominate the 4D seismic signal, posterior standard deviations are on average 0.05. These values rise to 0.25 in areas where the water change is obscured by pressure or gas-saturation changes.