Seismic reservoir characterization requires a transform of seismically derived properties such as P- and S-wave velocities, acoustic impedances, elastic impedances, or other seismic attributes into parameters describing lithology and reservoir conditions. A large number of different rock physics models have been developed to obtain this link. Their relevance is, however, constrained by the type of lithology, porosity range, textural complexity, saturation conditions, and the dynamics of the pore fluid. Because the number of rock physics parameters is often higher than the number of seismic parameters, this is known to be an underdetermined problem with nonunique solutions. We have studied the framework of inverse rock physics modeling which aims at direct quantitative prediction of lithology and reservoir quality from seismic parameters, but where nonuniqueness and data error propagation are also handled. The procedure is based on a numerical reformulation of rock physics models so that the seismic parameters are input and the reservoir quality data are output. The modeling procedure can be used to evaluate the validity of various rock physics models for a given data set. Furthermore, it provides the most robust data parameter combinations to use for either porosity, lithology, and pore fluid prediction, whenever a specific rock physics model has been selected for this cause.
Improved reservoir characterization and monitoring can be achieved by combining seismic and controlled‐source electromagnetic techniques. This requires developing coherent rock physics descriptions. In this paper we demonstrate consistent joint elastic‐electrical modelling according to the differential effective‐medium theory. We test our modelling against data from a compaction experiment on a set of 11 sandstone core samples from the same quarry location. The presented approach is analogous to calibrating a rock physics model to a particular reservoir based on data from possible well logs and core samples. For simplicity we choose to use multivariable non‐linear regression in the inversion. It shows that this technique is able to identify solutions that are physically sound. However, a more rigorous inversion method might be considered in future implementations. To identify the critical parameters we test the elastic‐electrical sensitivity of the various unknown variables involved. The most sensitive parameters identified are then perturbed during the modelling. The mineralogy consists mainly of quartz, which we assume to be spherical and kaolinite. We use the resistivity to calibrate the aspect ratio of the clay grains and estimate the porosity reduction due to compaction. These values are in turn used in inverse modelling of the bulk and shear moduli. The solid minerals make up the inclusion material in the differential effective‐ medium modelling for both the elastic and electrical properties. Hence, this formulation constitutes a consistent joint elastic‐electrical modelling scheme. We achieve good fits between the model results and the laboratory measurements for most of the samples. The reason for the less good fit with some of the samples might be due to measurement errors in the laboratory. This is supported by the observed abnormal stiffness compaction trends associated with those samples.
Identifying type of rocks and fluids from seismic-amplitude anomalies can be challenging because of seismic nonuniqueness and rock-physics ambiguities. Lithology and fluid predictions based on seismic properties therefore are often associated with uncertainties. On the Norwegian Shelf, clay-rich source rocks and hydrocarbon-filled sandstones often show similar AVO responses. A seismic screening method based on rock physics enables one to better discriminate between these different facies. This technique is demonstrated on seismic AVO data (i.e., acoustic impedance [AI] and V P /V S ) from the Norwegian Sea. Rockphysics models for organic-rich shales and gas sandstones are calibrated using nearby well data. Then these models are used for predictions of rock parameters away from well locations. From these predictions, the likelihood of presence of organic-rich shales versus gas sandstones can be evaluated, based on a rockphysics approach. However, there are many uncertainties in the accuracy of the calibrated models and the seismic image of the target area. Hence, predictions should be evaluated along with other geologic and geophysical information before firm conclusions about these anomalies are made.
We have developed a procedure for estimating the effective elastic properties of various mixtures of smectite and kaolinite over a range of confining pressures, based on the individual effective elastic properties of pure porous smectite and kaolinite. Experimental data for the pure samples are used as input to various rock physics models, and the predictions are compared with experimental data for the mixed samples. We have evaluated three strategies for choosing the initial properties in various rock physics models: (1) input values have the same porosity, (2) input values have the same pressure, and (3) an average of (1) and (2). The best results are obtained when the elastic moduli of the two porous constituents are defined at the same pressure and when their volumetric fractions are adjusted based on different compaction rates with pressure. Furthermore, our strategy makes the modeling results less sensitive to the actual rock physics model. The method can help obtain the elastic properties of mixed unconsolidated clays as a function of mechanical compaction. The more common procedure for estimating effective elastic properties requires knowledge about volume fractions, elastic properties of individual constituents, and geometric details of the composition. However, these data are often uncertain, e.g., large variations in the mineral elastic properties of clays have been reported in the literature, which makes our procedure a viable alternative.
Quantitative seismic interpretation has become an important and critical technology for improved hydrocarbon exploration and production. However, this is typically a resource-demanding process that requires information from several well logs, building a representative velocity model, and, of course, high-quality seismic data. Therefore, it is very challenging to perform in an exploration or appraisal phase with limited well control. Conventional seismic interpretation and qualitative analysis of amplitude variations with offset (AVO) are more common tools in these phases. Here, we demonstrate a method for predicting quantitative reservoir properties and facies using AVO data and a rock-physics model calibrated with well-log data. This is achieved using a probabilistic inversion method that combines stochastic inversion with Bayes' theorem. The method honors the nonuniqueness of the problem and calculates probabilities for the various solutions. To evaluate the performance of the method and the quality of the results, we compare them with similar reservoir property predictions obtained using the same method on seismic-inversion data. Even though both approaches use the same method, the input data have some fundamental differences, and some of the modeling assumptions are not the same. Considering these differences, the two approaches produce comparable predictions. This opens up the possibility to perform quantitative interpretation in earlier phases than what is common today, and it might provide the analyst with better control of the various assumptions that are introduced in the work process. IntroductionSeismic amplitude variations with offset (AVO) for reflections between two layers depend on the elastic properties and densities of both layers, which in turn are affected by hydrocarbon saturation and lithology. These amplitude variations can be modeled using the Zoeppritz equations (Mavko et al., 2009). AVO can be used as a direct hydrocarbon indicator by studying Erling Hugo Jensen 1 , Tor Arne Johansen 2, 3, 4 , Per Avseth 5, 6 , and Kenneth Bredesen 2, 7 the intercept R p crossplotted versus the gradient G. Typically, the data will exhibit a background trend of decreasing G with increasing R p , and a fluid factor can be defined as the perpendicular distance from this projection line to the data (Smith and Gidlow, 1987). Figure 1 shows the fluid factor for a vertical seismic section from the Norwegian Sea, slicing through and extending beyond a gas-sandstone discovery well (black dashed line). The fluid factor has been used to identify possible hydrocarbon prospects on the section. For example, we identify the hydrocarbon reservoir formations as well as some brightening right below the base Cretaceous unconformity (BCU) when moving off the structural high. Farther north, however, the graben anomalies have proven to be false (Avseth et al., 2016). Similarly, more detailed interpretation maps can be produced by highlighting various facies using the intercept versus gradient crossplot. Never...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.