Comment quantifier l’hétérogénéité des faciès sédimentaires ?

The best approach is to start by understanding the depositional environment and extracting a depositional facies log to reveal the geometric aspects of flow units. Next, extract a petro-facies log based on the petrophysical characteristics of the identified facies. Integrating depositional and petrophysical facies with seismic attributes helps establish a robust strategy for understanding facies variability. After integration, analyze the final facies log data to define the appropriate geostatistical approach for modeling facies distribution. This systematic approach guarantees a more precise and comprehensive distribution and mapping of facies.

Carbonate heterogeneity can be understood by integrating both static and dynamic data. Static data such as: - depositional environment - diagenesis processes - log and core data specially mini permeameter . - image log data - seismic data Dynamic data - PLT - Monitoring saturation behind casing TDT - 4D Seismic No one data set or tool can answer the question...

to quantify sedimentary facies heterogeneity is to use reservoir heterogeneity. Reservoir heterogeneity can be evaluated using various techniques such as geostatistics, seismic attributes, and well logs . Geostatistics is a statistical method used to analyze spatial data and to predict the distribution of a variable over space. Seismic attributes are special measures of kinematic, dynamic, geometric, or statistical characteristics obtained from seismic data and which contribute to structural, stratigraphic, and petrophysical interpretation .Well logs.are records of the. physical properties of rocks and fluids encountered in a borehole. These logs can be used to estimate the thickness, porosity, permeability, and lithology.of the reservoir.

The choice of method for quantifying sedimentary facies heterogeneity depends on the specific goals of your study, the type of data available, and the scale of heterogeneity you aim to capture. Various techniques are employed, including geostatistics, seismic attributes, and well logs: • Geostatistics, a statistical method for spatial data analysis, helps predict the distribution of reservoir properties. • Seismic attributes provide insights into structural, stratigraphic, and petrophysical features, while well logs directly measure properties like thickness, porosity, and permeability. • The integration of these methods offers a comprehensive approach to evaluating reservoir heterogeneity.

Normally it’s very important to identify and know the heterogeneity dimensions which can be vary from centimetres to kilometres, several methods can be used to help in understand the limitations of lithologhy facies, numerous of factors can be effected on sedimentary facies like chemical participation or weathering processing, knowing heterogeneity dimensions by facies proportion, facies diversity, facies association, facies connectivity and facies entropy are highly required to improve and enhance the reservoir properties and output oil and gas

C’est très bien développé et détaillé, mais à ajouter des zones de contact, type de ciment et les zones d’altérations ou de sommation .

Often geomodelers heavily rely on variogram analysis, investing considerable time in finding the optimal variogram model to fit data. While addressing trends by segmenting data into sub-zones or rock types proves beneficial in certain cases, it's important to acknowledge its limitations, particularly when dealing with datasets exhibiting non-stationary distribution. In such scenarios, persisting with variogram analysis might be counterproductive. A more effective approach involves conducting homoscedasticity and stationary analyses, incorporating necessary adjustments to discern the stationarity of the initial data. This approach is prevalent in the mining industry but appears overlooked in the petroleum domain for some reason.

It's crucial to emphasize that the effectiveness of using a calculated variogram to manage the distribution of a specific property is constrained by the spatial layout of samples, such as wells. Variogram ranges computed from sparse data points that are widely separated should be assigned less weight. A more robust approach involves using trend maps extracted from seismic attributes, porosity and thickness trends, or information from nearby analogs. This proves to be a more powerful tool for deducing reasonable ranges in the distribution of a given property.

In my opinion, quantifying facies heterogeneity using seismic data would be the best way in reservoir modelling, if facies logs were determined in well scales and a good correlation could be find between seismic attributes and facies codes.

by leveraging machine learning on geological data, you can not only learn patterns but also evaluate and quantify the heterogeneity of sedimentary facies using a variety of performance metrics. This process aids in gaining insights into the geological composition and spatial distribution of different facies within a given area.

Despite the numerous benefits that ML brings to facies identification and quantifying heterogeneity—such as managing correlations, sensitivities, noise cleanup, and more—there's a crucial need for effective product management. This involves rigorous testing of various algorithms and models, as well as exploration of additional datasets. ML's pivotal advantage lies in its ability to streamline and expedite these processes, especially when handling vast amounts of input data, ranging from wellbores and well logs to core samples and seismic attributes. The manual picking of facies types across hundreds of wells is a time-consuming task, making the initial draft provided by ML immensely valuable in accelerating the overall workflow.