Permeability Estimation From Well Log Responses

Abstract

Permeability is one of the most important characteristics of hydrocarbon bearing formations. Formation permeability is often measured in the laboratory from reservoir core samples or evaluated from well test data. However, core analysis and well test data are usually only available from a few wells in a field. On the other hand, almost all wells are logged. This paper presents a non-parametric model to predict reservoir permeability from conventional well log data using an artificial neural network (ANN). The ANN technique is demonstrated by applying it to one of Saudi Arabia's oil fields. The field is the largest offshore oil field in the world and was deposited in a fluvial dominated deltaic environment. The use of conventional regression methods to predict permeability in this case was not successful. The ANN permeability prediction model was developed using some of the core permeability and well log data from three early development wells. The ANN model was built and trained from the well log data and their corresponding core measurements by using a back propagation neural network (BPNN). The resulting model was blind tested using data which was taken from the modelling process. The results of this study show that the ANN model permeability predictions are consistent with actual core data. It could be concluded that the ANN model is a powerful tool for permeability prediction from well log data. Introduction Many oil reservoirs have heterogeneity in rock properties. Understanding the form and spatial distribution of these heterogeneities is fundamental to the successful exploitation of these reservoirs. Permeability is one of the fundamental rock properties, which reflects the rock's ability to transmit fluids when subjected to pressure gradients. While this property is very important in reservoir engineering, there is no specific geophysical well log for permeability, and its determination from conventional log analysis is often unsatisfactory(1). In general, porosity and permeability are independent properties of a reservoir. However, permeability is low if porosity is disconnected, whereas permeability is high when porosity is interconnected and effective. Despite this observation, theoretical relationships between permeability and porosity have been formulated, such as the Kozeny-Carmen theory. The Kozeny-Carmen theory relates permeability to porosity and specific surface area of a porous rock which is treated as an idealized bundle of capillary tubes. This theory, however, ignores the influence of conical flow in the constrictions and expansions of the flow channels and treats the highly complex porous medium in a very simple manner. Empirical relationships based on the Kozeny-Carmen theory have also been developed that relate permeability to other logs and/or log-derived parameters such as resistivity and irreducible water saturation(2). These relationships are applied only to the region above the transition zone or to the transition zone itself. Since core permeability data are available for most exploration and development wells, statistical methods have become a more versatile alternative in solving the problem of determining reservoir permeability. Regression is widely used as a statistical method to search for relationships between core permeability and well log parameters(3, 4).

Impacto de los Métodos Aplicados para la Determinación de la Permeabilidad y su Influencia en el Gerenciamiento del Reservorio en el Campo Shushufindi

“…Los registros se han convertido en una herramienta fundamental para la evaluación de formaciones, son corridos una vez culminada la etapa de perforación de un pozo. Los registros se basan en la medición de propiedades físicas, con una amplia variedad de herramientas, no existe un registro geofísico específico para la determinación directa de la permeabilidad, pero es posible su determinación mediante el análisis de registros convencionales (Shokir et al, 2006).…”

Predicting hydrofacies and hydraulic conductivity from direct-push data using a data-driven relevance vector machine approach: Motivations, algorithms, and application

“…A fundamentally different approach to model H‐G relations involves the application of statistical techniques, either simple parametric relations [e.g., Hyndman et al ., ] or more complex nonparametric methods [e.g., Mohaghegh et al ., ; Wong et al ., ; Lee and Datta‐Gupta , 1999; Chen et al ., ; Chen and Rubin , ; Paasche et al ., ; Shokir et al ., ; Dubois et al ., ; Al‐Anazi et al ., ; Elshafei and Hamada , ; Kharrat et al ., ; Al‐Anazi and Gates , 2010a, b; Dubreuil‐Boisclair et al ., ; Ruggeri et al ., ; Rumpf and Tronicke , ]. Unlike general theoretical or semiempirical models, statistical techniques are much more flexible and do not require prior knowledge about physical relations between various H‐G parameters or geological material.…”

“…Artificial neural networks (ANNs), which follow an empirical risk minimization of the training errors, are common form of learning machines that have been already considered. Different architectures of ANNs have been applied successfully for the prediction of lithofacies [ Chen and Rubin , ; Dubois et al ., ] or hydraulic properties in petroleum reservoirs [ Mohaghegh et al ., ; Wong et al ., ; Lee and Datta‐Gupta , ; Shokir et al ., ; Al‐Anazi et al ., ; Elshafei and Hamada , ; Kharrat et al ., ; Iturrarán‐Viveros and Parra , ] from cross hole or borehole geophysics data. However, despite their potential effectiveness, ANNs present some important drawbacks [ Camps‐Valls et al ., ]: (i) design and training often results in a complex, time‐consuming task, in which many parameters must be tuned; (ii) minimization of the training errors can lead to poor generalization performance; and (iii) performance can be degraded when working with small (sparse) data sets.…”

ANN-based prediction of cementation factor in carbonate reservoir

“…Many researchers had been predicted permeability with artificial neural networks from well logs data [3,4], generation of synthetic wireline logs from other logs [5,6], identification of lithological and depositional facies via competitive neural network and fuzzy logic [7], as well as estimation of reservoir permeability using integration of genetic algorithm and a coactive neuro-fuzzy inference system [8]. Archie in 1942 found the correlation between porosity (PHI), and resistivity formation factor (F).…”