Estimation of Rock Dynamic Elastic Property Profiles through a Combination of Soft Computing, Acoustic Velocity Modeling, and Laboratory Dynamic Test on Core Samples

Abstract

Abstract This paper presents a new approach for estimating rock elastic properties in wells, especially for wells with limited log suites. The approach is basically a combination of efforts that are put is a series of steps. Firstly, is to model a synthetic S-wave velocity profile for a key well, with support from laboratory acoustic measurement on core samples, enabling the establishment of profiles of elastic properties through the theory of elasticity. Secondly, the modeling and validating of relationship between P-wave velocity, Poisson ratio (as an example in this paper), porosity, water saturation, shale contents, and matrix density based on the log data available for the key well. Thirdly, prediction of all missing log suites (if any) for other wells using soft computing (artificial intelligence/neural network) that has been ‘trained’ using data from the key well and other wells that have more or less complete log data. Fourthly, estimation of rock elastic properties for all wells (except the key well) using soft computing. Finally, evaluation of results using comparison with the model validated in the key well. The method has been applied on 14 wells of an active oil reservoir in Java, Indonesia. Comparisons of porosity and water saturation values between results from standard log interpretation and results from the validated model serve as indicators for the success of the method. The reasonably good comparisons achieved have proved that the new approach is applicable, and the use of the model relationship avoids ‘blind’ estimation often practiced in reservoir characterization. Introduction Various operations in upstream petroleum industry require rock elastic properties, such as Poisson ratio, Young's modulus and bulk modulus, for data input in their designs. Hydraulic fracturing, anticipation for sand production, and wellbore stability are among the activities in the industry that require the data. Recent developments in seismic technology for reservoir characterization have also suggested the importance of rock elastic properties as properties that are potentially related to rock physical properties such as porosity and fluid saturation. In most cases, these required data are simply unavailable for the purpose. It is a matter of fact that not all petroleum wells, as the most immediate sources for the data, have even the standard log survey (i.e. logs normally dedicated to estimating rock physical properties such as porosity and water saturation) run on them. This is not to mention the scarcity of survey that can produce the rock elastic properties. For a particular reservoir, it is rare that either dynamic survey (e.g. full waveform log) or static survey (e.g. compression test on core samples) is conducted. Considering the data scarcity, a method that uses all available data is therefore desired. Past studies have offered many methods and abundant investigation data that range from well survey to laboratory measurement on samples, from fully empirical to models, and from analysis on single source to combination of approaches. For instance Charlez et al1 proposed an inversion method to estimate the properties from fracmeter survey in wells, Harrison et al2 presented a method to extract shear wave velocity from a combination of monopole and dipole sonic logs from which elastic/mechanical properties can be estimated, whilst direct investigations on rock samples are best presented by extensive data.3,4 Empirical relationships have also been taken as an approach5,4 as well as a combination with application of acoustic velocity models.6 As commonly accepted, none of empirical relations appears to have no general validity.