خط مشی دسترسیدرباره ماپشتیبانی آنلاین
ثبت نامثبت نام
راهنماراهنما
فارسی
ورودورود
صفحه اصلیصفحه اصلی
جستجوی مدارک
تمام متن
منابع دیجیتالی
رکورد قبلیرکورد بعدی
Document Type:Latin Dissertation
Language of Document:English
Record Number:53535
Doc. No:TL23489
Call number:‭1533953‬
Main Entry:Munirudeen Ajadi Oloso
Title & Author:Prediction of crude oil PVT properties by Soft Computing techniquesMunirudeen Ajadi Oloso
College:King Fahd University of Petroleum and Minerals (Saudi Arabia)
Date:2009
Degree:M.S.
student score:2009
Page No:142
Abstract:Characterization of Pressure-Volume-Temperature (PVT) properties of crude oil is important for many types of petroleum calculations, such as, determination of hydrocarbon flowing properties, gas-lift and pipeline design, calculation of oil recovery both from natural depletion and recovery techniques. Two of these important properties are the oil viscosity and gas/oil ratio. An experimental analysis which is both time-consuming and costly is used to determine these properties over the entire range of pressures. To solve the problem of going through these rigorous laboratory experimentations which gulp valuable production resources, time and money, equations of states (EOS) and empirically derived correlations have been used to predict these reservoir fluid properties. These two methods were used for a long time until Soft Computing (SC) /Artificial Intelligence (AI) techniques, basically Neural Networks, were introduced to improve the prediction performances. However, all the prediction methods up to date are for predicting single or multi-data points, even for PVT properties that are generated as curves In this study, we have developed a new approach for predicting PVT properties that need to be described by curves over specific ranges of reservoir pressures. This approach is demonstrated with oil viscosity and gas/oil ratio curves. First, a thorough study of the target reservoir properties based on the data collected from PVT laboratory analyses of crude oil were carried out. Also, a statistical analysis was conducted on the data to detect the outliers. We then explored the capabilities of different Soft Computing techniques for predicting these properties. Different prediction models using Support Vector Regression (SVR), Functional Networks (FN), Adaptive Neuro-Fuzzy Inference Systems (ANFIS) and Artificial Neural Networks (ANN) and also two hybrid models: Differential Evolution Algorithm with ANN (DE+ANN) and Genetic Algorithm with ANFIS (GA+ANFIS) have been developed. A very small root mean square error and absolute average percent error for the developed models were recorded. Any PVT property which can be described as a curve can easily be estimated using the outlined approach in this work. Therefore, this work will hopefully be a very fast and low cost method for predicting PVT properties for optimizing the oil production operation.
Subject:Applied sciences; Petroleum engineering; Systems science; 0790:Systems science; 0765:Petroleum engineering
Added Entry:A. E. Khoukhi, Moustafa
Added Entry:King Fahd University of Petroleum and Minerals (Saudi Arabia)