论文部分内容阅读
对于油藏动态计算和分析而言,压力、体积、温度(PVT)特性的重要性在于能精确计算如泡点压力、溶解气油比和原油地层体积系数等油藏参数。收集和分析多年积累的PVT资料来确定不同类型油气系统。几乎所有的这些相关分析都是用线性的、非线性多重回归技术或图象技术完成的。一旦人工神经网络训练成功,就为可靠的获得原油PVT特性结果提供一种新的可供选择的方法。在本研究中,我们提出为预测中东原油的PVT特性的神经网络模型。通过训练的网络数据代表了以前在建立中东原油PVT模型所收集的最大数据集。该神经网络模型可用来预测作为溶解气油比、天然气比重、原油比重和温度的一个函数的泡点压力和油层体积系数。本文用神经网络模型和其他相关分析方法,对中东原油样品进行预测的结果,作了详细的比较。
The importance of pressure, volume, and temperature (PVT) properties for reservoir dynamic calculations and analyzes is the ability to accurately calculate reservoir parameters such as bubble pressure, dissolved gas-oil ratio, and crude oil formation volumetric coefficient. Collect and analyze the accumulated PVT data over the years to identify different types of hydrocarbon systems. Almost all of these correlation analyzes are done using linear, non-linear multiple regression techniques or imaging techniques. Once the artificial neural network training is successful, it provides a new alternative method for reliably obtaining the results of PVT properties of crude oil. In this study, we propose a neural network model for predicting PVT characteristics of Middle East crude oil. The trained web data represents the largest dataset previously collected in the Middle East crude PVT model. The neural network model can be used to predict the bubble point pressure and reservoir volume factor as a function of dissolved gas-oil ratio, natural gas specific gravity, crude oil specific gravity and temperature. In this paper, neural network model and other related analysis methods, the Middle East crude oil samples to predict the results made a detailed comparison.