Estimation of the Reservoir Permeability by Petrophysical Information Using Intelligent Systems

Abstract

Permeability is one of the most important parameters in order to evaluate a hydrocarbon reservoir. The permeability of a formation is usually determined from the cores and/or well tests. It should be noted that cores and well test data are often only available from few wells in a reservoir while the logs are available from the majority of the wells. Therefore, the evaluation of permeability from well log data represents a significant technical as well as economic advantage. Many fundamental problems remain unsolved by most predictive models. This article introduces the use of an improved neural network trained by a back propagation learning algorithm to provide solution for the permeability prediction from well log data. An Iranian offshore gas field which is located in the Persian Gulf, has been selected as the study area in this article. Well log data are available on a substantial number of wells. Core samples are also available from a few wells. It was shown that the neural network system is the most effective method in predicting permeability from well logs.     

The Application of an Artificial Neural Network (ANN) and a Genetic Programming Neural Network (GPNN) for the Modeling of Experimental Data of Slim Tube Permeability Reduction by Asphaltene Precipitation in Iranian Crude Oil Reservoirs

Abstract

The authors introduce a new implementation of the neural network (ANN), genetic programming neural network (GPNN), and neuro-fuzzy (NF) technology in petroleum engineering. An intelligent framework is developed for calculating the amount of permeability reduction by asphaltene precipitation in Iranian crude oil reservoirs over a wide pressure, temperature, and solvent mole fraction range. Theoretical results and practical experience indicate that a feed-forward network can approximate a wide class of function relationships very well. In this work, a conventional feed-forward multilayer ANN, GPNN, and NF approach have been proposed to predict the amount of permeability reduction. The accuracy of the method is evaluated by predicting the amount of permeability reduction of various reservoir fluids not used in the development of the models. One of the ways in modeling such systems is using intelligent techniques, which need information about the systems, so, based on some intelligent learning methods, it can provide a suitable model. Furthermore, the performance of the model is compared with the performance of a simple model for permeability reduction prediction, a new correlation, and experimental data. Results of this comparison show that the proposed GPNN method first and then NF method is superior both in accuracy and generality, over the other models.     

Predicting the Bubble-Point Pressure and Formation-Volume-Factor of Worldwide Crude Oil Systems

Abstract

This paper presents models for predicting the bubble-point pressure (P _b ) and oil formation-volume-factor at bubble-point (B _ob) for crude oil samples collected from several regions around the world. The regions include major producing oil fields in North and South America, North Sea, South East Asia, Middle East, and Africa. The model was developed using artificial neural networks with 5200 experimentally obtained PVT data sets. This represents the largest data set ever collected to be used in developing P _b and B _ob models. An additional 234 PVT data sets were used to investigate the effectiveness of the neural network models to predict outputs from inputs that were not used during the training process. The network model is able to predict the bubble-point pressure and the oil formation-volume-factor as a function of the solution gas–oil ratio, the gas relative density, the oil specific gravity, and the reservoir temperature. In order to obtain a generalized accurate model, back propagation with momentum for error minimization was used. The accuracy of the models developed in this study was compared in details with several published correlations. This study shows that if artificial neural networks are successfully trained, they can be excellent reliable predictive tools to estimate crude oil properties better than available correlations. The network models can be easily incorporated into any reservoir simulators and/or production optimization software.     

Petroleum reservoir characterization with the aid of artificial neural networks

We introduce a new application of artificial neural network technology in the characterization of reservoir heterogeneity. Different reservoir properties, such as porosity, permeability and fluid saturation, in highly heterogeneous formations can be predicted with good accuracy using information deduced from readily available geophysical well logs. The methodology by which this is carried out is based on the intelligent and adaptive pattern recognition capabilities of an artificial neural network (three-layer feed forward, back propagation). The need for expensive processes to acquire porosity, permeability and fluid saturation data (such as well testing and extensive coring of the formation) may therefore be greatly reduced. Examples of several neural networks developed during this study are presented.     

新型电缆地层测试器渗透率反演方法及软件研究

结合新型电缆地层测试器FCT的特点,将整个预测试过程简化为3个线性系统来反演地层渗透率.在算法中考虑了表皮效应和管线存储效应校正;编制了线性系统下的常规分析程序和非线性系统下的神经网络方法程序.并针对BP算法收敛速度慢的特点,引入Levenberg-Marquardt算法对BP模型进行改进,用实际压力数据进行测试,与Geoframe解释结果相比,吻合良好.通过实际验证,所建立的神经网络模型,预测精度良好,具有较好的非线性映射能力.     

基于不同人工神经网络的催化裂化焦炭产率预测对比

在催化裂化装置（FCC）中,焦炭产率增加不但会使装置的总液收降低,而且会影响装置的热平衡,增加装置的操作难度。控制催化裂化装置焦炭产率十分重要,而其前提是能够准确预测装置的焦炭产率。催化裂化焦炭的生成和烧焦过程是一个连续的过程,影响参数众多且各参数之前互相影响,使用传统的方法建立多参数的预测模型具有一定的难度。本文利用人工神经网络(ANN)结合催化裂化装置的生产数据分别建立了GRNN神经网络预测模型和BP神经网络预测模型。对比分析结果表明,BP神经网络预测结果的准确度和稳定性优于GRNN神经网络。为进一步提高BP神经网络的预测效果,又分别使用了粒子群算法(PSO)和遗传算法(GA)对其进行优化。对比分析两种优化算法表明,两种优化算法均能提高BP神经网络的预测精度,综合考虑预测结果的准确性和稳定性两个方面,经遗传算法优化的BP神经网络预测模型优于经粒子群算法优化的BP神经网络预测模型。     

The Use of ANN and the Mathematical Model for Prediction of the Main Product Yields in the Thermal Cracking of Naphtha

Abstract

Thermal cracking of naphtha has such numerous reaction routes that the detailed reaction mechanism has not yet been determined. In this regard, a model of artificial neural networks (ANNs), using back propagation (BP), is developed for modeling thermal cracking of naphtha. The optimum structure of the neural network was determined by a trial-and-error method. Different structures were tried with several neurons in the hidden layer. The model investigates the influence of the coil outlet temperature, the pressure of the reactor, the steam ratio (H2O/naphtha), and the residence time on the pyrolysis product yields. A good agreement was found between model results and experimental data. A comparison between the results of the mathematical model and the designed ANN was also conducted and the relative absolute error was calculated. Performance of the ANN model was better than the mathematical model.     

PREDICTING THE EFFECT OF FEEDSTOCK ON PRODUCT YIELDS AND PROPERTIES OF THE FCC PROCESS

ABSTRACT

The mechanism of petroleum refining processes are too complex, and no thorough model has yet been developed. Neural networks represent an effective alternative to mathematical modeling of refinery operations if a sufficient amount of input-output data is available. In this paper, a feed forward neural network that models the Fluid Catalytic Cracking (FCC) process will be presented. The FCC process is the workhorse of the petroleum refining industry, making small and medium sized molecules out of big ones (gasoline and distillate out of gas oils). The input-output data to the neural network was collected from the literature on pilot and commercial plant operations and were obtained from actual refineries. Several network architectures were tried and the network that best simulates the FCC process was retained. This network is able to predict yields of products of the FCC unit as well as their properties. The network consists of one hidden layer of twenty neurons, an input layer of four neurons, and an output layer of twelve neurons. The predictions of the neural network model were compared to those of a commercial simulator of the FCC process, to non-linear regression models, and to published charts. The results show that the neural network model consistently gives better predictions.     

Prediction of Coke Yield of FCC Unit Using Different Artificial Neural Network Models

In fluid catalytic cracking(FCC) unit, it is greatly important to control the coke yield, since the increase of coke yield not only leads to the reduction of total liquid yield, but also affects the heat balance and operation of FCC unit. Consequently, it is significant to predict the coke yield accurately. The coke formation and burning reactions are affected by many parameters which influence each other, so it is difficult to establish a prediction model using traditional models. This paper combines the industrial production data and establishes a generalized regression neural network(GRNN) model and a back propagation(BP) neural network model to predict the coke yield respectively. The comparison and analysis results show that the accuracy and stability of the BP neural network prediction results are better than that of the GRNN. Then, the particle swarm optimization to optimize BP neural network(PSO-BP) and genetic algorithm to optimize the BP neural network(GA-BP) were further used to improve the prediction precision. The comparison of these models shows that they can improve the prediction precision. However, considering the accuracy and stability of the prediction results, the GA-BP model is better than PSO-BP model.     

Predicting the Bubble-Point Pressure and Formation-Volume-Factor of Worldwide Crude Oil Systems

This paper presents models for predicting the bubble-point pressure (P_b) and oil formation-volume-factor at bubble-point (B_ob) for crude oil samples collected from several regions around the world. The regions include major producing oil fields in North and South America, North Sea, South East Asia, Middle East, and Africa. The model was developed using artificial neural networks with 5200 experimentally obtained PVT data sets. This represents the largest data set ever collected to be used in developing P_b and B_ob models. An additional 234 PVT data sets were used to investigate the effectiveness of the neural network models to predict outputs from inputs that were not used during the training process. The network model is able to predict the bubble-point pressure and the oil formation-volume-factor as a function of the solution gas-oil ratio, the gas relative density, the oil specific gravity, and the reservoir temperature. In order to obtain a generalized accurate model, back propagation with momentum for error minimization was used. The accuracy of the models developed in this study was compared in details with several published correlations. This study shows that if artificial neural networks are successfully trained, they can be excellent reliable predictive tools to estimate crude oil properties better than available correlations. The network models can be easily incorporated into any reservoir simulators and/or production optimization software.     

人工神经网络技术在油田储层物性预测中的应用――以西峰油田为例

与传统的测井资料解释和信息处理技术相比较,在对非均质性较强、物性参数级差较大的储集层物性预测中,人工神经网络技术具有极强的自适应和自学习能力,其通过很强的非线性映射,能够精确地建立储集层参数与测井响应之间的非线性模型。在论述神经网络技术基本原理的基础上,对西峰油田延安组和延长组储层的物性参数(孔隙度和渗透率等)进行了预测,取得了较理想的结果。预测结果表明:渗透率参数级差不大(<102)时,预测精度高;渗透率的变化范围较大(>103)时,对具有高渗透率储层的预测精度高,而对具有低渗透率储层的预测值相对误差较大。     

The Prediction of Wax Precipitation by Neural Network and Genetic Algorithm and Comparison With a Multisolid Model in Crude Oil Systems

Abstract

This article introduce a new implementation of the neural network and genetic programming neural network technology in petroleum engineering. An intelligent framework is developed for calculating the amount of wax precipitation in petroleum mixtures over a wide temperature range. Theoretical results and practical experience indicate that feedforward networks can approximate a wide class of function relationships very well. In this work, a conventional feedforward multilayer neural network and genetic programming neural network (GPNN) approach have been proposed to predict the amount of wax precipitation. The introduced model can predict wax precipitation through neural network and genetic algorithmic techniques. The accuracy of the method is evaluated by predicting the amount of wax precipitation of various reservoir fluids not used in the development of the models. Furthermore, the performance of the model is compared with the performance of multisolid model for wax precipitation prediction and experimental data. Results of this comparison show that the proposed method is superior, both in accuracy and generality, over the other models.     

改进的开窗技术在利用测井资料预测渗透率中的应用

渗透率是油藏描述和油藏工程中一关键性的参数.文中描述了一种计算渗透率的新方法,即在岩心分析化验数据和相关测井曲线数据归一化的基础上,利用改进的开窗技术,借助反馈的神经网络方法逐点计算地层的渗透率.通过在胜坨油田的实际应用,证明该方法预测的渗透率与实际渗透率符合较好,具有推广应用的价值.     

人工神经网络方法预测润滑油基础油的抗氧性

用反向传播神经网络方法研究了润滑油放置氧弹时间与基础油化学组成之间的关系。研究表明，网络方法比常用的线性或指数关联法更能准确地关联和预报润滑油基础油的抗氧性。本文还对输入参数的选择及过拟合现象的预防方法进行了研究。     

基于人工神经网络的油水分离旋流器设计模型

以人工神经网络为手段，建立了油水分离旋流器设计模型。采用三层BP网络模型，成功地实现了根据处理物料物性参数和分离要求进行油水分离旋流器结构与操作参数全面设计的过程。通过设定足够大的神经网络训练次数，神经网络预测误差可逼近所需精度，完全满足油水分离旋流器参数设计要求。     

The Use of Optimization Procedures to Estimate Minimum Miscibility Pressure

An important factor in the design of gas injection projects is the minimum miscibility pressure (MMP). A new genetic algorithm (GA)–based correlation and two neural network models (one of them is trained by back propagation [BP] algorithm and another is trained by particle swarm optimization algorithm) have been developed to estimate the CO₂–oil MMP. The correlation and models use the following key input parameters: reservoir temperature, molecular weight of C⁺ ₅, and mole percentage of the volatiles and intermediate components (for the first time, the mole percentages are used as independent variables). Then results are validated against experimental data and finally compared with commonly used correlations reported in the literature. The results show that the neural network model trained by BP algorithm and the correlation that has been developed by GA can be applied effectively and afford high accuracy and dependability for MMP forecasting.     

利用反射地震资料和多尺度训练集的深度学习速度建模

随着地震勘探数据量的逐渐增大,常规地震速度建模方法在稳定性、精度和计算效率等方面均面临挑战。为此,提出一种利用反射地震资料和多尺度训练集的深度学习速度建模的方法,即将反射波形数据和速度谱联合作为全卷积神经网络的输入,并在网络中引入Dropout层提高泛化能力,结合多尺度训练集,实现从地震数据到速度模型的映射。为了测试该方法在不同地质构造条件下的效果和适用性,分别应用层状模型、孤立异常体模型和BP盐丘模型进行数值实验。实验结果表明,联合使用地震反射波形和速度谱作为深度学习特征数据集时,速度建模准确性优于仅采用地震反射波形或速度谱作为特征数据集的结果,并克服了单独使用反射波形导致建模不稳定和单独使用速度谱建模精度不足的缺陷;使用多尺度速度模型构建训练集的速度建模结果在异常体边界的准确性优于采用单尺度模型训练集;深度神经网络只需经过一次训练,就可以快速地对与训练集中速度结构相似的地下构造进行速度建模,比常规方法具有更高的计算效率。在构建大量速度模型时,该方法具有很好的推广价值。     

一种基于BP神经网络的完井液污染类型识别方法

目的 解决油田目的层钻井过程中完井液受盐水、残酸等污染后不能高效识别污染类型的问题。方法 对完井液进行不同质量占比的盐水、残酸污染测定,采用K-means聚类订正不同污染等级数据样本的标签。根据数据样本特征的获取难易度、隐藏层数目,训练不同的BP神经网络模型,并由留一交叉验证法检验模型的分类准确率。结果 数据样本拥有的特征越多,训练的BP神经网络分类准确率越高,隐层数目越多,分类准确率反而越低。选择包含“流变+老化+滤失+井名”4类特征的数据样本建立1隐藏层的BP神经网络模型,其平均分类准确率达到93.18%。结论 由流变、滤失等特征训练的BP神经网络模型可快速应用于试油现场,解决完井液污染类型识别问题,避免了试油现场因缺少大型仪器而无法鉴别完井液污染类型的难题。