The Design of New Soft Sensors Based on PCA and a Neural Network for Parameters Estimation of a Petroleum Reservoir

Abstract

A new set of soft sensors is presented based on principal component analysis (PCA) and artificial neural network (ANN) methodologies for parameters estimation of a petroleum reservoir. The crude diagrams of reservoir parameters provide valuable evaluation for petrophysical parameters. These parameters, however, are usually difficult to measure due to limitations on cost reliability considerations, inappropriate instrument maintenance, and sensor failures. PCA is utilized to develop new soft sensors to incorporate reliability and prediction capabilities of ANN. For this purpose, a PCA model is derived to reconstruct a parameter from other reservoir parameters using their redundancy relations. The developed soft sensors are applied to reconstruct parameters of Marun reservoir located in Ahwaz, Iran, by utilizing the available geophysical well log data. The experimental results demonstrate that the proposed hybrid PCA-NN algorithm is able to reveal a better performance than the PCA and the conventional back propagation–based NNs.     

Improved Permeability Prediction of a Heterogeneous Carbonate Reservoir Using Artificial Neural Networks Based on the Flow Zone Index Approach

Abstract

Permeability is one of the most important parameters required for reservoir characterization. Although core analysis provides more exact information, core data do not exist for all wells in the reservoir because coring is expensive and time consuming. Therefore, another approach should be sought for permeability determination. The objective of this study was to create an artificial neural network (ANN) model in order to use well log data to predict permeability in uncored wells/intervals. The well log, core, and other data were gathered from an Iranian heterogeneous carbonate reservoir. A flow zone indicator was then predicted using an ANN approach with well logs as input variables. The reservoir was thus classified into different zones based on hydraulic flow units to overcome the extreme heterogeneity. Then, a separate ANN training procedure was followed for each flow zone with log data as input variables and permeability as output. This improved method is capable of permeability prediction in heterogeneous carbonate reservoirs in uncored wells/intervals with an average error of less than 10.9%.     

A new methodology for identification of potential pay zones from well logs: Intelligent system establishment and application in the Eastern Junggar Basin, China

In recent years, as the exploration practices extend into more complicated formations, conventional well log interpretation has often shown its inaccuracy and limitations in identifying hydrocarbons. The Permian Wutonggou Formation hosts typical clastic reservoirs in the Eastern Junggar Basin. The sophisticated lithology characteristics cause complex pore structures and fluid properties. These all finally cause low well testing agreement rate using conventional methods. Eleven years' recent statistics show that 12 out of 15 water layers have been incorrectly identified as being oil or oil/water layers by conventional well log interpretation. This paper proposes a methodology called intelligent prediction and identification system(IPIS). Firstly, parameters reflecting lithological, petrophysical and electrical responses which are greatly related to reservoir fluids have been selected carefully. They are shale content(Vsh), numbered rock type(RN), porosity(Φ), permeability(K), true resistivity(RT) and spontaneous-potential(SP). Secondly, Vsh, Φ and K are predicted from well logs through artificial neural networks(ANNs). Finally, all the six parameters are input into a neuro-fuzzy inference machine(NFIM) to get fluids identification results. Eighteen new layers of 145.3 m effective thickness were examined by IPIS. There is full agreement with well testing results. This shows the system's accuracy and effectiveness.     

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.     

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.     

砂砾岩储层孔隙度和渗透率预测方法

垦西地区沙四段砂砾岩体岩块含量高,矿物多变,孔隙结构复杂,导致测井响应差异大,用传统的统计回归方法和测井解释方法计算储层孔隙度和渗透率精度很低.用基于测井相分析的参数回归法、多矿物模型测井最优化方法和BP神经网络法等方法分别计算砂砾岩孔隙度和渗透率,其目的是提高复杂岩性储层参数的解释精度,满足油藏描述和储量计算对参数精度的要求.通过与岩心分析结果对比,基于测井相分析的参数回归法较一般解释方法的解释精度高,而多矿物模型测井最优化法和BP神经网络法等非参数数学建模方法比前者的效果更好.该方法提高了现有测井信息的利用率,在垦西地区砂砾岩体应用中取得了良好的效果.     

神经网络技术用于测井解释的评述

对目前测井解释中广泛使用的神经网络技术提出一些粗浅的认识或评价,包括这项技术的目前研究现状,研究趋势和提高方法实用效果的可能途径等,以期今后能提高网络模型的预测能力,从而充分发挥网络技术在测井解释中的应用潜力。     

Neural Network Identification of Hydrocarbon Potential of Shaly Sand Reservoirs

Abstract

Hydrocarbon potential evaluation of shaly sand layers requires the adoption of certain shaly water model and also the selection of suitable conventional logging suite. The main target is how to get the accurate porosity and how to convert the apparent water saturation to true water saturation for given shaly sand layer. In this paper neural network approach is presented to replace the conventional interpretation of well logging data to better determine formation porosity and water saturation and to well identify hydrocarbon potential of clean and shaly sand layers. Two neural networks were constructed, one for prediction of porosity using six well logging data inputs: GR, LLD, RHOB, NPHI, PEF, and Δt); and water saturation using five well logging data inputs: GR, LLD, RHOB, NPHI and PEF. Shale volume was defined from GR data. Each neural network is trained using available logging data and validated using the core data before applying it to the entire well log. Neural network predicted formation porosity and water saturation for tested sections of one well. Also using the cut off values of porosity, water saturation and shale volume, the neural network defined the possible pay zones in the well. Network outputs have shown good matching with core data and the reference calculated petrophysical parameters. The developed network approach has successfully deduced porosity, water saturation and defined pay zones in a new well that projects its application for new wells.     

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.     

Introducing a New Method for Predicting PVT Properties of Iranian Crude Oils by Applying Artificial Neural Networks

Abstract

In this study artificial neural networks (ANNs) have been applied for the prediction of main pressure, volume, and temperature (PVT) properties, bubble point pressure (P_b), and bubble point oil formation volume factor (B_ob) of crude oil samples from different wells of Iranian oil reservoirs. Via a detailed comparison, the great power of ANNs with respect to traditional methods of predicting PVT properties, like Standing, Vasquez and Beggs, and Al-Marhoun, with higher prediction precision up to R² = 0.990 has been illustrated and the obtained parameters of ANNs for the application of prediction of other crude oil samples has been presented. The applied PVT data set in this study consists of 218 crude oil samples from Iranian reservoirs and for assurance of the applicability of the ANN model the PVT data set has been divided into 2 training (190 samples) and cross validation (28 samples) data sets and obtained ANNs from applying the training data set has been tested on the cross validation data set which has not been seen by the network during the training process. The obtained results for both training and cross validation data sets confirm the great prediction power of ANNs, for both data sets with respect to traditional PVT correlations.     

神经模糊混合系统在测井储层识别中的应用

储层识别是石油勘探中十分重要的基础工作,它可以为油田勘探和开发提供可靠的依据。神经模糊混合方法是通过对测井数据的学习,运用模糊逻辑与神经网络相结合的混合系统对测井数据进行提取和优化。根据来自不同油井的观测数据,采用一个二阶段的策略来决定该预测模型的结构和参数,从而对测井储层进行识别。给出了该混合方法预测的初步结果,为油井的开发提供了重要的参考。     

低阻油气层识别方法研究

在油气勘探开发中,地球物理测井资料解释的最基本任务之一是在钻孔剖面上准确地识别油(气)层、油水同层、水层、干层等。 新疆塔北地区三叠系的特殊油气层电阻率低于或接近水层电阻率，在电性上直接区分特殊油气层与水层很不现实;另外，该区油气层的特征和油气层的影响因素明显与其它油田存在差异，故解决问题方法和思路不同于其它油田。文章利用油气层、油水同层、水层和干层的测井曲线和储层参数，建立识别储层流体属性的判别模型，采用灰关联分析聚类法、BP人工神经网络等模式识别法，对实际测井资料进行了解释，识别结果与试油结果对比表明BP人工神经网络、奇异值分解等方法识别结果与实际结果基本一致，没有漏掉油气层，取得了好的解释效果。     

基于改进人工神经网络的气水层识别技术

在改进的神经网络训练算法的基础上．提出了利用神经网络快速识别气、水层的方法。为了迅速、准确地判断储层性质,选用了Kohonen自组织网络和BP神经网络,利用测井参数,建立了长庆气田气、水层识别模型。仿真计算与测井综合解释相对比,样本符合率高达81．3%。分析表明,该方法所需参数少、适用范围广,能定量识别出气水层,从而为制定有水气井改造措施提供较可靠的依据。     

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.     

The Estimation of Formation Permeability in a Carbonate Reservoir Using an Artificial Neural Network

Abstract

Reservoir permeability is an important parameter that its reliable prediction is necessary for reservoir performance assessment and management. Although many empirical formulas are derived regarding permeability and porosity in sandstone reservoirs, these correlations cannot be accurately depicted in carbonate reservoir for the wells that are not cored and for which there are no welltest data. Therefore, having a framework for estimation of these parameters in reservoirs with neither coring samples nor welltest data is crucial. Rock properties are characterized by using different well logs. However, there is no specific petrophysical log for estimating rock permeability; thus, new methods need to be developed to predict permeability from well logs. One of the most powerful tools that we applied by the authors is artificial neural network (ANN), whose advantages and disadvantages have been discussed by several authors. In particular, 767 data sets were used from five wells of Bangestan reservoir in a southwestern field of Iran. Depth, Neutron (NPHI), Density (RHOB), Sonic (DT) logs, and evaluated total porosity (PHIT) from log data were used as the input data and horizontal permeability obtained by coring was as target data. Sixty percent of these data points were used for training and the remaining for predicting the permeability (i.e., validation and testing). An appropriate ANN was developed and a correlation coefficient (R) of 0.965 was obtained by comparing permeability predictions and the actual measurements. As a result, the neural science can be used effectively to estimate formation permeability from well log data.     

人工神经网络：测井解释新的希望—神经网络系列研究报告之一

人工神经网络是一门新兴的信息处理技术。它在信息科学和工程技术领域已得到广泛的应用。它在测井解释领域的应用也取得了初步的成功,给测井解释领域诸多难题的解决带来了无限的希望。阐明了人工神经网络的基本原理。在分析介绍了人工神经网络的基本模型及其应用前景后,阐述了神经网络的具体实现办法,最后以一系列的应用实例说明了人工神经网络在测井解释领域的实际应用。     

A prediction method of borehole stability based on seismic attribute technology

Borehole instability in drilling engineering can bring about serious problems of drilling quality and safety. Based on the close relationships between seismic and well log information, the prediction method of borehole stability is presented to effectively control borehole instability. Conventional and nonlinear seismic attributes are extracted from borehole-side seismic traces of impending drilling well and drilled offset well respectively. Then the optimal attributes combinations sensitive to log properties are selected by using genetic algorithm and wavelet neural network technology together. A series of mapping models which reflect the nonlinear relationships between seismic attributes and acoustic and density log data of various formation intervals in drilled well are constructed through neural network modeling. With analysis of cutting logging data, seismic attributes of the formation under bit and corresponding mapping model can be used to predict acoustic and density log curves of this formation. Based on the predicted log data, log interpretation method, analysis technology of in-situ stress and mechanics model of borehole stability are employed to calculate in-situ stress, pore pressure, collapse pressure and fracture pressure, thus the safe drilling fluid density window which can keep borehole stable is determined. Prediction precision and real-time operation ability of the proposed method are satisfying, which have been proved in practical application in TR oil field.     

MDT测井在油田的应用

MDT测井在石油勘探中已发挥了重要的作用,与常规测井项目相比,MDT具有重要的技术优势,但是,MDT的测试费用与其它测井项目相比价格相对较高,这就要求针对不同的评价对象和评价目的,有的放矢地做好现场施工,减少无效测试点,提高测试成功率,在尽可能减小测井成本的同时,以求地质效果的最大化,以测井局部的高投入,换取整个勘探项目的高效益。为此,在该项技术的推广应用过程中.对其适用的地质条件、测前设计方法、测量注意事项进行了系统的研究,形成了一套行之有效的现场施工和解释方法,这对于类似盆地具有借鉴意义。     

在测井中用一种组合进化神经网络识别油水层

分析了基于进化算法的神经网络,指出基于遗传算法的神经网络具有强的全局搜索能力,基于进化规划的神经网络具有强的局部寻优能力。在此基础上,提出了综合上述两种神经网络优点的组合进化神经网络,并应用于油水层测井解释中,降低了误判率。     

Exact Lithologic Boundary Detection Based on Wavelet Transform Analysis and Real-Time Investigation of Facies Discontinuities Using Drilling Data

Abstract

Exact detection of lithologic boundaries is one of the main challenges in exploration, drilling operations, and geology. Investigation of facies discontinuities has been performed using petrophysical data regarding sharp changes along the wellbore. Due to the fact that recorded well logging signals contain lots of high-frequency waves (noise), detection of the layer boundaries comes with some uncertainties that should be eliminated by denoising those signals. Wavelet transform analysis is a good approach to denoise the signals and its ability has been proven in several studies. In this study, implementation of wavelet transform analysis resulted in an innovative approach for exact differentiation of neighborhood lithologic units.

Detection of boundaries between different layers, especially the ones in the vicinity of the reservoir during drilling operations, is one of the crucial issues in petroleum well engineering. This purpose is usually achieved by cutting analysis and geological maps, which are not accurate enough and may cause substantial problems. Unconfined rock compressive strength can be considered as an accurate criterion to detect geological boundaries. In this study, an artificial neural network (ANN) model is developed that can predict the unconfined rock compressive strength of formations being drilled by importing 10 drilling parameters as inputs. Because rock strength will experience sudden changes while entering the next layer, it can be used as a key parameter to determine boundaries.