Goto

Layer | Fill | Outline |
---|

Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|

Fill | Stroke |
---|---|

Hongkui, Ge (University of Petroleum, (East China) Dongying) | Yingsong, Lin (University of Petroleum, (East China) Dongying) | Shanzhou, Ma (University of Petroleum, (East China) Dongying) | Lili, Song (University of Petroleum, (East China) Dongying)

clastic parameter, coefficient, correlation, dynamic clastic parameter, dynamic loading, dynamic moduli, elastic parameter, frequency, friction, moduli, Poisson, Reservoir Characterization, rock elastic parameter, rock static dynamic clastic parameter, static poisson, static property, static test, strain amplitude, Upstream Oil & Gas, wave velocity

SPE Disciplines: Reservoir Description and Dynamics > Reservoir Characterization > Seismic processing and interpretation (1.00)

Tariq, Zeeshan (King Fahd University of Petroleum & Minerals) | Elkatatny, Salaheldin (King Fahd University of Petroleum & Minerals) | Mahmoud, Mohamed (King Fahd University of Petroleum & Minerals) | Abdulraheem, Abdulazeez (King Fahd University of Petroleum & Minerals)

Abstract Evaluating the elastic parameters of reservoir section are very essential for oil and gas industry in alleviating the risks associated with the drilling and production phases of the reservoir. These parameters are used in the optimization of the well placement, well bore instability, completion design, and draw-down limits to avoid sanding, hydraulic fracturing, reservoir subsidence and in many more applications. To carry out any aforementioned operations, a continuous profiles of rock elastic parameters are needed. Misleading estimation of elastic parameters may wrongly lead to heavy investment decisions and inappropriate field development plans. Retrieving reservoir rock samples throughout the depth of the reservoir section and performing laboratory tests on them are extremely expensive as well as time consuming. Therefore, these parameters are estimated from empirical correlations. Most of the previous models/correlations were developed using linear or non-linear regression techniques. Artificial intelligence tool once optimized for training can successfully model elastic parameters since these tools are capable of capturing highly complex and non-linear relationship between the input parameters and the target parameter. The objective of this study is to develop a robust and an accurate model for static Young's modulus based on the wireline logs as an input using three artificial intelligence tools (artificial neural network, adaptive neuro fuzzy inference system and support vector machine). The data on which these AI models are built comprises of more than 680 real field data points from different fields covering a wide range of values. Based on the minimum error and the highest coefficient of determination between actual and simulated data artificial neural network is selected as the proposed AI model to predict static Young's modulus. A comparison between the static Young's moduli predicted by the proposed model with the published models /correlations reveals that neural network model gives significantly less average absolute percentage error. Finally, a rigorous empirical correlation is developed using the weights of ANN model in order to make the AI black box model as a white box and universal that can be usable for field applications.

Artificial Intelligence, average absolute percentage error, bulk density, coefficient, correlation, dynamic young, elastic parameter, empirical correlation, estatic model, fuzzy logic, hidden layer, input parameter, machine learning, mathematical model, modulus model, neural network, Petroleum Engineer, predict static young, static young, travel time, Upstream Oil & Gas

Country:

- Europe (1.00)
- North America (0.93)

SPE Disciplines:

Technology:

- Information Technology > Artificial Intelligence > Representation & Reasoning > Uncertainty > Fuzzy Logic (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning > Support Vector Machines (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning > Regression (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Neural Networks (1.00)

Xu, Kai (Sinopec Geophysical Research Institute) | Sun, Zhentao (Sinopec Geophysical Research Institute) | Wang, Shixing (Sinopec Geophysical Research Institute) | Tang, Jinliang (Sinopec Geophysical Research Institute) | Zhang, Ruyi (Sinopec Geophysical Research Institute)

Reservoir characterization involves the estimation elastic parameters from well-log data and seismic data, nonlinear elastic parameters estimation based on machine learning is a popular method for reservoir characterization. The common process is to construct a comprehensive mapping relationship between seismic data including multi attributes and well data, but the prediction accuracy may be low when using one prediction model in different sedimentary environments. In order to solve this problem, we put forward the new method of multi-model prediction based on machine learning to estimate elastic parameters directly, in other words, we use corresponding deep neural network to estimate elastic parameters in different sedimentary environments. In this paper, our work mainly includes three parts: firstly, waveform classification based on Selforganization map, secondly, training and optimizing multi prediction models based on Long Short Term Memoryrecurrent neural networks, thirdly, elastic parameters estimation using multi-model. This method is successfully applied in model and real data, it shows the estimation results are more reasonable and effective. Presentation Date: Monday, October 12, 2020 Session Start Time: 1:50 PM Presentation Time: 4:45 PM Location: 351F Presentation Type: Oral

SEG-2020-3428172

algorithm, Artificial Intelligence, deep learning, elastic parameter estimation, estimate elastic parameter, estimation, estimation result, exploration geophysicist 10, input data, machine learning, neural network, parameter estimation, Reservoir Characterization, seg international exposition, seismic data, seismic waveform, Upstream Oil & Gas, well-log data

Oilfield Places: Africa > Angola > South Atlantic Ocean > Kwanza Basin > Block 18 > Greater Plutonio Field (0.99)

SPE Disciplines:

- Reservoir Description and Dynamics > Reservoir Characterization > Seismic processing and interpretation (1.00)
- Data Science & Engineering Analytics > Information Management and Systems > Neural networks (1.00)
- Data Science & Engineering Analytics > Information Management and Systems > Artificial intelligence (1.00)

Technology:

Abstract Rock mechanical parameters of reservoir rocks play an extremely important role in solving problems related to almost all operations in oil or gas production. A continuous profile of these parameters along the depth is essential to analyze these problems which include wellbore stability, sand production, fracturing, reservoir compaction, and surface subsidence. The mechanical parameters can be divided into three main groups, viz., elastic parameters, strength parameters, and in-situ stresses. Even the profile of in-situ stresses with depth is estimated using logs with elastic parameters as an essential input. The focus of this work is on the prediction of elastic parameters and their variation with the depth of a given reservoir. For an isotropic medium, there are two independent elastic parameters, viz., Young's modulus and Poisson's ratio. Generally, logging data consisting of density, compressional and shear wave velocities are used to estimate these parameters. However, these data provide dynamic elastic properties which are different from static values, especially in case of Young's modulus. To get continuous rock samples throughout the depth of the reservoir and conduct triaxial tests to determine the static values of these parameters is extremely expensive. Consequently, static values of Young's modulus and Poisson's ratio obtained from laboratory testing on rock samples acquired from selected intervals are used to calibrate the dynamic data obtained from logs. However, since the rock layers vary in their properties with depth, a realistic estimation of static elastic values of the rock is still a challenge. The problem is more prominent in limestone rocks compared to sandstone rocks. Further, shear velocity data is not always available from well logs, making the problem more difficult. An extensive experimental program was carried out first to obtain the static values of elastic parameters of reservoir rock samples at reservoir conditions of high pressure. Log data consisting of different variables such as density, velocity, and porosity from the same wells were also obtained. Three artificial intelligence methods viz. Neural Network, Fuzzy Logic and Functional Network, were used to obtain a continuous profile of static elastic parameters along the depth. The results obtained from these approaches were compared using log inputs. The strengths of each of these approaches are also discussed.

algorithm, Artificial Intelligence, corresponding plot, dataset, different artificial intelligence technique, elastic parameter, fuzzy logic, learnwh, log analysis, machine learning, mechanical properties, network model, neural network, neuron, Poisson, prediction, Reservoir Characterization, reservoir geomechanics, rock mechanical parameter, spe 126094, static value, training testing learning, Upstream Oil & Gas, variation, well logging, Wellbore Design, wellbore integrity, young modulus

Country:

- North America > United States (0.47)
- Asia (0.46)

SPE Disciplines:

- Well Drilling > Wellbore Design > Wellbore integrity (1.00)
- Reservoir Description and Dynamics > Reservoir Characterization > Reservoir geomechanics (1.00)
- Reservoir Description and Dynamics > Formation Evaluation & Management > Open hole/cased hole log analysis (1.00)
- Data Science & Engineering Analytics > Information Management and Systems > Artificial intelligence (1.00)

Technology:

- Information Technology > Artificial Intelligence > Representation & Reasoning > Uncertainty > Fuzzy Logic (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (1.00)
- Information Technology > Artificial Intelligence > Machine Learning > Neural Networks (1.00)

This work describes the application of artificial neural networks to directly estimate P-wave and S-wave velocity values from simulated seismic data. This application uses a neural network as an inversion operator to estimate a solution to the AVO inversion problem. Neural network results showed excellent correlation of P- and S-wave velocities and calculated Poisson’s ratio with their true values. Even in the presence of substantial noise corruption, the average percentage error between network predictions and true values was less than 10%. Neural network results were also compared with previously reported inversion results from two different approaches. In almost all cases, the neural networks demonstrated superior performance.

approximation, Artificial Intelligence, artificial neural network solution, AVO inversion problem, elastic parameter, inversion, inversion method, machine learning, Mallick, neural network, noise corruption, Poisson, reflection coefficient, Reservoir Characterization, s-wave velocity, Shuey, Upstream Oil & Gas, Zoeppritz equation

SPE Disciplines:

Technology: Information Technology > Artificial Intelligence > Machine Learning > Neural Networks (1.00)

Thank you!