Sparse Neural Networks for Inference of Interwell Connectivity and Production Prediction

SPE Journal ◽  
2021 ◽  
pp. 1-22
Author(s):  
Junjie Yu ◽  
Atefeh Jahandideh ◽  
Siavash Hakim-Elahi ◽  
Behnam Jafarpour
Keyword(s):  
Neural Network ◽  
Feature Learning ◽  
Production Performance ◽  
Real Field ◽  
Injection Wells ◽  
Well Production ◽  
Second Stage ◽  
Proxy Model ◽  
Connectivity Patterns ◽  
Interwell Connectivity

Summary A new neural network-based proxy model is presented for prediction of well production performance and interpretation of interwell connectivity in large oil fields. The workflow consists of two stages. The first stage uses feature learning to describe the general input-output relations that exist among the wells and to characterize the interwell connectivity. In the second stage, the identified interwell connectivity patterns are used as network topology to develop a multilayer neural network proxy model, with nonlinear activation functions, to predict the production performance of each producer. The estimation of connectivity patterns in the first stage serves as an interpretable feature-learning step to improve the effectiveness of the proxy model in the second stage. Identification of interwell connectivity is based on the selection property of the ℓ1-norm minimization by promoting sparsity in the estimated connectivity weights. The sparsity of the network is motivated by the domain knowledge that each production well is mainly supported by a few nearby injection wells. That is, a proxy model that allows each producer to communicate with all the other wells in the field is inherently redundant and must have an unknown sparse representation. The sparse structure of the connection weights in the resulting network is detected by promoting sparsity during the training process. Two synthetic numerical examples, with known solutions, are first used to demonstrate the functionality and effectiveness of ℓ1-norm regularization for interwell connectivity identification. The workflow is then applied to a real field waterflooding example in Long Beach to predict oil production and to infer interwell connectivity information. Overall, the workflow provides a proxy model that effectively combines the implicit physical information from simulated data with reservoir engineering insight to identify interwell connectivity and to predict well production trends.

Apply neural network for improving production planning at Samarang petrol mine

2016 ◽  
Vol 9 (2) ◽  
pp. 126-143
Author(s):  
Quoc Trung Pham ◽  
Thi Kim Dung Phan
Keyword(s):  
Neural Network ◽  
Production Planning ◽  
Petroleum Industry ◽  
Allocation Problem ◽  
Production Performance ◽  
Well Test ◽  
Operational Parameters ◽  
Content Type ◽  
Well Production ◽  
Non Linear

Purpose – Artificial neural network (ANN) is considered a good solution for building non-linear relationship between input and output parameters, which is suitable for solving production back allocation, which is the most important step for production planning of petroleum mine. The purpose of this paper is to suggest a solution for solving production back allocation problem at Samarang petrol mine based on ANN approach. Design/methodology/approach – In this study, well operational parameters’ surveillance was conducted and ANN was used to build relationships between operation parameters and production rates. Experimental method is used for testing and evaluating the possibility of using ANN for supporting production planning at Samarang mine. Findings – Consequently, the proposed ANN solution can increase the accuracy of predicted values and could be used for supporting production planning at Samarang mine. Because ANN uses well test data for training and predicting (without adding new devices), it could be a feasible and cheap solution. Research limitations/implications – There is a need for applying other methods, such as: support machine vector, non-linear autoregressive models, etc. for better evaluation of ANN solution. Practical implications – The ANN models helped operation engineers to understand well production performance and make decision to improve production plan in timely manner. This solution could be generalized for the whole mine or to similar petroleum mines in practice. Originality/value – This paper aims to propose a solution based on ANN for solving production back allocation problem of petroleum industry. The solution is tested at Samarang mine.

Economic benefit of shale gas exploitation based on back propagation neural network

2020 ◽  
Vol 39 (6) ◽  
pp. 8823-8830
Author(s):  
Jiafeng Li ◽  
Hui Hu ◽  
Xiang Li ◽  
Qian Jin ◽  
Tianhao Huang
Keyword(s):  
Neural Network ◽  
Shale Gas ◽  
Bp Neural Network ◽  
Large Scale ◽  
Linear Prediction ◽  
Back Propagation ◽  
Back Propagation Neural Network ◽  
Economic Benefits ◽  
Gas Well ◽  
Well Production

Under the influence of COVID-19, the economic benefits of shale gas development are greatly affected. With the large-scale development and utilization of shale gas in China, it is increasingly important to assess the economic impact of shale gas development. Therefore, this paper proposes a method for predicting the production of shale gas reservoirs, and uses back propagation (BP) neural network to nonlinearly fit reservoir reconstruction data to obtain shale gas well production forecasting models. Experiments show that compared with the traditional BP neural network, the proposed method can effectively improve the accuracy and stability of the prediction. There is a nonlinear correlation between reservoir reconstruction data and gas well production, which does not apply to traditional linear prediction methods

A Proxy Flow Modelling Workflow to Estimate Gridded Dynamic Properties and Well Production Rates by Deep Learning Algorithms

2021 ◽  
Author(s):  
Soumi Chaki ◽  
Yevgeniy Zagayevskiy ◽  
Terry Wong
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Dynamic Properties ◽  
Reservoir Management ◽  
Flow Modeling ◽  
Static Properties ◽  
Production Rates ◽  
Reservoir Properties ◽  
Well Production ◽  
Reservoir Simulator

Abstract This paper proposes a deep learning-based framework for proxy flow modeling to predict gridded dynamic petroleum reservoir properties (like pressure and saturation) and production rates for wells in a single framework. It approximates the solution of a full physics-based numerical reservoir simulator, but runs much more rapidly, allowing users to generate results for a much wider range of scenarios in a given time than could be done with a full physics simulator. The proxy can be used for reservoir management tasks like history matching, uncertainty quantification, and field development optimization. A deep-learning based methodology for accurate proxy-flow modeling is presented which combines U-Net (a variant of convolutional neural network) to predict gridded dynamic properties and deep neural network (DNN) models to forecast well production rates. First, gridded dynamic properties, such as reservoir pressure and phase saturations, are predicted from static properties like reservoir rock porosity and absolute permeability using a U-Net. Then, the static properties and the dynamic properties predicted by the U-Net are input to a DNN to predict production rates at the well perforations. The inclusion of U-net predicted pressure and saturations improves the quality of the well rate predictions. The proposed methodology is presented with the synthetic Brugge reservoir discretized into grid blocks. The U-Net input consists of three properties: dynamic gridded reservoir properties (such as pressure or fluid saturation) at the current state, static gridded porosity, and static gridded permeability. The U-Net has only one output property, the target gridded property (such as pressure or saturation) at the next time step. Training and testing datasets are generated by running 13 full physics flow simulations and dividing them in a 12:1 ratio. Nine U-Net models are calibrated to predict pressures/saturations, one for each of the nine grid layers present in the Brugge model. These outputs are then concatenated to obtain the complete pressure/saturation model for all nine layers. The constructed U-Net models match the distributions of generated pressures/saturations of the numerical reservoir simulator with a correlation coefficient value of approximately 0.99 and above 95% accuracy. The DNN models approximate well production rates accurately from U-Net predicted pressures and saturations along with static properties like transmissibility and horizontal permeability. For each well and each well perforation, the production rate is predicted with the DNN model. The use of the constructed proxy flow model generates reservoir predictions within a few minutes compared to the hours or days typically taken by a full physics flow simulator. The direct connection that is established between the gridded static and dynamic properties of the reservoir and well production rates using U-Net and DNN models has not been presented previously. Using only a small number of runs for its training, the workflow matches the numerical reservoir simulator results with reduced computational effort. This helps reservoir engineers make informed decisions more quickly, resulting in more efficient reservoir management.

A Two-Stage Combination Model for Wind Power Forecasting

2015 ◽  
Vol 737 ◽  
pp. 9-13
Author(s):  
Jun Zhang ◽  
Yuan Hao Wang ◽  
Ying Yi Li ◽  
Feng Guo
Keyword(s):  
Neural Network ◽  
Bp Neural Network ◽  
Wind Farm ◽  
General Trend ◽  
Two Stage ◽  
Combination Model ◽  
Second Stage ◽  
Combination Forecasting ◽  
The Mean ◽  
Power Sequence

With the wind farm data from the southeast coast this paper builds a two-stage combination forecasting model of output power based on data preprocessing which include filling up missing data and pre-decomposition. The first stage is a composite prediction of decomposed power sequence in which a time series and optimized BP neural network predict the general trend and the correlation of various factors respectively. The second stage is BP neural network with its input is the results of first stage. The effectiveness and accuracy of the two-stage combination model are verified by comparing the mean square error of the combination model and other models.

Neural-Network-Based Feature Learning: Auto-Encoder

2020 ◽  
pp. 195-217
Author(s):  
Haitao Zhao ◽  
Zhihui Lai ◽  
Henry Leung ◽  
Xianyi Zhang
Keyword(s):  
Neural Network ◽  
Feature Learning

scholarly journals Survey Analysis of Robust and Real-Time Multi-Lane and Single Lane Detection in Indian Highway Scenarios

2021 ◽  
Vol 309 ◽  
pp. 01117
Author(s):  
A. Sai Hanuman ◽  
G. Prasanna Kumar
Keyword(s):  
Neural Network ◽  
System Integration ◽  
Large Scale ◽  
Feature Learning ◽  
Semantic Segmentation ◽  
Lane Detection ◽  
Learning Approaches ◽  
Survey Analysis ◽  
Lane Recognition ◽  
Continuous Frames

Studies on lane detection Lane identification methods, integration, and evaluation strategies square measure all examined. The system integration approaches for building a lot of strong detection systems are then evaluated and analyzed, taking into account the inherent limits of camera-based lane detecting systems. Present deep learning approaches to lane detection are inherently CNN's semantic segmentation network the results of the segmentation of the roadways and the segmentation of the lane markers are fused using a fusion method. By manipulating a huge number of frames from a continuous driving environment, we examine lane detection, and we propose a hybrid deep architecture that combines the convolution neural network (CNN) and the continuous neural network (CNN) (RNN). Because of the extensive information background and the high cost of camera equipment, a substantial number of existing results concentrate on vision-based lane recognition systems. Extensive tests on two large-scale datasets show that the planned technique outperforms rivals' lane detection strategies, particularly in challenging settings. A CNN block in particular isolates information from each frame before sending the CNN choices of several continuous frames with time-series qualities to the RNN block for feature learning and lane prediction.

Sign in / Sign up

Export Citation Format

Share Document