Scribd Logo
Upload

Welcome to Scribd!

  • 126 views

    Final Year Project (Proposal Defense) (Autosaved)

    Uploaded by

    Ashwin Vel
    This document presents a project proposal to optimize history matching in reservoir simulation using artificial neural networks. The project will run from September 2016 to April 2017. The objectives are to create an ANN-enhanced dataset to reduce the number of simulation runs needed for history matching and provide useful insights into reservoir performance. Backpropagation algorithms will be applied to train the ANN model. Literature shows ANNs have been successfully used in petroleum engineering for applications like proxy modeling, well log interpretation, and reservoir characterization. MATLAB will be used to prepare the ANN model and data, while Schlumberger Eclipse will be used for simulations and history matching. The expected results are a trained ANN that can predict reservoir properties and produce a more geologically-driven

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd

    Final Year Project (Proposal Defense) (Autosaved)

    Uploaded by

    Ashwin Vel
    126 views10 pages
    This document presents a project proposal to optimize history matching in reservoir simulation using artificial neural networks. The project will run from September 2016 to April 2017. The objectives are to create an ANN-enhanced dataset to reduce the number of simulation runs needed for history matching and provide useful insights into reservoir performance. Backpropagation algorithms will be applied to train the ANN model. Literature shows ANNs have been successfully used in petroleum engineering for applications like proxy modeling, well log interpretation, and reservoir characterization. MATLAB will be used to prepare the ANN model and data, while Schlumberger Eclipse will be used for simulations and history matching. The expected results are a trained ANN that can predict reservoir properties and produce a more geologically-driven

    Original Description:

    Its about Artificial Neural Network

    Original Title

    Final Year Project (Proposal Defense) [Autosaved]

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document presents a project proposal to optimize history matching in reservoir simulation using artificial neural networks. The project will run from September 2016 to April 2017. The objectives are to create an ANN-enhanced dataset to reduce the number of simulation runs needed for history matching and provide useful insights into reservoir performance. Backpropagation algorithms will be applied to train the ANN model. Literature shows ANNs have been successfully used in petroleum engineering for applications like proxy modeling, well log interpretation, and reservoir characterization. MATLAB will be used to prepare the ANN model and data, while Schlumberger Eclipse will be used for simulations and history matching. The expected results are a trained ANN that can predict reservoir properties and produce a more geologically-driven

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    126 views10 pages

    Final Year Project (Proposal Defense) (Autosaved)

    Uploaded by

    Ashwin Vel
    This document presents a project proposal to optimize history matching in reservoir simulation using artificial neural networks. The project will run from September 2016 to April 2017. The objectives are to create an ANN-enhanced dataset to reduce the number of simulation runs needed for history matching and provide useful insights into reservoir performance. Backpropagation algorithms will be applied to train the ANN model. Literature shows ANNs have been successfully used in petroleum engineering for applications like proxy modeling, well log interpretation, and reservoir characterization. MATLAB will be used to prepare the ANN model and data, while Schlumberger Eclipse will be used for simulations and history matching. The expected results are a trained ANN that can predict reservoir properties and produce a more geologically-driven

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    of 10

    Final Year Project (Proposal Defense)

    Project Title: Optimizing History Match by utilization of Artificial


    Neural
    Network
    Period: September
    2016 April 2017

    Presented by : Ashwin Vel a/l Kalimuthu


    Date : 15/11/2016

    Introduction

    Relevancy of Project
    Problem Statement

    History matching requires large number of simulation runs because there are many
    uncertainties in reservoir characterization

    High non-linear behaviour of the reservoir

    Procedure of history matching is costly, time-consuming and difficult

    Automated/Assisted history matching lacks engineering judgement

    Introduction

    Relevancy of Project
    Objective of study:
    -

    Create an Artificial Neural Network (ANN) enhanced data set to use as input
    to reservoir simulation study.

    Reduce the number of simulations runs to save time of the expert

    Analysis of data using Artificial Neural Network

    Predict useful insights of the reservoir performance.

    Background Study
    Algorithm Applied

    Applies Backpropagation algorithm:

    Errors are propagated backwards

    Hidden Layer: acts a supervised learning layer

    Input and output is given and the network computes its own (error between them is
    calculated)

    The idea of backpropagation is to reduce this error

    Algorithm used: Scaled Conjugate Gradient/ Levenberg-Marquardt

    Literature Review
    Past Development

    APPLICATIONS OF ANN IN DIFFERENT


    FIELDS

    APPLICATIONS OF ANN IN PETROLEUM


    ENGINEERING

    Sales
    Forecasting

    (Yip, Hines and Yu 1997)

    Proxy model
    building

    (Zang, et. Al 2006)

    Medical
    Diagnosis

    (Amato, et al. 2013) (Lei and XingCheng 2010)

    Well long
    interpretation

    (Baldwin, Otte and Whealtley 1989)(JongSe and Jungwhan 2004)

    Stock market
    predictions

    (Adebiyi, et al. 2012)

    Well test data


    analysis

    (Al-Kaabi and Lee 1990)(Ershaghi, et al.


    1993)

    Customer
    research

    (Chattopadhyay, et al. 2012)

    Reservoir
    charactherization

    (Mohaghegh, et al 1995) (Ahmad, et al


    1997) (Singh, et al 2008)

    Industrial
    Process Control

    (Devadhas, Pushpakumar and


    Mary 2012)

    Inversion of seismic
    waveform

    (David 1993) (Yang and Huang 1991)

    Credit
    Evaluation

    (Baesens, et al. 2003)

    Prediction of PVT
    data

    (Gharbi and Elsharkawy 1997) (Briones, et


    al. 2009)

    Energy cost
    prediction

    (Yalcintas and Akkurt 2005)

    Identifying
    fractures and faults

    (Key, et al. 1997) (Sadiq and Nashawi


    2000) (Aminzadeh and deGrott 2005)

    Detecting
    hydrocarbons and
    forecast formation
    damage

    (Cheng-Dang, et al. 1994) (Aminzadeh


    and deGrott 2005) (Kalam, Al-Alawi and
    Al-Mukheini 1996)

    Methodology

    History Match using AI

    Methodology

    Preparing the Neural Network Structure


    INPU
    T

    HIDDEN
    LAYER

    ACTIVATION FUNCTION

    OUTPU
    T

    Data range: 10<k<100


    Data range:
    10<<100
    Data range:
    50<h<300

    Inputs

    Description

    Outputs

    Descriptio
    n

    (STB/D)

    Oil production difference from base


    case

    Permeabilit
    y

    (MSCF/D)

    Gas production difference from


    base case

    Porosity

    (STB/D)

    Water production difference from

    Thickness
    9

    Software Utilized

    Artificial Neural Network for History Match

    MATLAB
    Prepare data for ANN Model
    Design the ANN Model

    Schlumberger Eclipse
    Simulation and History Match
    10

    What to Expect?

    Artificial Neural Network for History Match

    Train and develop an Artificial Neural Network that can predict reservoir properties
    Produce more realistic reservoir modelling:
    -

    Incorporates more geology and reservoir information

    Model will be more geologically driven

    Reduces history matching time:

    11

    Reduces number of simulation runs

    Able to validate the model more accurately

    Provide better prediction data

    Acknowledgement
    FYP I Supervisor

    Berihun Mamo Negash

    Panels of Examiners

    Assoc. Prof. Dr Jang Hyun Lee

    Juhairi Aris bin Muhammad Shuhili

    Muhammad Azfar bin Mohamed

    Coordinator
    Friends
    Family

    REFERENCES
    Shahkarami, A., Mohaghegh, S. D., Gholami, V., & Haghighat, S. A. Artificial Intelligence (AI)
    Assisted History Matching.
    Srinivasan, K., & Ertekin, T. Development and Testing of an Expert System for Coalbed
    Methane Reservoirs Using Artificial Neural Networks.

    You might also like