PREDICTION OF STUCK PIPE USING ARTIFICIAL NEURAL NETWORK: A CASE STUDY ON NIGER DELTA FIELDS OF NIGERIA

0
407

ABSTRACT

Drilling is a process that involves the procurement of natural resources such as oil and gas which holds prime importance in today’s world, Drilling practices abounds with a number of complications and an efficient way of dealing with such problems is key to the continuity of the process.

One of such problems is stuck pipe, stuck pipe is a common problem in the industry and it accounts for major rig time loss known as Non Productive Time (NPT) and also accounts for billions of dollars wasted annually in the petroleum industry.

The purpose of this project to implement a powerful machine learning tool known as the Artificial Neural Network in the prediction of stuck pipe using Niger Delta fields as a case study, The ANN is a Matlab built in function and computational system inspired by the structure, processing method and learning ability of the human brain.

The ANN has the ability to take multiple inputs ( plastic viscosity, yield point and gel strength at 10 seconds and 10 minutes), a target ( mud weight ) to produce a single output which is the prediction of the occurrence of stuck pipe. This was successfully carried in this research study. It is therefore shown in this study that the ANN can be successfully used to predict the occurrence of stuck pipe. Thus, they can be utilized with real-time data representing the results on a log viewer which can help reduce the occurrence of getting stuck while drilling and all the complications that comes with this occurrence.

TABLES OF CONTENT

TITLE PAGE………………………………………………………………………………………………………………….. i

CERTIFICATION………………………………………………………………………………………………………….. ii

DEDICATION……………………………………………………………………………………………………………… iii

ACKNOWLEDGEMENT………………………………………………………………………………………………. iii

ABSTRACT…………………………………………………………………………………………………………………… v

LIST OF FIGURES……………………………………………………………………………………………………….. ix

LIST OF TABLES………………………………………………………………………………………………………….. x

CHAPTER ONE…………………………………………………………………………………………………………….. 1

1.0      INTRODUCTION…………………………………………………………………………………………….. 1

CHAPTER TWO………………………………………………………………………………………………………….. 12

CHAPTER THREE………………………………………………………………………………………………………. 21

CHAPTER FOUR………………………………………………………………………………………………………… 27

CHAPTER FIVE………………………………………………………………………………………………………….. 36

STUCK PIPE PREDICION SIMULATIONAND PREDICTION CODES USING ANN.. 41

APPENDIX 2………………………………………………………………………………………………………………. 53

STUCK PIPE PREDICION EXPRIMENTAL GRAPHS FOR INPUT AND OUTPUT…… 53

LIST OF FIGURES

Figure 1: Cross sectional view of differentially stuck pipe (http://www.drillingformulas.com)……… 4

Figure 2: Top view of a differentially stuck pipe (1997 Drillers Stuck pipe Handbook)………………. 4

Figure 3 : schematic of the ANN system…………………………………………………………………………….. 14

Figure 4 : Workspace for inputting time series…………………………………………………………………….. 24

Figure 5: Interface for making network architecture choices………………………………………………….. 25

Figure 6: Interface for choosing training algorithm………………………………………………………………. 25

Figure 7: Interface for choosing validation and testing data…………………………………………………… 26

Figure 8: Graph of Experimental value of plastic viscosity versus number of runs……………………. 29

Figure 9: Graph of experimental value of yield point versus number of runs…………………………… 30

Figure 10:Graph of experimntal value of gel (10 secs) versus number of runs…………………………. 30

Figure 11:Graph of experimental value of gel (10 mins) versus number of runs………………………. 31

Figure 12: Graph of experimental value of mud weight versus number of runs………………………… 32

Figure 13: A graph showing variation between the experimented and simulated value for NARX.. 33

Figure 14: A graph showing variation between the experimented and predicted value for NARX.. 33

Figure 15: A graph showing variation between the experimented and simulated value for NAR…. 34

Figure 16: A graph showing variation between the experimented and simulated value for NAR…. 34

Figure 17: A graph showing variation between the experimented and simulated value for NOI…… 35

LIST OF TABLES

Table 2.1: Input and output data for parameter selection………………………………………… 22

Table 4.1: MSE for Levenberg Marquadt……………………………………………………………… 27

Table 4.2: MSE for Bayesian Regularization…………………………………………………………… 27

Table 4.3: MSE for scaled Conjugate Gradient………………………………………………………. 28

CHAPTER ONE

     INTRODUCTION

Over several years the petroleum industry has been facing challenges associated with stuck pipe. Stuck pipe has caused a major drilling cost for the drilling industry worldwide and various cost estimates carried out have indicated that the cost of fixing stuck pipe issues exceeds $250 million per year (Bradley et al., 1991).

Problems of stuck pipe can range from minor inconveniences to increase in drilling cost up to major complications which will lead to altered drilling due to the inability to drill when this occurs resulting in major time loss.

A major key to the reduction of this phenomenon is the ability to correctly or even better, accurately predict the occurrence of stuck pipe.

Generally, stuck pipe is described as any restriction of upward or downward movement of drill string and/or pipe rotation and leads to a situation where the pipe cannot be freed from the hole without damaging the pipe, and without exceeding the drill rigs maximum allowed hook load. The portion of the drill string that cannot be rotated or moved vertically is known as the stuck pipe.

There are several causes of stuck pipe which include poor hole cleaning, key sitting, collapsed casing, junk, cement related problems, mobile formation, geo-pressured formation, fractured formation. However, the causes of stuck pipe can be classified under two broad categories which are mechanical and differential sticking.

     Mechanical sticking:

This is the limiting or prevention of motion of the drill string by anything other than differential pressure sticking. According to drillers stuck pipe handbook (1997) by Schlumberger, Mechanical Sticking can be caused by the following:

  1. Inadequate hole cleaning
  • Formation instability (brittle, sloughing, or swelling shales)
  • Key seating
  • Under gauge hole
  • Tectonically stressed formations
  • Plastic or mobile formations
  • Under pressured formations
  • Junk
  • Ledges and doglegs
  1. Collapsed casing/tubing
  1. Unconsolidated formations
  1. Large boulders falling into the hole
  1. Running large gauge tools
  1. Cement blocks
  1. Green cement

However, most cases of mechanical sticking can be avoided by proper well planning, optimal mud design and right directional planning.

     Differential sticking

Differential pipe sticking is one of the stuck pipe mechanisms that have had a major impact on drilling efficiency and well costs (Adams, 1977; Weakley, 1990; Wisnie and Zheiwei, 1994); it is in most drilling organization, the greatest drilling problem worldwide in terms of time and financial cost.

This is a condition whereby the drill string cannot be moved (rotate or reciprocated) along the axis of the wellbore. Differential sticking occurs when high contact force caused by low reservoir pressure, high wellbore pressures, or both, are exerted over a sufficiently large area of the drill string. It is important to note that the sticking force is a product of the differential pressure between the wellbore and the reservoir and the area that the differential pressure is acting upon, this means that a relative low differential pressure (δp) applied over a large working area can be effective in sticking the pipe as can a high differential pressure applied over a small area.