### Abstract

Estimation of the vertical force at the kick-off point (k.o.p) is of major concern to field engineers involved in horizontal drilling. Prior knowledge of the level of magnitude of the vertical force assists engineers in selecting appropriate hole paths to be drilled in order to minimize the risk of pipe failure. Current methods employed to approximate the vertical force are based on simple mathematical models that are not necessarily representative of field conditions. This paper presents a new approach based on the use of Artificial Neural Network (ANN) to predict the vertical forces at the k.o.p, which is required to push pipes through curved hole sections. The artificial neural network learns the relationship between field variables and the vertical forces through generated results using a finite element package and offers a quick and efficient way of estimating vertical forces at the k.o.p for various field conditions. The effect of pipe stiffness, hole radius (build-up rate), hole roughness, and the horizontal drag force applied at the end of build (e.o.b) are investigated. The finite element analysis and ANN results showed that the running force variation at the k.o.p increases as the horizontal force, buildup rate and drag increase. The results also showed that the pipe stiffness has negligible effect on the variation of running force at high buildup rate whereas a significant effect is observed at low buildup rate.

Original language | English |
---|---|

Pages (from-to) | 521-534 |

Number of pages | 14 |

Journal | Petroleum Science and Technology |

Volume | 19 |

Issue number | 5-6 |

DOIs | |

Publication status | Published - Jun 2001 |

### Keywords

- Artificial Neural Network
- Build-up rate (dogleg)
- High curvature well bores
- Horizontal drilling
- Vertical force at k.o.p

### ASJC Scopus subject areas

- Chemical Engineering(all)
- Energy Engineering and Power Technology
- Fuel Technology

## Fingerprint Dive into the research topics of 'Prediction of running forces in high curvature well bores using finite element analysis and artificial neural network'. Together they form a unique fingerprint.

## Cite this

*Petroleum Science and Technology*,

*19*(5-6), 521-534. https://doi.org/10.1081/LFT-100105271