Dissertation/Thesis Abstract

Drilling Hydraulics Optimization Using Neural Networks
by Wang, Yanfang, M.S., University of Louisiana at Lafayette, 2014, 73; 1592757
Abstract (Summary)

This study works with pump pressure prediction considering the dependent parameter ROP, formation variable like depth, and independent parameters like RPM, Torque, Differential Pressure, Hook Load, SPM, and mud properties. Reliable prediction of pump pressure provides an early warning of circulation problems, washout, underground blowout, and kicks helping the driller to make corrections and to safely avoid major problems.

In this study, a comprehensive overview of drilling optimization was provided. An ANN model to predict hydraulics was implemented through the fitting tool of MATLAB. We evaluated the training performance of the ANN model on the basis of mean MSE and efficiency coefficient `R'. Following the determination of the optimum model the sensitivity analysis of input parameters on the created model was investigated by using forward regression method. At last, the rest data from the selected well samples was applied for simulation to verify the quality of the developed model.

The simulation result was promising. This model can accurately predict pump pressure versus depth in analogous formations. In input sensitivity analysis, a total number of 9,360 networks were created. After this process, an overall ranking of sensitivity degree was then provided to show the impact of each individual input parameter on this model. The result was that Depth has top impact on this model and the followings were total stroke per minute (total SPM), differential pressure, etc. However, due to the limited dataset, this model should be used with cautions. For example, this model can apply vertical wells because dataset related to inclination angle and dog leg severity was not taken into account.

Indexing (document details)
Advisor: Salehi, Saeed
Commitee: Boukadi, Fathi, Guo, Boyun
School: University of Louisiana at Lafayette
Department: Petroleum Engineering
School Location: United States -- Louisiana
Source: MAI 54/06M(E), Masters Abstracts International
Subjects: Petroleum engineering
Keywords: Drilling optimization, Hydraulics, Matlab, Neural network
Publication Number: 1592757
ISBN: 9781321884494