Loading...

Simulation of Drilling Fluid Movement Using Neural Network Modeling and Comparison with CFD Methods

Ershadnia, Reza | 2017

26 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 50036 (06)
  4. University: Sharif University of Technology
  5. Department: Chemical and Petroleum Engineering
  6. Advisor(s): Jamshidi, Saeid
  7. Abstract:
  8. We report on numerical and intelligent modeling of three-dimensional laminar flow of non-Newtonian fluids driven by axial pressure gradient in annular media consisting of a coaxially rotating inner cylinder. This is an example of a rotating Couette flow with vast applications in food, oil and gas, and chemical engineering industries. We study the dynamics of pressure in and the velocities (in three directions X, Y and Z) of every grid in presence of rotation and without that by means of CFD. Afterward, we develop a novel predictive model based on an Artificial Neural Network (ANN). The hybrid ANN framework is trained and tested by over 100000 data which collected by CFD. We then perform numerous CFD simulations governed by Naiver-Stokes set of equations, which correspond to each of the laboratory cases. Comparing the results with those obtained from an ANN, we find the both of our numerical and intelligent models to be more accurate and with larger applicability domain. However, the ANN model is a supervised learning paradigm and hence results in the highest accuracy, and is more time-consuming compared to CFD modeling. Statistical and graphical error analyses in addition to global sensitivity analysis are conducted, which further demonstrate the robustness of our proposed models specially ANN modeling is very accurate for predicting the non-Newtonian flow characteristics in annular rotating systems
  9. Keywords:
  10. Drilling Fluid ; Computational Fluid Dynamics (CFD) ; Artificial Neural Network ; Fluid Motion ; Annular Space

 Digital Object List

 Bookmark

No TOC