Loading...

Developing an Innovative Semi Empirical Model of a Density Prob using Neutron Source

Rafizadeh, Abolfazl | 2025

75 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 58328 (46)
  4. University: Sharif University of Technology
  5. Department: Energy Engineering
  6. Advisor(s): Hosseini, Abolfazl
  7. Abstract:
  8. In this project, the design and simulation of a neutron-gamma tool for predicting density, hydrogen percentage, drilling mud thickness, and barite percentage were carried out. For this purpose, neutron and gamma transport in a borehole environment was modeled using the MCNPX simulation software, and the impact of various parameters, such as mud thickness and barite weight percentage, on measurement accuracy was investigated. The simulated data were then compared with experimental results to validate the models. Subsequently, semi-empirical models were developed that improved the prediction accuracy of the tool response using simulation data and the physical principles of neutron and gamma transport. Next, a convolutional neural network (CNN) was designed and trained as an artificial intelligence model for predicting different formation properties. In this section, 256 different neural networks with various input combinations were trained, and their performance was evaluated. The use of optimized combinations of neutron and gamma detectors significantly improved the model accuracy. Additionally, optimization techniques such as the Adam algorithm for learning rate adjustment and Polynomial Decay for gradual learning rate reduction in the final stages of training contributed to improving the model's performance. Finally, the project demonstrated that combining semi-empirical models with neural networks can effectively increase the accuracy of predictions in neutron-gamma well-logging. These results can contribute to the optimization of neutron-gamma tools and the development of new methods in hydrocarbon exploration
  9. Keywords:
  10. Density Measurements ; Convolutional Neural Network ; Neural Networks ; Monte Carlo N-Particle (MCNP)Code ; Neutron-Gamma Density ; Gamma-Gamma Density ; Densitometer Tool

 Digital Object List

 Bookmark

No TOC