Loading...

Sealability Prediction of Premium Threaded Connection Based on Machine Learning

Amiri, Mohammad | 2025

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58314 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Khodaygan, Saeed
  7. Abstract:
  8. The use of proprietary threaded connections in various industries, including the oil and gas industry for the extraction of oil and gas from deep underground, is increasingly expanding due to the need to reach greater depths. These types of connections, due to their unique design which includes trapezoidal thread geometry, torque shoulder to protect against connection failure during makeup process, and metal-to-metal seal to sealing the connection exhibit strong resistance to high pressure high temperature (HPHT) condition and combined cycling loads during operation. One of the most critical performance parameters of these connections is their sealing capability. Various standards have been established to assess the performance of these connections, the most important of which is API 5C5 (ISO 13679), which evaluates this aspect through a series of severities testing protocols. However, a key challenge with these tests is their high cost and time-consuming nature. The aim of this thesis is to develop an artificial neural network using machine learning algorithms to predict the sealing performance of proprietary connections based on the input of influential parameters. This neural network will be developed using the Multi-Layer Perceptron (MLP) algorithm. To train and then evaluate the accuracy of this algorithm in predicting the sealing performance of the connections, a required dataset will be generated using numerical methods and finite element analysis of the connections performance under various operating conditions. The generated data will then be validated against reliable results from previous research. After creating the required dataset, an appropriate portion will be allocated to the training dataset for algorithm learning, and another portion will be used as the test dataset to evaluate the performance of the developed neural network
  9. Keywords:
  10. Finite Element Analysis ; Artificial Neural Network ; ISO 13679 (API 5C5)Standard ; Metal to Metal Sealing ; Multi-Layer Perceptron Algorithm ; Premium Threaded Connection

 Digital Object List

 Bookmark

No TOC