Predictive Maintenance Water Injection Motor Menggunakan Multiple Regression Analysis

Ridwan, Ridwan (2025) Predictive Maintenance Water Injection Motor Menggunakan Multiple Regression Analysis. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022231002-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Predictive maintenance sangat penting dalam industri hulu minyak dan gas untuk meminimalkan downtime peralatan dan meningkatkan keandalan. Meskipun data real-time memberikan infromasi yang penting, potensi dalam pengambilan keputusan pemeliharaan seringkali terbatas oleh kompleksitas data, ketidakpastian, dan tingginya biaya diagnosis. Banyak industri masih bergantung pada metode analisis kondisi konvensional, yang sering mengakibatkan jadwal pemeliharaan yang tidak efisien dan kegagalan tak terduga. Studi ini mengembangkan model prediktif untuk memperkirakan waktu menuju kegagalan water injection motor menggunakan multiple linear regression berdasarkan data operasional aktual, termasuk getaran, temperatur, kandungan air, dan kandungan silikon. Untuk meningkatkan akurasi prediksi, digunakan pendekatan stepwise regression yang dikombinasikan dengan fungsi kuadratik. Berbeda dengan banyak studi yang berfokus pada deteksi kerusakan, model ini bertujuan untuk memprediksi time to failure guna mendukung perencanaan pemeliharaan yang proaktif dan tepat sasaran. Model yang dioptimalkan mencapai nilai R-squared sebesar 0,481 dan root mean square error sebesar 50,7, yang menunjukkan tingkat akurasi rendah namun dapat diterima untuk perencanaan pemeliharaan secara praktis. Vibrasi dan temperatur muncul sebagai prediktor paling signifikan terhadap waktu kegagalan, sementara kandungan air dan silikon memberikan pengaruh yang lebih kecil. Implementasi model ini meningkatkan ketersediaan peralatan dari 92,44% menjadi 95,96% serta keandalan pada 1.000 jam dari 86,11% menjadi 90,67%. Pendekatan praktis berbasis data ini mendukung pengambilan keputusan pemeliharaan yang lebih baik dan efisiensi operasional tanpa memerlukan investasi tambahan pada infrastruktur pemantauan.
==================================================================================================================================
Predictive maintenance is critical in the upstream oil and gas industry to minimize equipment downtime and improve reliability. Although real-time data provides valuable insights, its full potential in maintenance decision-making is often limited by data complexity, uncertainty, and high diagnostic costs. Many industries still depend on conventional condition analysis methods, which often lead to inefficient maintenance schedules and unexpected failures. This study develops a predictive model to estimate time to failure of water injection motors using multiple linear regression based on actual operational data, including vibration, temperature, water content, and silicon content. To enhance prediction accuracy, a stepwise regression approach combined with a quadratic function was applied. Unlike many studies that focus on fault detection, this model aims to predict time to failure for proactive and precise maintenance planning. The optimized model achieved an R-squared value of 0.481 and a root mean square error of 50.7, indicating a low but acceptable level of accuracy for practical maintenance planning. Vibration and temperature emerged as the most significant predictors of time to failure, while water content and silicon had minor influence. The implementation of this model increased equipment availability from 92.44% to 95.96% and reliability at 1,000 hours from 86.11% to 90.67%. This practical, data-driven approach supports better maintenance decisions and operational efficiency without requiring additional investment in monitoring infrastructure.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Predictive Maintenance, Time to failure, Multiple Regression, Getaran, Temperatur, Water Injection Motor, Vibration, Temperature
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK2785 Electric motors, Induction.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Mr Ridwan Ridwan
Date Deposited:	24 Jul 2025 09:51
Last Modified:	24 Jul 2025 09:51
URI:	http://repository.its.ac.id/id/eprint/121371

Actions (login required)

View Item