Optimasi Matriks Risiko Risk Based Inspection Menggunakan Artificial Neural Network Pada Item High Temperature Shift Converter (1-R-201) Di PT. ABC

Danarta, Ficho Alfa (2026) Optimasi Matriks Risiko Risk Based Inspection Menggunakan Artificial Neural Network Pada Item High Temperature Shift Converter (1-R-201) Di PT. ABC. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5011221054-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

Petrokimia tergolong industri dengan jumlah peralatan proses kompleks dan area operasi korosif yang memungkinkan terjadinya kegagalan melalui mekanisme kerusakan berupa kebocoran, pecah, atau ledakan. Salah satu item peralatan proses adalah high temperature shift converter (HTSC) (1-R-201) milik PT. ABC yang berfungsi meningkatkan kadar gas hidrogen (H₂) dan karbon dioksida (CO₂) melalui reaksi antara gas karbon monoksida dengan steam (H₂O). High temperature shift converter (HTSC) (1-R-201) beroperasi di temperatur 360-430 °C dan tekanan 30 kg/cm²G. Penelitian ini bertujuan melakukan penilaian risiko berbasis API RP 581 serta mengembangkan model Artificial Neural Network (ANN) untuk prediksi kategori Probability of Failure (POF), Consequence of Failure (COF), dan risiko. Model artificial neural network dibangun menggunakan 642 data dan 13 fitur input parameter operasi/desain dan karakteristik material. Sampel data dibagi ke tiga subset, data training (79,8%), data validasi (10,1%), dan data testing (10,1%). Hasil asesmen berbasis API 581 menunjukkan risiko HTSC di kategori medium akibat faktor kerusakan thinning saat beroperasi mengikuti ketentuan desain. Sedangkan pengembangan ANN menghasilkan prediksi risiko HTSC di kategori low, medium, dan medium-high dengan overall accuracy 81,5%. Performa terbaik dicapai saat prediksi POF dengan akurasi 100%, sedangkan akurasi prediksi COF sebesar 74%. Pengembangan ANN menghasilkan peningkatan risiko seiring kenaikan temperatur dan tekanan operasi melampaui ketetapan desain. Pada kondisi tersebut, mekanisme kerusakan dominan bergeser dari thinning ke High Temperature Hydrogen Attack, yang membuat peningkatan risiko medium ke medium-high. Analisis SHapley Additive exPlanations (SHAP) menunjukkan temperatur operasi, tekanan operasi, dan efektivitas inspeksi sebagai fitur yang berkontribusi besar terhadap prediksi POF dan COF, sehingga menjadi faktor utama yang memengaruhi risiko kegagalan HTSC.
======================================================================================================================================
The petrochemical industry involves complex process equipment operating under corrosive environments, making it susceptible to failures such as leakage, rupture, and explosion. One of the critical process units is the High Temperature Shift Converter (HTSC) (1-R-201) at PT. ABC, which increases hydrogen (H₂) and carbon dioxide (CO₂) concentrations through the water-gas shift reaction between carbon monoxide and steam (H₂O). The HTSC operates at temperatures of 360–430 °C and a pressure of 30 kg/cm²G. This study aims to assess the equipment risk based on API RP 581 and to develop an Artificial Neural Network (ANN) model for predicting the categories of Probability of Failure (POF), Consequence of Failure (COF), and overall risk. The ANN model was developed using 642 datasets with 13 input features representing operating conditions, design parameters, and material characteristics. The dataset was divided into training (79.8%), validation (10.1%), and testing (10.1%) subsets. The API RP 581 assessment classified the HTSC as medium risk, primarily due to the dominance of the thinning damage mechanism under design operating conditions. The developed ANN successfully predicted low, medium, and medium-high risk categories with an overall accuracy of 81.5%. The model achieved its best performance in POF prediction with 100% accuracy, while COF prediction reached 74% accuracy. Furthermore, the ANN predicted an increase in risk as operating temperature and pressure exceeded the design limits, accompanied by a shift in the dominant damage mechanism from thinning to High Temperature Hydrogen Attack (HTHA), resulting in the escalation of risk from medium to medium-high. SHapley Additive exPlanations (SHAP) analysis identified operating temperature, operating pressure, and inspection effectiveness as the most influential features affecting POF and COF predictions, highlighting their critical roles in determining the failure risk of the HTSC.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	API 581, Artificial Neural Network, High Temperature Shift Converter, Matriks Risiko, Risk-Based Inspection, SHapley Additive exPlanations
Subjects:	Q Science Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Material & Metallurgical Engineering > 28201-(S1) Undergraduate Thesis
Depositing User:	Ficho Alfa Danarta
Date Deposited:	23 Jul 2026 09:01
Last Modified:	23 Jul 2026 09:01
URI:	http://repository.its.ac.id/id/eprint/137056

Actions (login required)

View Item