Integrasi Algoritma Xgboost Dan Api 581 Untuk Evaluasi Risiko Korosi Dan Estimasi Sisa Umur Pada Peralatan Statis Unit Pemrosesan Gas Berbasis Data Operasi Real-time

Gunawan, Kohan (2026) Integrasi Algoritma Xgboost Dan Api 581 Untuk Evaluasi Risiko Korosi Dan Estimasi Sisa Umur Pada Peralatan Statis Unit Pemrosesan Gas Berbasis Data Operasi Real-time. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5011221062_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Peralatan statis pada unit pemrosesan gas beroperasi di bawah fluktuasi termodinamika dan lingkungan aqueous sour ekstrem akibat paparan gas H2S dan CO2. Pendekatan perhitungan korosi statis konvensional seringkali gagal menangkap efek sinergi dari variabel dinamis, memicu bias pada estimasi kegagalan. Penelitian ini mengintegrasikan Machine Learning eXtreme Gradient Boosting (XGBoost) dengan kerangka Risk-Based Inspection API 581 untuk mengevaluasi risiko penipisan (thinning) secara presisi pada Feed-Product Exchanger dan Feed-Gas Chiller. Data operasional real-time (tekanan, suhu, laju alir, fraksi uap, pCO2, pH2S) diproses oleh XGBoost untuk memprediksi laju korosi dinamis, yang selanjutnya disubstitusikan ke dalam persamaan mekanika fraktur probabilistik API 581. Hasil komparasi pada Feed-Product Exchanger membuktikan keunggulan sensitivitas model; perhitungan statis mencatat laju korosi aktual komponen Tube sebesar 0,05 mm/tahun dengan Probability of Failure (PoF) 0,0044 dan Damage Factor (DF) 113,00, yang menempatkannya pada kategori level menengah di pemetaan Matriks Risiko (Risk Matrix). Setelah dievaluasi dengan XGBoost, laju korosi prediktif Tube terkoreksi naik menjadi 0,0555 mm/tahun, memicu lonjakan DF menjadi 194,62 dan menggeser posisinya ke arah kuadran risiko yang lebih kritis. Lebih lanjut, evaluasi model pada Feed-Gas Chiller mengungkap anomali ekstrem akibat pencucian kondensat asam; laju korosi Tube melonjak tajam hingga 0,1388 mm/tahun. Angka ini memproyeksikan penyusutan sisa umur menjadi 5,14 tahun dan menghasilkan lonjakan DF hingga 992,50 beserta PoF 0,0393. Secara definitif, hasil prediksi ini mengunci peralatan Feed-Gas Chiller pada kuadran risiko tertinggi (High Risk) pada Matriks Risiko fasilitas.
=================================================================================================================================
Static equipment in gas processing units operates under thermodynamic fluctuations and extreme aqueous sour environments due to H2S and CO2 gas exposure. Conventional static corrosion calculation approaches often fail to capture the synergistic effects of dynamic variables, triggering bias in failure estimation. This research integrates the eXtreme Gradient Boosting (XGBoost) Machine Learning algorithm with the API 581 Risk-Based Inspection framework to precisely evaluate thinning corrosion risks on a Feed-Product Exchanger and a Feed-Gas Chiller. Real-time operational data (pressure, temperature, flow rate, vapor fraction, pCO2, pH2S) is processed by XGBoost to predict the dynamic corrosion rate, which is subsequently substituted into the probabilistic fracture mechanics equations of API 581. The comparison results on the Feed-Product Exchanger prove the model's superior sensitivity; static calculations recorded an actual corrosion rate for the Tube component of 0.05 mm/year with a Probability of Failure (PoF) of 0.0044 and a Damage Factor (DF) of 113.00, placing it in the medium-level category on the Risk Matrix mapping. After being evaluated with XGBoost, the predictive corrosion rate of the Tube was corrected upwards to 0.0555 mm/year, triggering a DF surge to 194.62 and shifting its position towards a more critical risk quadrant. Furthermore, the model evaluation on the Feed-Gas Chiller reveals extreme anomalies due to acid condensate washing; the Tube's corrosion rate spiked sharply to 0.1388 mm/year. This figure projects a shrinkage in remaining life to 5.14 years and produces a DF surge up to 992.50 along with an absolute PoF of 0.0393. Definitively, these prediction results lock the Feed-Gas Chiller equipment into the highest risk quadrant (High Risk) on the facility's Risk Matrix.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	API 581, Laju Korosi, Matriks Risiko, Probability of Failure, Risk-Based Inspection, XGBoost, : API 581, Corrosion Rate, Probability of Failure, Risk Matrix, Risk-Based Inspection, XGBoost.
Subjects:	T Technology > T Technology (General) T Technology > T Technology (General) > T174 Technological forecasting T Technology > T Technology (General) > T174.5 Technology--Risk assessment.
Divisions:	Faculty of Industrial Technology > Material & Metallurgical Engineering > 28201-(S1) Undergraduate Thesis
Depositing User:	Kohan Gunawan
Date Deposited:	24 Jul 2026 03:51
Last Modified:	24 Jul 2026 03:51
URI:	http://repository.its.ac.id/id/eprint/136845

Actions (login required)

View Item