Kajian Longitudinal Tren Incident Dan Accident Pada Kapal Tanker Analisis Akar Masalah Menggunakan Metode Bayesian Network

Prasetya, Wulan (2026) Kajian Longitudinal Tren Incident Dan Accident Pada Kapal Tanker Analisis Akar Masalah Menggunakan Metode Bayesian Network. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6019241023-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Penelitian ini mengkaji tren longitudinal incident dan accident pada sektor machinery kapal tanker PT. Pertamina International Shipping (PIS) periode 2023-2025 sebagai analisis tren awal berbasis tiga tahun pengamatan, dengan pemahaman bahwa periode ini belum cukup untuk menyimpulkan pola jangka panjang secara definitif. Data mencakup 154 insiden (35/2023; 73/2024; 46/2025) dengan laju insiden per kapal 1,13 / 4,86 / 3,07 sebagai bentuk normalisasi exposure yang tersedia, dan 75 insiden terkonfirmasi sebagai kegagalan Main Engine. Model BN hierarkhis empat lapis menggunakan Noisy-OR — dipilih untuk mereduksi ekspansi kombinatorik CPT dari 16 nilai menjadi 5 parameter terkalibrasi — mengidentifikasi root cause dominan: spare part tidak standar (RC01 = 63,8%), FW Cooling tidak terjaga (RC02 = 31,0%), ketidakpatuhan pre-heating (RC03 = 3,4%), dan keterbatasan kompetensi kru (RC05 = 1,7%), dengan gap maintenance (RC04) hadir sistemik pada 100% kasus. Validasi menghasilkan P(ME) = 0,548 terhadap empiris 75/154 = 0,487 (selisih 0,061), mengindikasikan kesesuaian awal yang memadai dengan data empiris; konsistensi internal dikonfirmasi melalui analisis sensitivitas per faktor root cause (sub-bab 4.8.1).Probabilitas accident meningkat dari P(Acc) = 0,198 (2023) menjadi 0,215 (2025) dengan potensi penurunan 78,6% apabila seluruh root cause dikelola optimal. Penelitian ini menyimpulkan terdapat indikasi tren awal deteriorasi risiko yang progresif dan sistemik pada periode pengamatan. Model Bayesian Network yang dikembangkan efektif sebagai alat prediksi risiko berbasis data untuk mendukung pengambilan keputusan keselamatan yang lebih proaktif dan terukur.
================================================================================================================================
This study examines incident and accident trends in the tanker machinery sector of PT. Pertamina International Shipping (PIS) from 2023 to 2025 — framed as an initial trend assessment based on three years of observation, with the understanding that this period is not yet sufficient to conclusively establish long-term patterns. The data includes 154 incidents (35/2023; 73/2024; 46/2025) with incident-per-vessel rates of 1.13 / 4.86 / 3.07 as the available form of exposure normalization, and 75 incidents confirmed as Main Engine failures. A four-layer hierarchical BN model using Noisy-OR — selected to reduce CPT combinatorial expansion from 16 values to 5 calibratable parameters — identified the dominant root causes: non-standard spare parts (RC01 = 63.8%), poor FW Cooling quality (RC02 = 31.0%), pre-heating non-compliance (RC03 = 3.4%), and limited crew competency (RC05 = 1.7%), with maintenance gaps (RC04) present in 100% of cases as a baseline aggravating factor. Validation yielded P(ME) = 0.548 against empirical 75/154 = 0.487 (gap of 0.061), indicating adequate initial agreement with empirical data; internal consistency was further confirmed through per- factor sensitivity analysis (sub-section 4.8.1). The predicted accident probability increased from P(Acc) = 0.198 (2023) to 0.215 (2025) with a potential risk reduction of 78.6% if all root causes are optimally managed This study concludes that there are initial indications of progressive and systemic risk deterioration within the observation period. The developed Bayesian Network model is effective as a data-driven risk prediction tool to support more proactive and measured safety decision-making.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Accident, Bayesian Network, Incident, Kapal Tanker, Longitudinal, Noisy OR Accident, Bayesian Network, Incident, Tanker, Longitudinal, Noisy-OR
Subjects:	T Technology > T Technology (General)
Divisions:	Faculty of Marine Technology (MARTECH) > Naval Architecture and Shipbuilding Engineering > 36101-(S2) Master Thesis
Depositing User:	Wulan Prasetya
Date Deposited:	01 Aug 2026 07:41
Last Modified:	01 Aug 2026 07:41
URI:	http://repository.its.ac.id/id/eprint/142919

Actions (login required)

View Item