Analisis Scouring Pipa Bawah Laut Berbasis Pendekatan Machine Learning (Studi Kasus di Perairan Laut Jawa)

Laksono, Candra Rizky (2026) Analisis Scouring Pipa Bawah Laut Berbasis Pendekatan Machine Learning (Studi Kasus di Perairan Laut Jawa). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5046221018-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Kebutuhan gas bumi di Pulau Jawa terus meningkat seiring pertumbuhan industri, pembangkit listrik, dan kebutuhan domestik, sehingga mendorong pengembangan proyek gas lepas pantai terintegrasi di Perairan Laut Jawa yang melibatkan lapangan MDA dan MBH. Jalur pipa bawah laut tersebut berada pada kondisi batimetri, sedimen, arus musiman, dan gelombang yang bervariasi sehingga rentan mengalami scouring. Penelitian ini bertujuan menentukan hasil perhitungan analitik scouring pada jalur pipa MDA–MBH, mengembangkan model prediksi berbasis machine learning, serta mengevaluasi akurasinya dalam memprediksi lebar dan kedalaman scour terhadap data survei aktual. Penelitian menggunakan data sekunder berupa karakteristik pipa, batimetri, kecepatan arus, gelombang, dan parameter geoteknik dasar laut pada 20 titik pengamatan tahun 2020 dan 2022. Analisis dilakukan melalui perhitungan kondisi awal pergerakan sedimen, bilangan Keulegan–Carpenter, parameter Shields, serta persamaan empiris untuk memperoleh scour width dan scour depth. Model regresi linear kemudian dibangun menggunakan Orange Data Mining melalui tahap pembersihan data, pemilihan variabel, pelatihan, dan pengujian. Kinerja model dievaluasi menggunakan Root Mean Square Error dan Mean Absolute Percentage Error. Hasil menunjukkan bahwa prediksi machine learning mengikuti pola data aktual, meskipun masih terdapat penyimpangan pada beberapa titik. Pada tahun 2020, prediksi lebar scour menghasilkan RMSE 7,75 m dan MAPE 20,67%, sedangkan kedalaman scour menghasilkan RMSE 0,26 m dan MAPE 15,33%. Pada tahun 2022, model menghasilkan RMSE 4,65 m dan MAPE 11,44% untuk lebar, serta RMSE 0,15 m dan MAPE 10,80% untuk kedalaman. Dengan demikian, regresi linear dapat digunakan sebagai pendekatan tambahan berbasis data, tetapi pengembangan dataset dan variabel masih diperlukan untuk meningkatkan keandalan prediksi.
====================================================================================================================================
Natural gas demand on Java Island continues to increase with the growth of industry, power generation, and domestic consumption, encouraging an offshore gas project in the Java Sea involving the MDA and MBH fields. The subsea pipeline route is exposed to varying bathymetry, sediments, seasonal currents, and waves, making it susceptible to scouring. This study aims to determine analytical scour characteristics along the MDA–MBH pipeline, develop a machine-learning prediction model, and evaluate its accuracy in predicting scour width and depth against survey data. Secondary data included pipeline characteristics, bathymetry, current velocity, wave parameters, and seabed geotechnical properties at 20 observation points in 2020 and 2022. The analysis covered sediment mobility, the Keulegan–Carpenter number, the Shields parameter, and empirical estimation of scour width and depth. A linear regression model was developed in Orange Data Mining through data cleaning, variable selection, training, and testing. Model performance was evaluated using Root Mean Square Error and Mean Absolute Percentage Error. Results show that machine-learning predictions generally followed actual data patterns, although deviations remained at several points. In 2020, predicted scour width produced an RMSE of 7.75 m and MAPE of 20.67%, while scour depth produced an RMSE of 0.26 m and MAPE of 15.33%. In 2022, the model produced an RMSE of 4.65 m and MAPE of 11.44% for width, and an RMSE of 0.15 m and MAPE of 10.80% for depth. Therefore, linear regression can serve as an additional data-driven approach, although further dataset development is required to improve prediction reliability.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Laut Jawa, Machine Learning, Pipa Bawah Laut, Regresi Linear, Scouring ============================================================ Java Sea, Machine Learning, Subsea Pipeline, Linear Regression, Scouring
Subjects:	T Technology > T Technology (General) > T174 Technological forecasting T Technology > T Technology (General) > T385 Visualization--Technique T Technology > T Technology (General) > T57.5 Data Processing T Technology > TC Hydraulic engineering. Ocean engineering > TC147 Ocean wave power. T Technology > TC Hydraulic engineering. Ocean engineering > TC175.2 Sediment transport
Divisions:	Faculty of Marine Technology (MARTECH) > Offshore Engineering > 54040-(S1) Undergraduate Thesis
Depositing User:	Candra Rizky Laksono
Date Deposited:	30 Jul 2026 07:15
Last Modified:	30 Jul 2026 07:15
URI:	http://repository.its.ac.id/id/eprint/139859

Actions (login required)

View Item