Sistem Deteksi Kebocoran Pipa Hidrogen Sulfida dengan Metode Artificial Neural Network pada Fasilitas Central Processing Plant Perusahaan Kilang Minyak

Al Khothib, Abdurrohman (2025) Sistem Deteksi Kebocoran Pipa Hidrogen Sulfida dengan Metode Artificial Neural Network pada Fasilitas Central Processing Plant Perusahaan Kilang Minyak. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040211027-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

Kebocoran pada jaringan perpipaan gas masih menjadi permasalahan di Sistem Proses Central Processing Plant Blok A Aceh karena dapat menurunkan efisiensi distribusi, mengancam keselamatan pekerja, dan mencemari lingkungan. Metode deteksi konvensional, termasuk sistem berbasis serat optik, sering kali terlalu sensitif terhadap gangguan eksternal, sehingga memicu false alarm yang menurunkan kepercayaan operator. Dalam kondisi nyata, kebocoran tidak selalu dapat terdeteksi secara langsung oleh sistem. Oleh karena itu, biasanya kebocoran dapat diketahui oleh sistem monitoring dan patroli rutin yang memantau sistem proses, serta sensor gas ditempatkan di permukaan dekat dengan pemukiman warga untuk memberikan peringatan tambahan. Penelitian ini mengusulkan sistem deteksi kebocoran berbasis Artificial Neural Network (ANN) dengan memanfaatkan selisih tekanan dan laju aliran antara sisi hulu dan hilir pipa gas hidrogen sulfida di Alur Siwah Central Processing Plant, Blok A, PT XYZ. Dataset lapangan sebanyak 23.146 baris dibersihkan menggunakan teknik listwise deletion dan standardization, menghasilkan 16.369 sampel terlabel (bocor dan normal). Model terbaik terdiri atas dua hidden layer dengan jumlah neuron masing-masing 128 dan 32, fungsi aktivasi ReLU, optimizer Adam, batch size 64, dan dropout 0,3. Dengan skema pembagian data 70% untuk pelatihan dan 30% untuk pengujian, model mencapai akurasi 99,31%, F1-score 0,99, dan MAE 0,0124 pada 4.911 sampel uji. Model diintegrasikan ke dalam antarmuka graphical user interface (GUI) yang menampilkan data secara real time, histori kebocoran, serta alarm audio. Hasil tersebut menunjukkan bahwa pendekatan Artificial Neural Network mampu mendeteksi kebocoran pipa gas dengan tingkat akurasi tinggi.
=================================================================================================================================
Gas pipeline leakage remains a critical issue in the Central Processing Plant (CPP) system at Block A, Aceh, as it can reduce distribution efficiency, pose risks to worker safety, and harm the environment. Conventional detection methods, including fiber-optic-based systems, are often overly sensitive to external disturbances, leading to frequent false alarms that undermine operator confidence. In real-world conditions, leaks are not always detected directly by the system. Therefore, leak detection typically relies on routine process monitoring, patrol inspections, and surface-mounted gas sensors placed near residential areas for additional warnings. This study proposes a leak detection system based on Artificial Neural Network (ANN) by utilizing the pressure and flow rate differences between the upstream and downstream sections of the hydrogen sulfide (H₂S) gas pipeline at the Alur Siwah Central Processing Plant, Block A, operated by PT XYZ. A field dataset comprising 23,146 rows was cleaned using listwise deletion and standardization techniques, resulting in 16,369 labeled samples (leak and normal conditions). The best-performing model consists of two hidden layers with 128 and 32 neurons respectively, using ReLU activation functions, Adam optimizer, a batch size of 64, and a dropout rate of 0.3. With a data split of 70% for training and 30% for testing, the model achieved an accuracy of 99.31%, an F1-score of 0.99, and a mean absolute error (MAE) of 0.0124 on 4,911 test samples. The model was integrated into a graphical user interface (GUI) that displays real-time data, leak history, and audio alarms. These results demonstrate that the Artificial Neural Network approach is capable of detecting gas pipeline leaks with a high level of accuracy.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Deteksi Kebocoran, Artificial Neural Network, Hidrogen Sulfida, Leak Detection, Artificial Neural Network, Hydrogen Sulfide
Subjects:	Q Science T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing.
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Abdurrohman Al-khothib
Date Deposited:	08 Aug 2025 07:14
Last Modified:	08 Aug 2025 07:14
URI:	http://repository.its.ac.id/id/eprint/127992

Actions (login required)

View Item