Model Estimasi Probabilitas Kebocoran Pada Pipa Distribusi Gas Menggunakan Regresi Logistik

Pamungkas, Digdho Andika (2026) Model Estimasi Probabilitas Kebocoran Pada Pipa Distribusi Gas Menggunakan Regresi Logistik. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5010221142-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Kebocoran pipa distribusi gas akibat gangguan pihak ketiga merupakan salah satu penyebab kegagalan yang paling dominan sekaligus paling sulit dievaluasi, terutama pada wilayah dengan tingkat urbanisasi tinggi seperti Kota Surabaya. Penelitian ini bertujuan mengestimasi probabilitas kebocoran pada setiap segmen pipa distribusi gas berbahan Polyethylene di Area Surabaya berdasarkan atribut pipa dan tingkat gangguan pihak ketiga menggunakan metode regresi logistik. Data yang digunakan merupakan data historis redspot periode Januari 2025 hingga Januari 2026 sebanyak 204 segmen, yang terdiri dari 92 segmen bocor (45,1%) dan 112 segmen ekspose (54,9%). Variabel independen yang digunakan meliputi diameter, panjang segmen, usia pipa, dan activity level, dengan variabel dependen berupa status kebocoran biner. Hasil analisis deskriptif menunjukkan bahwa panjang segmen dan intensitas gangguan pihak ketiga merupakan faktor yang paling membedakan kondisi kebocoran, sedangkan diameter dan usia pipa berpengaruh secara tidak langsung. Uji signifikansi menunjukkan model signifikan secara keseluruhan (Uji G, G = 34,168; p < 0,001) dan seluruh variabel independen berpengaruh signifikan secara parsial (Uji Wald, p < 0,05). Model menghasilkan accuracy sebesar 85,37% dan rata-rata AUC 5-fold cross validation sebesar 0,8169 yang tergolong kategori considerable, serta terbukti tidak mengalami overfitting. Penerapan model pada 18 segmen redspot tahun 2026 menghasilkan estimasi probabilitas kebocoran antara 0,2042 hingga 0,6922, yang selanjutnya digabungkan dengan jumlah Sambungan Rumah Tangga (SRT) per kecamatan menjadi Pipeline Evaluation Index (PEI) melalui pembobotan berbasis expert adjustment, sebagai gambaran yang lebih menyeluruh mengenai tingkat kerentanan segmen pipa. Hasil penelitian ini diharapkan dapat menjadi informasi pendukung bagi pengelolaan integritas jaringan pipa distribusi gas secara lebih proaktif.
=================================================================================================================================================
Leakage of gas distribution pipelines caused by third-party activities is one of the most dominant yet most difficult failure causes to evaluate, particularly in highly urbanized areas such as the city of Surabaya. This study aims to estimate the probability of leakage for each Polyethylene gas distribution pipeline segment in the Surabaya Area based on pipe attributes and the level of third-party activity using the logistic regression method. The data used are historical redspot records from January 2025 to January 2026, comprising 204 segments, consisting of 92 leaking segments (45.1%) and 112 exposed segments (54.9%). The independent variables include diameter, segment length, pipe age, and activity level, with a binary leakage status as the dependent variable. Descriptive analysis shows that segment length and the intensity of third-party activity are the most distinguishing factors of leakage conditions, whereas diameter and pipe age exert an indirect influence. Significance testing indicates that the model is significant overall (G-test, G = 34.168; p < 0.001) and that all independent variables are partially significant (Wald test, p < 0.05). The model achieves an accuracy of 85.37% and an average 5-fold cross-validation AUC of 0.8169, which falls into the considerable category, and is shown to be free from overfitting. Applying the model to 18 redspot segments in 2026 produces estimated leakage probabilities ranging from 0.2042 to 0.6922, which are then combined with the number of Household Connections (SRT) per district into a Pipeline Evaluation Index (PEI) through expert-adjustment-based weighting, providing a more comprehensive representation of pipe segment vulnerability. The results of this study are expected to serve as supporting information for a more proactive integrity management of gas distribution pipeline networks.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Pipa Distribusi Gas, Regresi Logistik, Gangguan pihak ketiga, Probabilitas Kebocoran, Gas Distribution Pipelines, Logistic Regression, Third-Party Interference, Leakage Probability
Subjects:	T Technology > TS Manufactures T Technology > TS Manufactures > TS174 Maintainability (Engineering) . Reliability (Engineering)
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Industrial Engineering > 26201-(S1) Undergraduate Thesis
Depositing User:	Digdho Andika Pamungkas
Date Deposited:	03 Aug 2026 02:13
Last Modified:	03 Aug 2026 02:13
URI:	http://repository.its.ac.id/id/eprint/141859

Actions (login required)

View Item