Perancangan Sistem Deteksi Kesalahan Pada Pompa Torak Glycol Di Offshore Sipl Gresik Berbasis Ekstraksi Ciri Tekanan Dan Laju Aliran.

Arfianti, Rida Ayu (2022) Perancangan Sistem Deteksi Kesalahan Pada Pompa Torak Glycol Di Offshore Sipl Gresik Berbasis Ekstraksi Ciri Tekanan Dan Laju Aliran. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 02311840000122-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (3MB)

Abstract

Kesalahan sering terjadi pada pompa torak dikarenakan pengaruh lingkungan yang keras seperti pada offshore. Hal tersebut dapat menyebabkan kerusakan serta kebocoran pada valve dan kesalahan bearing pada pompa yang dapat memicu kegagalan fungsi operasi serta diperlukannya cost yang tidak sedikit untuk melakukan maintenance. Penelitian kali ini dilakukan perancangan sistem deteksi kesalahan dengan metode ekstraksi ciri guna mendeteksi kesalahan seal leak dan bearing fault yang terjadi di offshore SIPL Gresik. Tahapan yang dilakukan dalam perancangan sistem deteksi kali ini ada empat tahapan yang dimulai dari pemodelan pompa torak glycol. Hasil validasi data perusahaan menggunakan metode MAPE diperoleh nilai error rata-rata 6.58%. Selanjutnya, dilakukan simulasi untuk empat skenario yaitu: healthy, leakage, bearing, dan leakage & bearing. Dari hasil simulasi diperoleh data sinyal domain waktu yang diolah kedalam domain frekuensi untuk diambil power spectrum menggunakan fast fourier transform (FFT). Pada setiap skenario, sinyal domain waktu maupun frekuensi dilakukan ekstraksi ciri sebanyak 14 parameter dengan teknik statistik. Data tersebut selanjutnya digunakan sebagai masukan classifier support vector machines (SVM) dan diperoleh nilai akurasi 98.21%.
===================================================================================================================================
Faults often occur in reciprocating pumps due to harsh environmental influences such as offshore. This can cause damage and leaks in the pump valve and bearing faults which can trigger operating malfunctions and require significant costs for maintenance. In this research, a fault detection system designed uses a feature extraction method to detect faults seal leaks and bearing that occurred offshore of SIPL Gresik. There are four steps in the design of the detection system, starting from the glycol reciprocating pump modeling. The results of company data validation using the MAPE method obtained an average error value of 6.58%. Furthermore, simulations are carried out for four scenarios are: healthy, leakage, bearing, and leakage & bearing. From the simulation results obtained time domain signal data that is processed into the frequency domain to take the power spectrum using fast Fourier transform (FFT). In each scenario, both time and frequency domain signals were extracted with 14 parameters using statistical techniques. The data is then used as input for classifier support vector machines (SVM) and obtained an accuracy value of 98.21%.

Item Type:	Thesis (Other)
Additional Information:	RSF 621.381 536 Arf p-1 2022
Uncontrolled Keywords:	Deteksi Kesalahan, Laju Aliran, Pompa Torak, Tekanan. Fault Detection, Flow, Pressure, Reciprocating Pump.
Subjects:	T Technology > TJ Mechanical engineering and machinery > TJ903 Pumping machinery.
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Mr. Marsudiyana -
Date Deposited:	12 May 2026 01:39
Last Modified:	12 May 2026 01:39
URI:	http://repository.its.ac.id/id/eprint/133142

Actions (login required)

View Item