Nu'ila, Anis Umi (2026) Pemodelan Reliability Hot Well Pump berbasis Data Proses dengan Deteksi Anomali Mahalanobis Distance pada Pembangkit Listrik Tenaga Panas Bumi. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5009221053-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5009221053-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (4MB) | Request a copy |
Abstract
Hot well pump merupakan peralatan kritis pada Pembangkit Listrik Tenaga Panas Bumi (PLTP) yang berfungsi mengalirkan kondensat dari kondensor menuju cooling tower. Penurunan performa atau kegagalan pada hot well pump dapat memengaruhi kontinuitas operasi pembangkit dan menurunkan efisiensi sistem secara keseluruhan. Metode reliability umumnya berbasis data historis kegagalan memiliki keterbatasan dalam menggambarkan kondisi aktual peralatan yang berubah secara dinamis selama beroperasi. Penelitian ini mengembangkan metode pemantauan kondisi dan pemodelan reliability berbasis data proses untuk menghasilkan informasi reliability yang lebih representatif terhadap kondisi aktual hot well pump. Data proses yang digunakan meliputi differential pressure discharge-suction, flow rate, temperatur bearing drive end (DE), temperatur bearing non-drive end (NDE), vibrasi pompa, arus motor, dan tegangan suplai. Tahapan penilitian meliputi pengolahan data awal awal, deteksi anomali menggunakan mahalanobis distance, analisis distribusi, perhitungan reliability statik dan dinamik, serta integrasi hasil ke dalam MATLAB GUI. Metode mahalanobis distance digunakan untuk mendeteksi adanya anomali berdasarkan penyimpangan pola hubungan antar variabel dengan mempertimbangkan kovarians data. ata non anomali digunakan dalam analisis distribusi dan penentuan persamaan reliability statik. Hasil penelitian menunjukkan bahwa mahalanobis distance mampu mendeteksi anomali pada data proses hot well pump. Seleksi variabel meningkatkan nilai covariance matrix menjadi di atas 0,9 dan menghasilkan threshold yang lebih representatif sebesar 7,779 dibandingkan 12,017 pada model 1. Nilai mahalanobis distance pada kedua model mengikuti distribusi lognormal sehingga digunakan dalam pemodelan reliability. Model reliability model 2 memberikan hasil yang paling mendekati kondisi aktual, yaitu 259 hari normal dan 35 hari trip dibandingkan kondisi aktual 261 hari normal dan 33 hari trip. Integrasi ke dalam MATLAB GUI mendukung visualisasi hasil deteksi anomali dan nilai reliability.
================================================================================================================================
Hot well pump is a critical piece of equipment in a geothermal power plant (GPP), serving to transport condensate from the condenser to the cooling tower. A decline in performance or failure of the hot well pump can affect the continuity of the plant’s operations and reduce the overall efficiency of the system. Reliability methods, which are generally based on historical failure data, have limitations in describing the actual condition of equipment, which changes dynamically during operation. This research develops a condition monitoring and reliability modelling method based on process data to generate reliability information that is more representative of the actual condition of the hot well pump. The process data used includes discharge-suction differential pressure, flow rate, drive-end (DE) bearing temperature, non-drive-end (NDE) bearing temperature, pump vibration, motor current, and supply voltage. The research stages include initial data processing, anomaly detection using Mahalanobis distance, distribution analysis, calculation of static and dynamic reliability, and integration of the results into a MATLAB GUI. The Mahalanobis distance method is used to detect anomalies based on deviations in the relationship patterns between variables, taking data covariance into account. Non-anomalous data are used in distribution analysis and the determination of static reliability equations. The research results show that the Mahalanobis distance is capable of detecting anomalies in hot well pump process data. Variable selection increased the covariance matrix value to above 0.9 and produced a more representative threshold of 7.779, compared to 12.017 in Variation 1. The Mahalanobis distance values in both variations follow a log-normal distribution and were therefore used in reliability modelling. The reliability model for Variation 2 provided results that most closely matched the actual conditions, namely 259 normal days and 35 trip days, compared with the actual conditions of 261 normal days and 33 trip days. Integration into the MATLAB GUI supports the visualisation of anomaly detection results and reliability values.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Condition Monitoring, Hot Well Pump, Mahalanobis Distance, MATLAB GUI, Reliability Condition Monitoring, Hot Well Pump, Mahalanobis Distance, MATLAB GUI, Reliability Condition Monitoring, Hot Well Pump, Mahalanobis Distance, MATLAB GUI, Reliability Condition Monitoring, Hot Well Pump, Mahalanobis Distance, MATLAB GUI, Reliability |
| Subjects: | T Technology > T Technology (General) T Technology > TA Engineering (General). Civil engineering (General) T Technology > TA Engineering (General). Civil engineering (General) > TA169 Reliability (Engineering) |
| Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis |
| Depositing User: | Anis Umi Nu'ila |
| Date Deposited: | 31 Jul 2026 03:25 |
| Last Modified: | 31 Jul 2026 03:25 |
| URI: | http://repository.its.ac.id/id/eprint/140434 |
Actions (login required)
 |
View Item |