Model Prediksi Unplanned Well Down Pada Operasi Heavy Oil Untuk Pengambilan Keputusan Operasional

Saputra, Aryo (2025) Model Prediksi Unplanned Well Down Pada Operasi Heavy Oil Untuk Pengambilan Keputusan Operasional. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6032241097-Master_Thesis.pdf Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Penurunan produksi Heavy Oil (HO) di PT XYZ dipengaruhi oleh sumur mati (well down) yang bersifat tidak terencana (unplanned), yang berpotensi menimbulkan Loss Production Opportunity (LPO) dan berdampak signifikan pada target produksi. Penelitian ini bertujuan mengembangkan dan membandingkan dua metode peramalan deret waktu yaitu ARIMA dan Holt-Winters Exponential Smoothing (HWES). Kedua metode tersebut digunakan untuk memprediksi jumlah kejadian well down yang disebabkan oleh kegagalan Mechanical Pumping Unit (MPU) di masa mendatang. Pengujian akurasi menggunakan metrik Mean Absolute Percentage Error (MAPE) menunjukkan bahwa Model SARIMA (2,2,1) adalah model terbaik dan paling akurat dengan nilai MAPE 4.5619%, jauh melampaui kinerja model HWES (MAPE tertinggi 27.3697%). Model terpilih ini memprediksi total kerugian finansial yang mungkin terjadi (Base Case) sebesar Rp 13,35 Miliar dalam satu tahun, dengan potensi kerugian maksimum (Worst Case) mencapai Rp 41,80 Miliar. Hasil peramalan ini menjadi landasan untuk tiga rekomendasi strategis utama yaitu pengendalian risiko finansial berbasis batas atas 95% untuk membenarkan anggaran upgrade, perencanaan target produksi realistis yang disesuaikan dengan LPO prediksi, integrasi KPI dengan LPO rata-rata historis sebagai ambang batas peringatan dini (early warning), dan juga sebagai justifikasi investasi operasional.
==================================================================================================================================
The decline in Heavy Oil (HO) production at PT XYZ is influenced by unplanned well down events, which potentially create a Loss Production Opportunity (LPO) and significantly impact production targets. This research aims to develop and compare two time series forecasting methods, ARIMA and Holt-Winters Exponential Smoothing (HWES) to predict the number of well down events caused by Mechanical Pumping Unit (MPU) failures in the future. Accuracy testing using the Mean Absolute Percentage Error (MAPE) metric showed that the SARIMA (2,2,1) Model is the best and most accurate model with a MAPE value of 4.5619%, far surpassing the performance of the HWES models (highest MAPE 27.3697%). This selected model predicts the total potential financial loss in the Base Case scenario to be Rp 13.35 billion over one year, with a maximum potential loss in the Worst Case reaching Rp 41.80 billion. The forecasting results serve as the foundation for three key strategic recommendations are financial risk control based on the Upper Bound 95% to justify the MPU upgrade budget, realistic production target planning adjusted according to the predicted LPO, KPI integration using the historical average LPO as the early warning threshold.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Heavy Oil, Unplanned, Well down, ARIMA, Holt-Winters Exponential Smoothing
Subjects:	T Technology > TJ Mechanical engineering and machinery > TJ174 Maintenance and repair of machinery
Divisions:	Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT)
Depositing User:	Aryo Saputra
Date Deposited:	22 Jan 2026 03:01
Last Modified:	22 Jan 2026 03:01
URI:	http://repository.its.ac.id/id/eprint/130051

Actions (login required)

View Item