Pemodelan Regime Pembebanan Pembangkit Secara Harian Menggunakan Metode Markov Switching Autoregressive : Studi Kasus Di PLTMG Arun

Prasetya, Andyk Probo (2024) Pemodelan Regime Pembebanan Pembangkit Secara Harian Menggunakan Metode Markov Switching Autoregressive : Studi Kasus Di PLTMG Arun. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6032221028-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (2MB) | Request a copy

Abstract

Kebutuhan energi listrik di Indonesia terus tumbuh setiap tahunnya, diperkirakan penjualan listrik meningkat sekitar 4,9% dan jumlah pelanggan mencapai 24,4 juta menurut RUPTL 2021 – 2030. PLTMG Arun di Aceh memainkan peran krusial dalam Sub Sistem Aceh (Sumbagut - Aceh), dengan kapasitas saat ini mencapai 184 MW, menyumbang 9,41% dari kebutuhan beban Sub Sistem Sumbagut sebesar 1.956 MW. Penelitian ini bertujuan untuk menyelidiki dan mendeteksi perubahan dalam Regime pembebanan pembangkit listrik secara harian, bulanan, dan tahunan menggunakan Model Markov Switching Autoregressive, dengan fokus pada PLTMG Arun. Perubahan dalam Regime pembebanan pembangkit dapat memberikan informasi berharga kepada pengelola pembangkit dan dispatcher untuk membuat keputusan yang lebih baik dan efisien. Penelitian ini mendalami bagaimana Model Markov Switching Autoregressive dapat mengidentifikasi perubahan Regime yang mempengaruhi pola operasional pembangkit, memberikan pemahaman yang lebih baik tentang dinamika pembebanan, serta menentukan kriteria efektif untuk memilih model yang paling sesuai. Model MS(4)AR(1) dipilih sebagai model terbaik dengan Akaike Information Criterion (AIC) terendah 15.688,27 menunjukkan efisiensi dalam menjelaskan variasi data tanpa kompleksitas berlebih. Forecast kebutuhan beban pembangkit selama 30 hari menunjukkan fluktuasi signifikan, dengan beban tertinggi pada hari ke-2 (167,48 MW) dan terendah pada hari ke-18 (9,29 MW). Total selisih antara nilai forecast (1.835,54 MW) dan realisasi (1.828,70 MW) hanya sebesar 6,84 MW, dengan MAPE rata-rata 18,63%, menunjukkan akurasi prediksi yang dapat diterima. Dengan memahami durasi dan karakteristik setiap regime, perusahaan dapat meningkatkan efisiensi sumber daya dan keberlanjutan energi, meminimalkan risiko operasional di masa depan, berkontribusi pada pengembangan metode analisis pembebanan dan pengambilan keputusan yang lebih efisien oleh pengelola pembangkit.
===================================================================================================================================
The electricity demand in Indonesia continues to grow annually, with electricity sales expected to increase by approximately 4.9% and the number of customers reaching 24.4 million, according to the 2021-2030 RUPTL. The Arun Gas Power Plant (PLTMG) in Aceh plays a crucial role in the Aceh Sub-System (Sumatra North - Aceh), with its current capacity of 184 MW contributing 9.41% to the Sub-System's total load of 1,956 MW. This study aims to investigate and detect changes in the load regime of power generation on a daily, monthly, and yearly basis using the Markov Switching Autoregressive (MSAR) Model, with a focus on the Arun Power Plant. Changes in the load regime can provide valuable information for plant operators and dispatchers to make more informed and efficient decisions. This research delves into how the Markov Switching Autoregressive Model can identify regime shifts that affect power plant operation patterns, offering a better understanding of load dynamics, and determining effective criteria for selecting the most appropriate model. We selected the MS(4)AR(1) model as the best model, demonstrating its efficiency in explaining data variations without excessive complexity, with the lowest Akaike Information Criterion (AIC) of 15,688.27. Forecasts of the power plant's load demand over a 30-day period showed significant fluctuations, with the highest load on the second day (167.48 MW) and the lowest on the eighteenth day (9.29 MW). The total difference between the forecasted value (1,835.54 MW) and the actual realization (1,828.70 MW) was only 6.84 MW, with a mean absolute percentage error (MAPE) of 18.63%, indicating an acceptable level of prediction accuracy. By understanding the duration and characteristics of each regime, companies can enhance resource efficiency, improve energy sustainability, minimize operational risks in the future, and contribute to the development of load analysis methods and decision-making processes that support more efficient power plant management.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Markov Switching Autoregressive, Time Series, Load Profile, Dispatcher.
Subjects:	Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry)
Divisions:	Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT)
Depositing User:	Andyk Probo Prasetya
Date Deposited:	14 Jan 2025 03:56
Last Modified:	14 Jan 2025 03:56
URI:	http://repository.its.ac.id/id/eprint/116115

Actions (login required)

View Item