Optimisasi Konsumsi Energi Pada Electrical Submersible Pump (ESP) Menggunakan Algoritma Particle Swarm Optimization (Studi Kasus: Dataset Sumur X Lapangan Y)

Fiddiin, Muhammad Rasyid (2025) Optimisasi Konsumsi Energi Pada Electrical Submersible Pump (ESP) Menggunakan Algoritma Particle Swarm Optimization (Studi Kasus: Dataset Sumur X Lapangan Y). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5009211137-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Electrical Submersible Pump (ESP) merupakan metode pengangkatan buatan banyak digunakan kedua setelah pompa angguk, tetapi memiliki konsumsi listrik yang tinggi. Padalapangan minyak & gas offshore, ESP tercatat dapat menyumbang hingga 67,3% dari total konsumsi listrik pada fasilitas produksi. Optimisasi sistem ESP diperlukan guna menurunkan konsumsi energinya yang tinggi. Konsumsi energi dimodelkan menggunakan Artificial Neural Network (ANN), dengan target berupa konsumsi energi, yaitu rasio antara konsumsi energi harian (kWh) terhadap produksi fluida harian (bbl). Data terlebih dahulu dibersihkan menggunakan Principal Component Analysis (PCA) untuk mendeteksi dan mengeliminasi outlier, kemudian dianalisis dengan Partial Least Squares (PLS) guna mengevaluasi relevansi variabel terhadap output. Dari total 13 variabel, sebanyak 11 variabel dinyatakan relevan dan digunakan sebagai input pada model ANN. Arsitektur model terdiri atas satu input layer(11node), satu hidden layer (16 node), dan satu output layer (1 node). Data dibagi ke dalam tiga subset, yaitu 70% untuk pelatihan (training), 20% untuk validasi, dan 10% untuk pengujian (testing). Model ANN menunjukkan performa baik dengan nilai RMSE, MAE, dan R² masingmasing sebesar 0,3637; 0,1932; dan 0,9424 pada data validasi, serta 0,1695; 0,0969; dan 0,9826 pada data uji. Hasil optimisasi menunjukkan bahwa kombinasi parameter operasi optimum, yaitu frekuensi sebesar 44,9 Hz dan diameter choke sebesar 128 (1/64"), mampu menurunkan konsumsi energi sebesar 12,79% dibandingkan dengan kondisi awal.
======================================================================================================================================
The Electrical Submersible Pump (ESP) is a widely used artificial lift method second only to the sucker rod pump but has high electricity consumption. In offshore oil & gas fields, ESP is recorded to contribute up to 67.3% of the total electricity consumption in production facilities. To reduce energy consumption, optimization is necessary. Energy consumption was modeled using an Artificial Neural Network (ANN), with the target being scaled energy consumption, which is the ratio between daily energy consumption (kWh) to daily fluid production (bbl). The data was first cleaned using Principal Component Analysis (PCA) to detect and eliminate outliers and then analyzed with Partial Least Squares (PLS) to evaluate the relevance of variables to the output. From a total of 13 variables, 11 variables were declared relevant and used as inputs to the ANN model. The model architecture consists of one input layer (11 nodes), one hidden layer (16 nodes), and one output node. The model architecture consists of one input layer (11 nodes), one hidden layer (16 nodes), and one output layer (1 node). The data was divided into three subsets, 70% for training, 20% for validation, and 10% for testing. The ANN model showed good performance with RMSE, MAE, and R² values of 0.3637, 0.1932, and 0.9424 in the validation data, and 0.1695, 0.0969, and 0.9826 in the test data, respectively. The optimization results show that the optimum combination of operating parameters, namely a frequency of 44.9 Hz and a choke diameter of 128 (1/64"), can reduce energy consumption by 12.79% compared to the initial conditions.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Artificial Neural Network, Electrical Submersible Pump, Konsumsi Energi, Particle Swarm Optimization, Artificial Neural Network, Electrical Submersible Pump, Energy Consumption, Particle Swarm Optimization
Subjects:	Q Science > QA Mathematics > QA278.5 Principal components analysis. Factor analysis. Correspondence analysis (Statistics) Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Industrial Technology > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Rasyid Fiddiin
Date Deposited:	05 Aug 2025 07:09
Last Modified:	05 Aug 2025 07:09
URI:	http://repository.its.ac.id/id/eprint/127426

Actions (login required)

View Item