Primantara, Ari (2025) Pengembangan Model Integrasi Metode Six Sigma dan Data Analytics berbasis Machine Learning untuk Peningkatan Efisiensi Pengantongan di Industri Pupuk. Doctoral thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 7032222003-Doctoral.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
7032222003-Doctoral.pdf - Accepted Version Restricted to Repository staff only Download (31MB) | Request a copy |
Abstract
Peningkatan persaingan industri pupuk urea di dunia membuat harga jual pupuk urea non subsidi semakin menurun. Karena itu peningkatan efisiensi diperlukan pada proses produksi pupuk urea non subsidi di industri pupuk untuk bisa menurunkan biaya produksi. Salah satu potensi efisiensi didapatkan dari penurunan jumlah losses produk pada area pengantongan. Penelitian ini mengusulkan model integrasi Six Sigma dan Data Analytics berbasis Machine Learning pada proses pengantongan di PT Petrokimia Gresik. Metode Six Sigma digunakan pada tahap Define dan Measure untuk mengukur variabilitas berat pada sistem pengantongan, perhitungan kapabilitas proses dan nilai sigma. Tahap selanjutnya adalah analyze yang mencakup descriptive dan predictive analytics, yang melibatkan penerapan algoritma Support Vector Regresssion (SVR), Artificial Neural Network (ANN), Linear Regression (LR), dan Random Forest Regressor (RFR). Hasil analisis menunjukkan RFR sebagai model paling akurat dengan R² 0,9638, RMSE 0,0496, MAE 0,0388, dan MAPE 0,0978. Model lain, ANN dengan R² 0,9277, RMSE 0,0727, MAE 0,0561, dan MAPE 6,4901; SVR dengan R² 0,9171, RMSE 0,0765, MAE 0,0454, dan MAPE 0,1738; sedangkan LR tercatat dengan R² 0,9110, RMSE 0,0762, MAE 0,0522, dan MAPE 0,2359. Selanjutnya yaitu improve sebagai persiapan untuk tahap control. Pada tahap control, penelitian ini mengembangkan Smart Bagging System (SBS), yaitu sistem kontrol pengantongan berbasis teknologi digital yang mengintegrasikan IoT, sensor real-time, dan algoritma RFR untuk mengatur dua variabel CTQ yaitu waktu buka tutup gate valve dan tekanan angin mesin pengantongan. SBS menjaga berat pupuk di rentang 50,2 – 50,4 kg. Hasil implementasi SBS meningkatkan sigma level dari 1,03 dengan defect rate 68,15% menjadi 3,48 dengan defect rate 2,37%. SBS juga menurunkan nilai DPMO dari 681,523 menjadi 23,730. Dengan kapasitas produksi hingga 3.000 ton/hari (±60.000 bag/hari), analisis ekonomi menunjukkan SBS sangat layak, dengan net present value (NPV) positif, internal rate of return (IRR) 230%, dan benefit cost ratio (BCR) 12,02 dan payback period tercapai dalam 32 hari. Penelitian peningkatan efisiensi pengantongan dengan model integrasi Six Sigma dan Data Analytics berbasis Machine Learning bertujuan untuk memperoleh berat pupuk sesuai standar pada range 50,2 – 50,4 kg.
====================================================================================================================================
The increasing competition in the global urea fertilizer industry has led to a decrease in the selling price of non-subsidized urea fertilizer. Therefore, increasing
efficiency in the production process of non-subsidized urea fertilizer is necessary in the fertilizer industry to reduce production costs. One of the potential efficiencies
is obtained from reducing the number of product losses in the bagging area. This study proposes an integration model of Six Sigma and Data Analytics based on Machine Learning in the bagging process at PT Petrokimia Gresik. The Six Sigma method is used in the Define and Measure stages to measure weight variability in the bagging system, process capability calculations and sigma values. The next stage involves analyzing data, which includes descriptive and predictive analytics. This analysis utilizes Support Vector Regression (SVR), Artificial Neural Network (ANN), Linear Regression (LR), and Random Forest Regressor (RFR) algorithms.
The results of the analysis show RFR as the most accurate model with R² 0.9638, RMSE 0.0496, MAE 0.0388, and MAPE 0.0978. Other models, ANN with R² 0.9277, RMSE 0.0727, MAE 0.0561, and MAPE 6.4901; SVR with R² 0.9171,
RMSE 0.0765, MAE 0.0454, and MAPE 0.1738; while LR was recorded with R² 0.9110, RMSE 0.0762, MAE 0.0522, and MAPE 0.2359. The next step is to improve as preparation for the control stage. At the control stage, this study
developed the Smart Bagging System (SBS), a digital technology-based bagging control system that integrates IoT, real-time sensors, and the RFR algorithm to regulate two CTQ variables, namely the gate valve opening and closing time and the bagging machine air pressure. SBS maintains the fertilizer weight in the range of 50.2 - 50.4 kg. The results of SBS implementation increased the sigma level from
1.03 with a defect rate of 68.15% to 3.48 with a defect rate of 2.37%. SBS also reduced the DPMO value from 681.523 to 23.730. With a production capacity of up to 3,000 tons/day (±60,000 bags/day), economic analysis shows that SBS is very
feasible, with a positive net present value (NPV), an internal rate of return (IRR) of 230%, and a benefit cost ratio (BCR) of 12.02 and a payback period achieved in 32
days. Research on increasing bagging efficiency with the Six Sigma integration model and Machine Learning-based Data Analytics aims to obtain fertilizer weight
according to standards in the range of 50.2 - 50.4 kg.
Actions (login required)
 |
View Item |