STRATEGY TO INCREASE PT FREEPORT INDONESIA SLURRY LINE AVAILABILITY AND LIFETIME USING MACHINE LEARNING

• Main Author: Rizki Febrianto, Mohammad
• Format: Theses
• Language: Indonesia
• Subjects: Manajemen umum
• Online Access: https://digilib.itb.ac.id/gdl/view/80956
• Institution: Institut Teknologi Bandung

The copper concentrate slurry lines availability is important in maintaining sustainable process production for PT Freeport Indonesia. Yet the leaking trendline of the slurry pipe due to the thin pipe wall from 2019 to 2023 is increasing, indicating that the availability of the copper concentrate slurry line become lower. Furthermore, segments between Mile Post 34 – Mile Post 6, typically projected to have a ten-year service life, are experiencing accelerated wear, leading to discrepancies between expected and actual lifespan. Such deviations from anticipated durability metrics significantly impact the slurry transport system's efficiency, underscoring the need for refined predictive maintenance strategies. This research aims to predict slurry line thickness using a machine learning model and use the model to optimize pumping operational parameters. The independent variables considered include slurry flow, density, slurry running hours, water running hours, water flow, and particle size (150 #). In this study, a dataset comprising 58 samples was divided, with 70% allocated for training and 30% for testing. The predictive models that are being used are Decision Tree Regression, Neural Network, and Support Vector Regression. The best model produced 0.114 MSE and 4.3% MAPE with no overfit nor underfit is SVR. It can be inferred from the model that slurry flow and particle size together with running hour affect the line thickness significantly. It is in line with the theory and previous research. On the other side, the slurry density doesn’t affect pipeline thickness significantly. The optimization using Solver with the Generalized Reduced Gradient method can minimize the slurry line thickness reduction while fulfilling the target production. From the optimized model, a replacement plan has been defined. The slurry line can be replaced after 8 years since the MP 33 – MP 29 installment and 9 years since the MP 28 – MP 22 installment. The future research should consider incorporating a larger and more diverse set of data from slurry line operations.