Optimization of Energy Savings and Economic Feasibility in Industrial Processes Using Frequency Inverters, Centrifugal Pumps, and Machine Learning

This research was carried out at the Tres Valles Sugar Company located in Cantarranas, Honduras, and focuses on reducing energy consumption and improving economic feasibility in industrial processes through the implementation of variable frequency drives (VFDs). The study seeks to address the challenge of energy efficiency in critical industrial sectors, which play a significant role in global energy consumption, particularly in emerging economies. The proposed approach utilizes affinity laws of centrifugal pumps and recurrent neural networks (LSTM) to more accurately predict energy consumption in a closed-loop control system, replacing traditional flow control methods with VFD-based systems. This contributes to achieving the United Nations’ Sustainable Development Goals (SDGs) by promoting more efficient, reliable, and affordable energy systems in industrial applications, supporting cleaner energy use and reducing environmental impact. The study highlights the potential for significant energy savings and economic benefits in developing economies, where energy efficiency is key to sustainable development. Results indicate an estimated annual energy saving of 21,249.15 kWh and a payback period of approximately 1.84 years. The model’s accuracy, with an average discrepancy of only 2.41% between predicted and actual energy consumption after VFD installation, underscores its value for energy management in industrial processes. This work demonstrates that the integration of advanced technologies like VFDs and machine learning can lead to more sustainable, energy-efficient solutions in critical infrastructure sectors.