Objectives

• Optimize energy consumption in an industrial refrigeration facility.

• Improve operational efficiency of compressors and system components.

• Apply Artificial Intelligence for more efficient and autonomous operation.

Methodology and Approach

• mplementation of Cirrus AI, a dynamic optimizer based on AI.

• Collection of operational data over several weeks under varying conditions.

• Activation of the system in automatic mode on alternate weeks for comparison.

• Use of machine learning models trained with collected data to optimize in real time parameters such as set-points, start-up sequences, and valve operations.

Problems

• Static control of the facility without considering variable operating conditions.

• Higher-than-necessary electricity consumption.

• Frequent and unnecessary compressor startups.

• Prolonged operation under partial load, increasing wear.

• Inefficient start/stop sequencing.

Results

• Energy savings of 7.65% initially and 10.47% in later, more mature stages.

• Reduction in specific electricity consumption (kWh per ton of refrigerated product).

• Greater compressor stability, fewer unnecessary startups, extended equipment lifespan.

• Adjusted set-points and NH₃ flow that improved COP (coefficient of performance).

• Increased adaptability of the model as more data became available.

• Preparation for next phases including new flowmeter installations and multivariable optimization models.