This study reports the operational performance analysis of an installation composed by five plate heat exchangers, arranged in parallel, belonging to a glass factory cooling system. Develop an methodology for operational diagnosis, that can be used by the partner company, is the overall objective of this research. The energy balance between hot and cold fluids, the logarithmic mean of temperature differences method for countercurrent flows and the effectiveness-number of transfer units method were used to correlate pressures, flow rates and temperatures recorded in the process, and consolidate a statistical base with heat transfer rates, the amount of heat transferred, the effectiveness of the operation and other thermophysical parameters over a defined period. The influence that the pressure drop and the equipments in operation have on the effectiveness of the system is investigated. To achieve this, computational and statistical tools were used to develop supervised machine learning models. An soft sensor was modeled with linear regression to correlate the effectiveness with the total pressure drop and its coefficient of determination was 0.8663, and a signal classifier was modeled with the K-Nearest Neighbors algorithm and its predictions had an accuracy of 0.9426 in relation to the observed operational classes.