Predictive Maintenance for the Efficient Use of Industrial Agitators in Distillers and Reactors

Abstract

Distillation is one of the most common processes in companies of the chemical sector. Its overall performance cannot be based only on physical-chemical processes carried out, but as a whole, including maintenance and operation costs and trying to achieve the best possible energy efficiency. By means of predictive maintenance techniques, such as vibration analysis, ultrasound, thermography, etc., it is possible to anticipate equipment failures, avoiding unnecessary stops and significant losses, and identify dangerous or inappropriate functional conditions to achieve proper operating dynamics. This article describes a methodology to analyze the functional conditions of the agitators of large distillers and reactors. Based mainly on vibration analysis techniques, this methodology tries to find the best operating conditions for the equipment, both electromechanically and energetically, and identifying, if that were the case, those that are dangerous for the equipment. This methodology is currently being applied in a large process plant, and has been extended to gradually integrate all the agitators of the industrial plant, obtaining a substantial improvement in its energy efficiency, as well as a significant reduction of failures, particularly in the mechanical seals, a very expensive and essential element to maintain the correct pressure and temperature required for the process.

Optimisation of Maintenance Policies Based on Right-Censored Failure Data Using a Semi-Markovian Approach

“…The literature on maintenance is vast, including Reliability Engineering [4][5][6][7][8][9][10][11][12], maintenance policies and modelling [13][14][15][16][17][18][19][20], and optimal preventive maintenance intervals [21][22][23][24]. However, fewer papers address the problem of uncertainty in lifetime distribution [25][26][27][28], either by analysing several time-based maintenance policies having uncertainty in the parameters of the lifetime distribution [29] or using right-censored data [30], and adopt the Markovian approach of transition between states (operational, preventive, and corrective maintenance intervention) [31][32][33][34][35][36].…”