Equipment condition monitoring: The problematicon statistical control charts

Every system is subject, along its life cycle, to several degradation processes that progressively degrade its state and increase its probability of failure (reliability reduction). This is true for the generality of systems—mechanical, electrical, software, human or organizational. If nothing is made—if there is no maintenance—every system will eventually fail. In order to start the condition control and improve its reliability, fixed sensors should be chosen in order to collect data of vibration, oil and water pressure, and temperature and particle size- among others. The actual equipment condition must be known, estimated or predicted from the collected data [1]. Unless the state or condition of system is directly observed, that condition is a latent variable in the sense of statistical theory—a variable not directly observed but with observable effects in the manifest variables, as the just mentioned observed variables associated to sensors. To provide support for the decision maker— at last for critical selected systems—it has been shown that some control statistical techniques are effective to ascertain trends and predict needs of future interventions out of observed data. When data is not appropriate, the combination of statistical techniques with simulations and can be considered. With this work we intend to show that control charts can be decisive as instruments of control of equipment monitoring and functioning [4], although there are some problems when applied to the conditioned maintenance; some of those problems can be overcome using a EWMAQ modified chart, adjusting its parameters [3].