T ² charts applied to mechanical equipment condition control

The use of CBM (Condition Based Maintenance) on critical equipment allows on-line monitoring of equipment vibration level, for instance. The analysis of unstructured data generated by vibration sensors can cause false alarms, in particular when several variables are analyzed. Therefore, T ² multivariate statistical control charts can play an important role in monitoring the condition of critical production equipment, allowing early detection of imminent failure and trigger a maintenance task in order to protect equipment from failure and destruction. When applying these control charts it is important to check whether the observed variables are independent. When this assumption is violated, we suggest modeling the data through the ARIMA (p, d, q) and the consequent application of T ² control charts to model residues/ forecasting errors. This paper applies this methodology to study the condition of an electric pump, subject to a set of forced disturbances that could predictably trigger damage if, after the detection of anomalies by T ² control charts, suitable measures were not taken. So it will allow to an effective early detection of anomalies and taking proactive actions that reduce the probability of an unexpected failure.