Condition monitoring can be described as a process of monitoring machine condition, for example: vibration levels. Condition monitoring is a major part of predictive maintenance and is used to detect early vibration problems so that parts can be ordered and maintenance can be scheduled. Why do we need to perform condition monitoring? When a machine fails, production is stopped. Downtime can be very expensive and secondary damage can occur. Instead of simply replacing a bearing, the entire rotor may need to be replaced. The repair costs can increase. In addition, it takes longer to repair, resulting in longer downtime and higher labor costs.
Large amount of vibration data are collected during monitoring program. It must be analyzed so that early vibration problems can be detected and decisions can be made regarding repairs. This process takes a lot of energy and time, thus it is very costly. Vibration analysts usually invest time to set up good alarm levels for the machines so that the data can be scanned in the computer for measurement points that are not within acceptable limits.
It is often difficult to set the alarm limits that are correct for machines. Alarm limits can be calculated from actual vibrations levels on machinery. A common rule is that if the vibration levels are as high as twice of their original baseline measurement, a problem exists. The baseline measurement is ideally a measurement that we believe best represents how the machine should vibrate. It is a measurement with which we can compare against to determine if a change has occurred.
We can use calculated alarm limits once there is a history of data and set alarm limits based on a change above reference levels for overhauled machines and machines known to be running well. A common rule is: 50% increase is significant and needs investigation; 150% increase is substantial and needs investigation with priority.
We can set baseline in two ways. Firstly, we can take measurement of first set of data and set it as the baseline. Many software programs easily compare the current vibration data with the baseline measurement. Secondly, we can take a reading after a machine has been repaired or overhauled (it has had a chance to break-in), and set it as the baseline.