Creating baselines and alarms
The best alarm limit or baseline is created from data from the machine itself tested under a predetermined set of conditions including a set load, run speed, test setups and test points. If one keeps test conditions and test procedures constant over time then one can trend data to determine if the condition of the machine or asset has changed or is changing. One may ask what happens if one is unaware of the mechanical condition when the machine is first tested, what if it already has some mechanical fault? The answer is simple, if the machine is running and has been running without problems for some time, then this is a good starting point to look for changes in condition, even if the machine may already have some problem. What we are most interested in knowing is if the condition is changing and if the machine is getting worse or failing therefore we can use “today” as a starting point and monitor changes from there.
The best baselines are made from data collected from the machine itself under predefined standard test conditions. Although one set of data collected in this way makes a good starting point for determining future changes in machine condition, a better baseline can be created by averaging together data collected from the same machine under the same test conditions over time. Rather than using a single test as a baseline, a statistical average of data deemed to be “good” and representative of the machine in good health provides a more accurate picture of the machine in its normal range of operation. The statistical average will compensate a bit for slight changes in operating conditions including slight variations in temperatures, pressures or process conditions and will thereby create a more accurate reference for the healthy machine under normal conditions.
COmpare Identical Machines
A common problem when setting up baselines for a machine is that one often doesn’t really know what levels and patterns are “normal” and acceptable for the machine. A second common problem is that one tests a machine and knows it has problems, but due to this fact, the data is not a good candidate for a baseline reading because it represents what the machine looks like with these faults, not what it looks like when it is healthy.
A remedy to both of these situations is to identify other machines in the plant that are identical to this one (preferably before testing any of them) and compare them to each other. In order to do this one must be sure that the entire machine is really identical, meaning the driver, transmission and driven components are identical as are the types of foundations the machines are mounted on. If the machines are to be compared then one must be certain to test each machine in the same exact location and under the same predetermined set of operating conditions (speed and load etc.)
The existence of identical machines offers a huge opportunity for setting up a monitoring program more efficiently and it is definitely worth the time to investigate if there are machines in the plant that are in fact identical and to test these machines in the same way and compare them to each other.
After Balancing or Aligning the Machine
After balancing or aligning machines, be sure to collect vibration data under predefined set of operating conditions in order to create or update baselines. When aligning a machine, one will check for and remedy soft foot, and one will eventually know, via the alignment tools, that the machine is in fact aligned properly. Similarly with balancing, one will take a starting run and trial runs and will eventually come to a point where the machine is balanced to the best of one’s ability to balance it or within accepted limits. This is the best time to create or update a vibration baseline for the machine, since one knows the machine is free from these common faults. Be sure to define standard test conditions (speed, load and test point) for the machine so it can be tested the same way in the future.
After an overhaul
After overhauling a machine it is important to determine if it is still the “same” machine. It is also important to check if the baseline needs to be updated. Regarding whether or not the machine is the “same”; if for example, the motor was replaced with a new motor of similar size but of a different make or model, then the “machine” is not the same as it was before. Forcing frequencies will need to be reinvestigated and a new baseline will need to be created.
If the machine was overhauled, the bearings were replaced and the machine was balanced and aligned, then it is still considered the same machine as before. The baseline however must still be revisited to see if it can be improved. Perhaps the original baseline was created when the machine had some signs of mechanical wear, but the data for the baseline was used anyway because it was the best available under the circumstances and it was an adequate starting point to look for changes in machine condition. Now that the machine is freshly balanced and aligned the new data may provide a more accurate portrait of what the machine should look like when it is in good health. If this is the case, delete the old baseline and create a new one from new data tested under predefined standard test conditions.
