Jonathan Newell discovers how the interpretation of acoustic emissions goes beyond NDT to encompass condition assessment of slow moving bearings.
Monitoring the condition of bearings using vibration sensors is a well established technique that has been providing good results to maintenance engineers for years – but what if the bearing runs at a frequency below that which vibration measurement fails to be appropriate? Slow moving bearings of around 60rpm or lower and low-speed reciprocating bearings require a different approach to detecting defects that could result in eventual failure. Such failure can have significant cost implications in high value capital plant and machinery.
Acoustic Emission (AE) measurement is a passive technique for listening to the degradation of materials before any vibration develops and is being used successfully to assess the condition of slow moving bearings of down to less than one revolution per minute, drawing on its long heritage as a Non-Destructive Testing (NDT) technique of assessing the condition of such structures as aircraft composites, suspension bridges and offshore oil platforms.
Interpreting bearing noise
To assess a bearing, an array of AE sensors is attached to the bearing housing and connected to analysis software such as that used by LSB-PAC by Mistras Group which provides information from which the condition of the bearing can be assessed.
The software interprets the various noises, their frequency and energy profile to distinguish between different possible defects, such as the growth of subsurface cracks, the existence of debris or the deformation of rotating components within the bearing structure.
Emissions from corrosion product break-up can indicate the corrosion of tracks or balls, surface fretting emissions imply insufficient lubrication and other noise profiles can lead to the detection of cracking or plastic deformation of bearing material such as the balls, track and cage.
Mistras Group’s General Manager, Tim Bradshaw explains this by saying that the various pops and crackles emanating from the bearing can be identified similarly to the way that the human ear can interpret the difference between the sounds of a pencil being snapped and a sheet of paper being torn.
By using an array of sensors, directional information is also available which gives an indication of where the defect is located, whether within the static or rotating structure or the rolling elements of the bearing.
The assessment of a bearing using AE isn’t as simple as plugging the sensors in and the name of the defect popping up on a computer monitor. Interpretation of the results is a task that requires considerable expertise.
“There are a number of challenges,” says Bradshaw. “For example, if there is a large amount of high frequency background noise, for example as produced by electrical drive motors, which is much more than from hydraulic motors. For this reason, for each installation, it’s necessary to take the context of the local environmental conditions into account.”
“However, once these factors have been accounted for, the grading technique is very similar, whatever the application, whether a wind turbine or a large industrial rotating turret. In all cases, the assessment can be done immediately without the need for comparison to a “known good” profile for the bearing.
Assessment versus monitoring
Once assessed, decisions need to be made on what actions are needed in the future. A good bearing will need monitoring and a bad bearing will need replacement, repair, lubrication or continuous monitoring to see if the problem deteriorates further.
When a defect is detected, detailed analysis involves the identification and elimination of extraneous noise, grading the severity of the bearing signature and identifying any features that may indicate the root cause of the problem.
If continuous monitoring is required, it can be done remotely, or, for inaccessible bearings such as on vessel propulsion systems and wind turbines, by permanently installed sensors, or by so-called patrol monitoring to enable an operator to sample bearings in multiple installations.
For permanent installations and patrol monitoring, the analysis can be pared down to those parameters which can be linked to an alarm once a threshold has been breached. In these cases, the monitoring task is easier if the bearing behaviour is known and predictable and other parameters have been taken into account to prevent false alarms.
“A simple example is whether the bearing is actually running or not,” says Bradshaw. “Very low noise levels could mean a perfect bearing or one that isn’t running – therefore the operational state of the bearing needs to be included in the monitor.”
Acoustic Emissions for NDT
Mistras has been involved in the assessment of acoustic emissions in industrial environments since the 1980s, predominantly as a method of Non-Destructive Testing, still the mainstream use of the technology.
As an inspection method, AE analysis can determine fatigue crack growth, assess defects in pressure vessels and even measure the growth of corrosion in inaccessible metal containers. The progress rate of corrosion in tanks for the oil industry is the quietest profile that Mistras’ Tank-Pac product has listened to. Rather than assessing the thickness of the base of the tank, the product can detect the signature sound pattern of the growth of the corrosion product, needing only around an hour’s listening to make an assessment of the degree of corrosion present in the tank.
Such fine listening precision is now finding wider application in the condition monitoring field, With the growth of wind turbines and other uses for slow moving bearings in propulsion and manufacturing industry, it represents a new way of reducing the high maintenance costs associated with the replacement of defective bearings with poor accessibility.
“Cost savings can be quickly achieved by preventing and reducing unplanned stoppages in addition to catastrophic bearing failures and by allowing maintenance to be planned during low-peak periods,” concluded Bradshaw.
Bearings in wind turbines
One application for AE assessment of bearing condition is for wind turbines, which have three “sets” of bearing types which have very different functions. The main gearbox contains a group of bearings which run at different speeds and AE typically isn’t used in this application, and vibration monitoring is favoured.
The main bearings run at up to around 13rpm and are an ideal application for permanently installed AE sensors for monitoring the condition, since the bearing has poor access and is usually part of an installation that has many wind turbines in a group at an isolated location. By linking the AE sensors to parametric alarms, maintenance departments can be alerted to changes in the condition of the bearing and can make a decision on service requirements.
The pitch bearing on the wind turbine blades is another interesting application. It is a positional bearing without full rotation. It is slow moving and covers an angle of just 87 degrees. However, its condition is critical to the efficient and safe operation of the wind turbine. If the bearing on one of the blades doesn’t operate correctly, the whole turbine balance will be affected and a failure to “feather” the blade in high wind conditions could result in overspeed of the turbine. This bearing has a noise profile associated with certain defects like any other and using Acoustic Emissions analysis can detect potential problems with the bearing before it fails.