It used to be a relatively common practice for plants to wait for equipment to fail, and then repair it. But with greater pressures to realize improved productivity and profitability, such "reactive" approaches to maintenance have largely fallen by the wayside in favor of more proactive strategies.
A range of predictive technologies can be implemented to detect developing machinery faults at an early stage, and before they become problematic. These can include performance parameter sensors, specialized monitoring devices and analytical and data management software to capture, trend, diagnose and report timely information on the operating conditions of machinery assets. The key to choosing appropriate predictive technologies is to understand how a particular machine fails, what symptoms will be visible and detectable before it fails, and how fast that machine will deteriorate.
Among the parameters to monitor, the analysis of machinery vibration and lubricant can help chart mechanical integrity and operational health.
Measuring effects of vibration
Many machinery problems manifest as vibration, which is widely considered the best operating parameter to assess a machine’s condition. Vibration can detect machine fault conditions such as:
- Unbalance
- Misalignment
- Oil film bearing instabilities
- Rolling bearing degradation
- Mechanical looseness
- Structural resonance
- Soft foot
- Rotor bow
- Cracked rotors
Vibration measurements are also quick and fairly non-intrusive since the operating equipment is undisturbed.
Typically, the user compares the overall vibration level to a predetermined alarm level and, if the alarm level is exceeded, action can be taken before equipment failure occurs. There are also industry standards, such as ISO, which provide allowable vibration severity levels for specific types of equipment and operating speeds. These standards can be used as guidelines to establish equipment vibration alarm levels.
A variety of hand-held vibration monitoring tools can also do the job, ranging from low-cost vibration pens and meters measuring overall vibration levels to more sophisticated portable analyzers that collect time and frequency content of vibration, along with other predictive information. These are packaged into a feature-rich, compact size with loads of storage.
Most pens and meters measure overall vibration velocity over a frequency range of 10 Hz to 1 kHz (ISO Vibration Standard 10816-1), which is considered the best range for judging rotational and structural problems (imbalance, misalignment, looseness and stress applied to components). More sophisticated portable data collectors/analyzers collect and store machinery vibration data over several different frequency ranges (sometimes up to 20 kHz) and display high resolution FFT frequency spectra and time domain waveforms on a high-definition LCD screen.
Collected vibration measurements can be analyzed on the spot or downloaded to a specialized software application on a computer work station or network for analyzing, long-term trending, and reporting abnormalities. The ability to both detect vibration problems and diagnose specific machinery faults with a portable data collector/analyzer is greatly enhanced due to their FFT spectrum analysis capabilities.
Online surveillance monitoring systems (available in hard-wired and wireless configurations) can complement the efforts, facilitating a proactive approach to reliability with around-the-clock monitoring of machinery, regardless of location. These systems collect data continuously or over a predetermined data-collection period from permanently installed sensors and then relay the findings to a host computer for analysis and/or directly to a plant’s controls system for immediate action.
Detecting defects in bearings
The vibration measured at a machine’s bearings will open a window into equipment health, because most machine problems have distinct vibration symptoms. For example, as a bearing deteriorates, there are specific vibration symptoms easily recognized by an experienced analyst.
A bearing can degrade due to:
- Improper lubrication
- Contaminated lubrication
- Heavier loading than anticipated, often caused by other machinery problems such as imbalance, misalignment or a bent shaft
- Improper handling or installation
- Surface fatigue
- Misapplication
The "good news" is that a failing bearing produces vibration symptoms that are detectable well in advance of failure. If detected, properly analyzed, and the progression of the damage is monitored accordingly, these signals provide maintenance personnel adequate time to correct the cause of the bearing problem (effectively extending the bearing’s service life) or, if necessary, time to replace or repair the bearing during scheduled downtime.
A two-step approach can be utilized for detecting and analyzing bearing faults: early detection and in-depth fault analysis.
- Fault Detection: This should be performed as easily and reliably as possible. Workers can configure vibration monitoring devices to generate alarms when bearing conditions change. Detection systems must be configured properly and structured to communicate quickly.
- Fault Analysis: After an alarm indicates a change in bearing condition, in-depth analysis should be performed to confirm the fault, determine severity, diagnose root cause, and determine how long before the bearing will fail. Trends from original detection measurements should be analyzed, and more measurements and inspections may be required at additional machine locations to obtain more detailed and insightful information.
Analyzing condition of lubricant
Lubricant inspection and analysis serve as a particularly practical method to help detect problems with machinery assets, especially since many characteristics can be examined visually.
- Water contamination can be observed with clarity in a standing sample
- Ferrous materials (filings, metal dust) can be detected using a magnet drawn up the side of a glass jar containing lubricant diluted with a solvent
- Flow and discoloration can be noted in a bull’s eye sight glass
- Non-ferrous particles can be evaluated by residue on filter paper
- Viscosity can be monitored using simple in-plant tools
Beyond the day-to-day observations, though, lubricant analysis as a Predictive Maintenance activity should target at least three critical areas:
- Machinery wear particles: Small wear debris can be measured by standard emission spectroscopy techniques. Larger, severe wear particles should be measured periodically as well. In machinery where the predominant lubrication regime is hydrodynamic (full fluid film) and the wearing components are non-ferrous bearing surfaces (such as with sleeve and pad bearings), Rotrode Filter Spectroscopy (RFS) is appropriate. For machines with rolling element or steel gear component wear as the primary failure modes, the appropriate method is direct reading ferrography (DR). Analysis of the wear particles provide clues as to which components may be degrading, and aid in determining time to failure.
- Contamination: Selection of an analytical method for measuring and evaluating contamination will depend on the machine, lubricant type and environment. Contamination can be present in four different forms: gaseous, fluid, semi-solid, or solid. A rule of thumb is to select test methods relevant to the probability that a specific contaminant can enter the machine’s lubricating system or be produced within the machine.
- Oil lubrication degradation: Standard analytical methods for measuring oil degradation include increases or decreases in viscosity or changes in alkalinity and/or acidity measured by the pH. When changes occur (from degradation and not contamination), the lubrication is overdue for changing, and sludge and varnish have already begun to form in the machine.
- Grease lubrication degradation: Grease degradation is more difficult to detect. Samples are difficult to obtain and analytical equipment and methods to analyze grease properties can be quite sophisticated. The simplest way is to trend operating temperatures along with vibration measurements.
Before embarking on any predictive maintenance program, a clearly defined maintenance strategy should be in place. Decisions to apply related technologies should be prioritized according to the risks associated with equipment failure, the possible financial consequences, the impact on the safety of personnel, production processes and the environment. As an essential first step, a benchmark assessment of equipment at facilities can contribute effectively to the most appropriate design, implementation and management of machinery assets.