How to prevent unexpected machine failures through condition monitoring

Whether your company is thinking about starting a machine condition monitoring program or expanding one, certain questions inevitably arise: What will this program do for us? What equipment do we need and what will it cost? How do we start?

Knowing the answers to such questions before starting a condition monitoring program can mean the difference between saving thousands of dollars or making a large investment that won’t pay off.

What is condition monitoring?

Broadly speaking, condition monitoring is the process of collecting operating information from a machine while it’s running to detect changes in the machine’s condition that may lead to failure. When such changes are detected, users analyze this data to determine the cause of the problem, pinpoint the faulty component, and predict when it will fail.

With sufficient warning of an impending problem, maintenance personnel can inspect the equipment, evaluate options, and take the best corrective action. They can schedule repairs at a convenient time long before the machine fails.

This method of combining machine condition monitoring with predictive maintenance reduces the possibility of catastrophic failure, keeps machinery running for longer periods, reduces downtime, and minimizes spare parts inventories.

The condition monitoring approach sure beats the old run-to-failure method with its unexpected downtime and costly repairs! It also avoids some of the drawbacks of preventive maintenance programs, which typically consist of periodic machine inspection, followed by overhaul and parts replacement to prevent failure. Though preventive maintenance reduces unplanned shutdowns, repairs may be premature, causing unnecessary downtime and cost.

Almost any manufacturing process can benefit from a condition monitoring program. Such programs keep production on schedule by preventing unplanned downtime and repairs that would otherwise increase costs, reduce productivity, and ultimately lose customers.

Condition monitoring is even more vital for small companies because a catastrophic failure or long-term shutdown can have a greater economic impact here than on a larger company.

In an essential facility, such as a power plant where a shutdown would cause serious consequences, you can’t afford to be without condition monitoring.

What do you measure and how?

Condition monitoring requires sensors at strategic locations to detect machine data. These sensors can be permanently fixed to a machine, or carried with portable monitoring equipment. Typical measurement parameters include:

Vibration. The most common measurement, vibration helps to identify roller bearing defects, imbalance, misalignment, loose parts, and structural problems. It is usually measured in frequency, or Hz.

Several types of vibration sensors are available, and the following will help you select those that best suit your machine’s operation.

Displacement, or eddy, probes can be used to sense the position (in mils or microns) of a rotating shaft with respect to another component, such as a bearing housing. Typically mounted in a drilled hole in the bearing housing, these probes detect shaft imbalance, misalignment, shaft bow, and bearing fluid film instability. They are best suited for measuring low-frequency vibrations, typically up to about 2 kHz, though some units can handle up to 15 kHz. Displacement probes require an external power source.

Velocity sensors measure the rate at which displacement changes (in./sec). These sensors monitor bearing or machinery housing vibration and detect conditions such as bearing fatigue, turbine or fan blade problems, gear problems, and loose parts. They’re especially effective in low to mid-frequency vibration ranges (10 to 1,500 Hz).

Velocity sensors have moving parts that can stick or fail. Therefore, they should be used only where accompanied by other sensor types, particularly on critical machinery.

Accelerometers measure the rate of change of velocity in a frequency range from near zero to 40 kHz or more.

Continue on page 2

Because they have no moving parts, accelerometers are generally more reliable and last longer than velocity sensors. They are available with or without signal conditioning electronics for use with monitoring equipment.

Acoustics. Some machine problems occur at ultrasonic frequencies (about 150 to 500 kHz), which are above the range of conventional accelerometers. But new high-frequency acoustical sensors that use Spectral Emitted Energy (SEE) technology can detect defects in the ultrasonic range.

This type of sensor detects the impact between two components, such as a bearing ball and outer race, when they penetrate the lubricant hydrodynamic film between them (metal-to-metal contact). It also can detect excessive dynamic loading, contamination, or bearing defects.

The sensor detects defects when they are very small, hence at an early stage. With this early warning, users can often correct operational conditions (improper loading, minor misalignment, improper lubricants) to extend bearing life.

Temperature. When a bearing starts to fail, excessive friction causes its temperature to rise. Thermocouple sensors on the bearing housing or in the lubricant detect this temperature rise, which signals impending bearing failure. When combined with other data, temperature measurement (thermography) helps you recognize problems at an early stage.

Lubrication. Oil analysis also provides important clues to machine condition. Foreign substances degrade the lubricating properties of oil and cause bearing failures. Measuring the quantity and size of foreign particles helps you determine if a wear problem exists, and even which part of the machine is wearing and how fast.

Though there are many ways to test lubricants, the three most common methods used in condition monitoring are:

Spectrochemical analysis, which indicates lubricant quality based on the condition of its additives and on materials suspended in the lubricant. This type of analysis also provides viscosity data plus information on the presence of harmful dissolved solids and water.

Wear particle analysis, or ferrography, which quantifies large (>5 microns), small (<5 microns), and total number of wear particles, both ferrous and suspended nonferrous types. This technique is sometimes followed by microscopic examination to determine the type of wear.

Filtered particle (or contaminant) counts, based on particle size.

The best results are usually achieved by combining these three oil-analysis methods.

Combining parameters. In addition to vibration, temperature, acoustics, and lubrication, other measurements such as rotating speed, pressure, or flow, and testing methods such as non-destructive testing (NDT) may help in detecting early signs of deterioration.

Choosing the appropriate measurement parameters requires a careful evaluation of the entire machine. It’s an ongoing process that often requires fine tuning. As a rule of thumb, vibration, temperature, and lubricant measurements are good indicators of rotating machinery condition. Pressure, temperature, and flow help in assessing machine performance (efficiency).

By correlating the data from multiple parameters, engineers and technicians can better understand the behavior of the machine system, locate the faults, and determine the mechanical and operating conditions that caused the faults.

Whatever type of data is collected, it is important to evaluate changes in values (rather than a particular value) as a means of predicting failure. If any dynamic value, vibration or otherwise, suddenly increases, you should:
• Verify its validity.
• Examine the parameter’s trend history.
• Examine trends in related data.

These can pinpoint when a condition started to develop and provide clues to the probable cause.

What equipment do you need?

Many equipment options are available, from hand-held data collectors to sophisticated instruments with analysis software. Costs range from a few hundred dollars for a hand-held vibration pen to $20,000 for a datalogger and related software. Extensive online data acquisition and customized software systems cost even more.

Hand-held meters. For vibration and SEE readings, hand-held instruments, Figure 1, are inexpensive, lightweight, and easy to use. Some even come in pen size. These portable instruments offer: • Quick inspection where deterioration in machine condition is suspected. Their simplicity lets supervisors and operators check factory floor equipment during daily tours of production areas.

Continue on page 3

• Monitoring of both vibration changes and lubrication programs.

Most hand-held devices, however, are limited in the type and range of measurements they can perform. Unlike larger devices, hand-held units typically lack data storage capability, so they require manual record keeping. More advanced hand-held units incorporate memory and software that permits limited, computerbased trending.

Portable data collectors-analyzers. These small, lightweight instruments, Figure 2, perform more extensive functions than hand-held meters. For example, they record vibration data for later analysis, freeing users from manual recording chores. Many types display vibration trend plots and abnormal measurements.

An operator carries a data collector to each measurement point on a machine and measures vibration with a probe or permanently attached sensors. Multiplechannel data collectors permit several types of analysis, such as acceleration enveloping and SEE, and they can be used for other functions.

For example, an operator may use a data collector not only to inspect machinery, but also to enter other process measurements, such as temperature or flow, during daily rounds of a plant. The data collector then performs plant performance trends or is linked to a database management software program for storage and further analysis.

Online data acquisition and analysis. Online systems continuously measure data via sensors that are permanently attached to strategic measuring points, thus providing an ongoing means of detecting machine problems. Each data acquisition unit monitors and collects data from a number of sensors. Then it uploads information to a computer for display or later analysis by a database management software program.

Some units have a protection feature that trips a machine Off when measurements reach a preset level, thus preventing damage. These devices are useful for continuously monitoring critical machines or those with high consequences of failure.

Software programs. Condition monitoring software programs aid in the collection, management, and analysis of machinery data. They perform various types of analysis, Figure 3, such as plotting data trends to show changes in a machine’s condition or comparing data levels with alarm set points.

New, sophisticated programs, called Computerized Maintenance Management Software programs (CMMS), identify costs associated with a particular machine, including parts, repairs, and scheduling, and integrate this information with the company’s overall planning process.

The type of software you choose depends on the type and number of machines being monitored, and the type of monitoring equipment. When evaluating a software program, make sure it has the ability to customize reports and adapt to your plant operation, not the other way around. Also look for a program that can interface with other software and new equipment, so it won’t become obsolete as your condition monitoring program expands.

Part 2 of this series will tell how to implement a condition monitoring program and evaluate its results.

Doug Johnson is a training manager at SKF Condition Monitoring, San Diego, Calif.