Take two and call me in the morning

Use prodictive technologies like a tag team that wrestles machine problems right out the door

Just what is this thing we call condition monitoring? Two of the best writings about it are very concise.

...most failures give some warning of the fact that they are about to occur. This warning is called a potential failure, and is defined as an identifiable condition which indicates that a functional failure is either about to occur or is in the process of occurring." -- John Moubray, Reliability
Centered Maintenance.

"The objective of condition monitoring is to provide information that will keep machinery operating longer at the least overall cost. Condition monitoring and predictive maintenance are changing maintenance from an expensive, never-ending exercise in emergency fire fighting to an efficient, organized process of least cost precision action. Contrary to some popular beliefs, the objective of condition monitoring is not to establish new records for the number of measurements that are being recorded or to demonstrate analytical brilliance in diagnosing difficult problems." -- John Mitchell, Machinery Analysis and Monitoring.

The four principles of condition monitoring of machinery are straightforward. First, the goal is to identify changes in the condition of a machine that indicates some potential failure. Second, it identifies physical characteristics that collectively indicate the current condition of the machine. Third, each characteristics is measured, analyzed, and recorded so as to reveal trends. Fourth. over a period of time, the progress of these trends represents the deterioration of machine condition and can be used to determine maintenance actions.

The four principles of condition monitoring of machinery are straightforward.

Studies by the U.S. Navy and Tracor Applied Sciences on equipment ranging in size from 15 to 4,000 hp showed that monitored equipment has one-sixth the catastrophic failure rate of equipment that was not monitored. In addition, the probability of detecting an impending failure ranged between 92 and 95 percent with a false alarm probability of between 5 and 8 percent when the proper monitoring interval and alarm values are selected.

Vibration
For most facilities, vibration analysis is the cornerstone of the condition monitoring and predictive maintenance programs. Vibration analysis has been proven to be the most successful predictive technology for rotating equipment. to increasing equipment availability and reliability. Maximizing the finite life associated with rolling element bearings and optimizing equipment production life requires , minimizing excessive wear caused by misalignment, imbalance, and resonance. Routine and consistent gathering of vibration data is vital to the process of analyzing and trending machinery health. Vibration monitoring establishes and verifies acceptance standards of rebuilt or newly installed equipment.

Vibration analysis takes two forms. The first form, machinery protection or protective monitoring, detects sudden changes in condition that could lead to catastrophic failure. This form is particularly valuable for machinery that could represent a threat to the health and safety of people, or cause an environmental incident.

Machinery protection systems monitor key measurements continuously, comparing readings against alarm levels. Should a measured variable exceed an alarm value, such systems may take a number of actions from sounding an audible warning to tripping the machine to prevent further deterioration. Whatever it is that machinery protection systems do, they must do it quickly and reliably. Protection systems for critical applications are constrained by a widely recognized standard known as API670, published by the American Petroleum Institute.

The second form of vibration analysis is known as predictive monitoring or predictive maintenance. This form identifies the earliest onset of incipient failures. Doing so allows predicting the likely progress of the failure. This allows planning a suitable response.

Predictive maintenance systems are much more diverse than protection systems. Although they typically have less-demanding standards for measurement accuracy and speed, much more data processing is needed to turn these measurements into useful information. Data acquisition systems may be portable or on-line (sometimes referred to as surveillance systems), and may also accept measurements from other types of system such as machinery protection and
control systems.

Oil analysis
The first time someone suggests oil analysis methods, the inevitable questions is "why?" You don't monitor the oil in your car--you just change it every three thousand miles or three months. Three main reasons to monitor the oil in your machine are cost, valuable information, and the effect on the life of the machine. We can take a car and $20 to the local lube shop and get the oil changed. With many of our machines requiring 55 gallons or more of specially formulated oil, the cost to change industrial oil can be incredible. With the current EPA regulations, it can cost almost as much to dispose of the oil as buying new oil. Thus, the cost of simply changing the
oil increases even more.

More importantly, however, the oil contains valuable information about the condition of the machine. If a bearing is beginning to spall, then where do the chucks of metal end-up? Finally, as people began to monitor the condition of the oil they found that keeping the oil clean and free from contaminants extends the life of both the oil and the machine.

Oil analysis takes three forms. The first form of oil analysis is fluid contaminant analysis and control. It specifies whether or not contaminants entered the oil system. Knowing this information about contamination allows correcting potential problems before significant damage occurs.

Fluid contaminant analysis measures particle count, viscosity, and water content in the oil. Changes in these parameters indicate a contaminant entering the system. The contaminants could be ingested from outside the system, for example water or dirt. On the other hand, the system can be generating contaminants like bearing particles. After identifying the type of contaminant, one can take appropriate corrective action.

Fluid properties analysis involves analyzing the chemical properties of the oil to determine the degree of oil degradation. Using elemental spectroscopy, FTIR spectroscopy, and other wet chemistry tests determine the chemical properties of the oil. Normally, these measurements are performed in an off-site laboratory. From the chemical properties, you can determine whether the additives are breaking down and no longer providing protection, for example oxidation resistance. You can also find out if the oil is no longer lubricating the machine appropriately.

Wear particle analysis involves the in-depth analysis of the particle in the oil to determine whether bearings are spalling, gears are chipping, or dirt is being ingested. Experienced microscopists separate the particles from the oil to examine them closely. The color and shape of the particle identifies the exact source.

Thermography
Thermographic analysis uses the heat generated and transmitted by a machine to determine the condition of the machine. Temperature differentials relative to ambient temperature are used to prioritize deficiencies. The real power of thermography is that it quickly locates and monitors, in real time, both maintenance and production problems.

This technology works alone in many applications and detects problems that cannot be identified with any other means. Thermographic information can be particularly important in cases like electrical circuits and connections that may show no visible signs of deterioration until moments before failure. Modern thermographic equipment allows effective scanning and problem detection of very difficult problems. Infrared monitoring and testing is non-intrusive and is performed with equipment in service at normal operating conditions. The thermography applications for electrical
systems include:

• main transformers,
• motor control centers,
• circuit breakers,
• distribution panels,
• connections,
• cable trays, and
• control systems.

• rotating equipment bearings,
• electric motor and pump casings,
• couplings,
• steam traps,
• condensers, heat exchangers, piping,
• steam and compressor leaks,
• valves,
• brick furnaces,
• process applications,
• insulation deterioration, and
• vacuum leaks.

• pressurized gas, air leaks,
• vacuum leaks,
• boiler tube, heat exchanger leaks,
• steam traps,
• valve seat leaks,
• bearing lubrication timing,
• bearing faults, and
• compressor valve leakage.

An off-line surge test detects stator failures. While the motor is disconnected, an external surge of electric is injected into the stator windings. Measurements identify winding shorts.

A second on-line current test detect rotor problems. This test involves making a high resolution measurement of the input current while the motor operates normally. The measurements identifies cracks or high resistant joints in the rotor and determines the positioning of the rotor in the stator.

Ultrasound reveals corona, arcing, and tracking, which may not show up using thermography.

Typically, the past experience method is used. However, as the reliability of machines becomes more important, a modified version of Reliability Centered Maintenance determines the best methodologies to maintain the desired level of reliability.

A U.S. power plant was monitoring a pump using monthly vibration and oil contaminant monitoring and bi-annual oil properties analysis. During a monthly survey, the vibration monitoring program detected an unusually high vibration level. Upon further examination, the vibration analysts diagnosed the problem as misalignment and submitted a work request. Completely unknown to the vibration analysis, an oil analysis specialist pulled a sample and received an extremely high contaminant level and also submitted a work request. Fortunately, the area engineer received the requests simultaneously and was smart enough to do some further investigating before issuing either work order.

As it turned out, when oil had recently been added to this machine, the reservoir was overfilled. This caused the crud from the inside top of the reservoir to pollute the oil and reservoir. Simply changing the oil as suggested by the oil analysis specialist or re-aligning the machine as suggested by the vibration analysts would not have corrected the problem.

Instead, the oil needed to
be changed and the system cleaned before fresh oil would correct the problem. The alignment problem was related to the viscosity of the contaminated oil changing the stiffness of the bearing. The lesson to be learned is that condition monitoring techniques need to be used in parallel and not in isolation.