Heinz P. Bloch
Consulting Engineer
Montgomery, Tex.
Joseph R. Carroll
Industrial Consultant
Grand Junction, Colo.
A carefully chosen, periodically implemented, preventive maintenance (PM) program can sometimes make more economic sense than predictive maintenance because not all equipment responds to purely predictive techniques.
Predictive maintenance uses projected data or trends to determine the trouble-free service life of equipment. In contrast, preventive maintenance is doing the minimal routine work necessary to ensure the equipment remains in proper operating condition.
Increasingly, modern process plants are intent on deferring equipment maintenance until the need for maintenance has been clearly established. To determine maintenance needs, these plants typically apply predictive, instrument-based condition appraisal techniques, including vibration and corrosion monitoring, lube-oil analysis, and continuous determination of machine efficiency deterioration.
Predictive maintenance methods have become an indispensable part of maintenance planning and shutdown strategies of modern process plants. The predictive approach has been used for many decades on rotating machinery, such as pumps, compressors, steam turbines, and electric motors.
Information is gathered over a long period on such items as vibration severity. Lubricating oil samples are collected, analyzed, and trended; and in some cases, considerably more sophisticated measurements, such as noise, temperature, or vibration-signature analyses are included in predictive maintenance routines.
Very often though, predictive or condition-based maintenance is practiced to the exclusion of preventive maintenance. This can be a serious and costly mistake.
An example would be waiting for lubrication deficiencies in rolling element bearings to manifest themselves through excessive vibration. In this case, it would be far more cost-effective to simply implement a regular program of periodic lubrication.
Another very important area where preventive maintenance neglect has caused extended equipment outages relates to steam-turbine trip systems. Here, periodic exercising of trip valve stems and proper lubrication of sensitive linkages will prove inexpensive and far more effective than most predictive approaches.
A modern process plant is, therefore, confronted with the task of determining whether preventive maintenance, predictive maintenance, or a combination of both is the best approach to various operations. Some specific examples of equipment types show when, where, and why it may be in the best interest of operating and maintenance departments to carry out traditional PM.
When a preventive maintenance program is developed, it must involve all personnel responsible for carrying out and meeting all of the program's goals.
Maintenance effectiveness surveys can serve to sort out which of the two approaches is more appropriate in a given situation. Conducted by two or more engineers experienced in both maintenance management and equipment reliability assessment, these surveys can provide rapid and valuable information on how to best utilize all available maintenance resources.
PREDICTIVE MAINTENANCE AREAS
Some types of equipment are well-suited to predictive maintenance techniques. An example is metals inspection for corrosion or erosion in vessels and piping systems.
A typical equipment inspection program is based on the collection of extensive data involving the rate of corrosion or erosion of specific equipment. These data are then extrapolated to determine the remaining life of the equipment.
The resulting predictions can, and often do, set the run length of the equipment and determine a good portion of future maintenance to be undertaken.
A good metals-inspection program can result in modifications to operating conditions or selection of alternative materials. Furthermore, such a program provides reasonable assurances that the equipment, given similar operating environments, will operate reliably and safely for the predicted period.
A similar predictive approach is valid for most pump maintenance programs. Data are taken, trended, and sometimes analyzed to evaluate the hydraulic and mechanical performance of these machines.
The trended data provide a reasonable certainty that the equipment will operate satisfactorily. More importantly, predictive techniques can indicate the probability of an incipient failure if timely maintenance is not performed.
The analysis does not avoid repair work, but in the majority of cases prevents more costly repairs or even catastrophic failures. This essentially is one of the major goals of an effective predictive maintenance program for pumps and, to some extent, motors.
Other major objectives are to extend the operating life of the equipment and to allow for proper work execution planning. Achieving all of these objectives certainly reduces overall maintenance costs.
Obviously, predictive programs have the advantage, in that measurements and data can be collected while the equipment is in service. In-service data can then be used to predict the future with a high degree of credibility.
For some equipment, however, predictive maintenance is difficult because on-line data are hard to gather. These types of equipment require visual inspection or offline measurements to collect the data necessary to determine physical condition.
AREAS WHERE PM IS BETTER
For some equipment, predictive maintenance may not work well because of the potential seriousness of a failure. Steam turbine stop valves or trip and throttle (T/T) valves are examples of this type of equipment, where rotor overspeed events can be catastrophic.
Most process plants will disassemble line valves for inspection and repair when problems, such as excessive seat or external leakage or sluggish operation, are apparent. But unless turbine stop valves and T/T valves are periodically tested and inspected, the valves may not work when called upon.
For these types of equipment, maintenance must emphasize preventive techniques. For instance, lube oil purification and periodic exercising of the screw spindle of a common T/T valve (Fig. 1, Item 36) are two effective preventive maintenance techniques.
Water-contaminated lube oil is prevalent in steam turbine hydraulic systems. With wet, contaminated lube oil on the hydraulic side of the oil cylinder of the T/T valve (Fig. 1, Item 73), corrosion products have sometimes lodged between the piston rings and cylinder wall.
In one recent catastrophic failure at a medium-size U.S. refinery, water-contaminated lube oil was indirectly responsible for causing major damage to a large steam turbine. The problem began with vibration caused by imbalance in the coupling between the turbine and the driven equipment.
Vibration caused the coupling bolts to shear, instantly unloading the turbine and resulting in an overspeed condition. The overspeed condition activated the T/T valve, but the piston stem (Fig. 1, Item 64) did not move because of corrosion products on the cylinder wall, caused by water in the lube oil. Steam to the turbine was not shut off, allowing the turbine rotor to speed up until it disintegrated.
Oil purification, by using a system depicted in Fig. 2, could have prevented corrosion products from building up on the cylinder wall by keeping the oil dry.1 The purification system, designed to be permanently installed on the oil reservoir, continuously removes water from the lube oil by means of air stripping.
Of course, a predictive program of lube oil analysis could have assisted in determining lube oil quality.2 But a thorough cost comparison with a lube oil purification system would show an advantage for the preventive purification technique (Table 1).
Many T/T valves also fail to trip because impurities in the steam collect on sliding parts in the steam passages of the valve. The extent to which these deposits have collected, or the extent that they will impede valve operation, is very difficult to predict.
One sure way to verify that the valve will operate when called upon, is to periodically exercise the valve over its full range of travel. Fig. 1 shows that a TfT valve has a hand wheel that allows stroking or partial closing of the valve stem with the turbine in normal operation.
T/T valves are typically oversized. Therefore, partially closing the valve with the turbine on-line will not cause the turbine to lose speed.
Exercising the valve verifies the integrity of the sliding surfaces and also wipes any deposits off of the surfaces before they buildup to serious levels.
Two other equipment categories that do not respond well to predictive maintenance are electrical equipment and instrumentation components. In many instances, electrical or electronic equipment gives little or no warning of an imminent failure.
Circumstances which may ultimately lead to failure are usually present beforehand, but may defy detection by predictive means. Examples include: Corrosion of electrical connections and contacts, deterioration of insulation, the presence of moisture, loose connections, overheating, poor initial design or construction, exceeding design conditions, momentary overload, or combinations of all of these.
In equipment that is not normally energized, such as shut-down systems or automatic start equipment, a component may be inoperative, but the condition may go undetected until the component is required to operate.
It is unrealistic to expect that all failures can be avoided. But with a well-developed preventive maintenance program that is faithfully implemented and managed, much higher reliability will be achieved.
Unfortunately, many organizations are still skeptical Of preventive electrical maintenance, while others make only token efforts. Effectiveness audits by the authors show that even though a preventive maintenance program was in place, little or no preventive effort was carried through to completion.
CENTRIFUGAL PUMPS
Predictive vibration-monitoring techniques for the determination of bearing defects, rotor imbalance, shaft misalignment, and other defects have proven effective for centrifugal pumps. But relying on predictive techniques alone is the same as waiting for defects to announce themselves instead of preventing defects from developing.
Although there is evidence that periodic oil changes will significantly reduce bearing failure events, not many process plants practice oil replacement intervals as recommended in Fig. 3.3
Failure to change the lube oil periodically in typical centrifugal pumps allows water, corrosion products, and other contaminants to contact various bearing components, thus causing accelerated wear. In some cases, lube oil is replaced only when pumps are taken to the repair shop. And often, many pumps require repair simply because the lube oil was not changed in accordance with proper practices.
In a recent study, the authors reviewed bottom-line maintenance and repair costs for pumps using properly applied conventional lubrication techniques and recommended oil-change intervals. These were found to compare quite favorably with oilmist lubrication techniques.
Of course, oil mist lubrication looked favorable when compared with conventional lubrication techniques when conventional techniques were not applied correctly, or when there were no periodic oil-change intervals.
ELECTRIC MOTOR BEARINGS
The two most prevalent reasons for electric motorbearing failures are overgreasing and inadvertent mixing of lithium-soap grease and polyurea synthetic grease. Lithium-soap grease is commonly used in 85% of U.S. process plants, and polyurea synthetic grease is supplied in an estimated 85% of electric motor bearings in the U.S., so the potential to mix incompatible greases is high.
A large number of electric-motor bearings are prone to failure due to over greasing because the shielded side of the bearing is installed away from the grease cavity instead of adjacent to the grease cavity (Fig. 4). This allows excess grease to enter the element cage.
Others fail because regreasing is done without removing the drain plug. The drain hole expels spent or excess grease during regreasing, thus removing old grease and preventing overgreasing.
To avoid grease compatibility and overgreasing problems, some engineers have indicated a preference for lifetime-lubricated electric motor bearings. But manufacturers' bearing life expectancy tables show that lifetime-lubricated bearings rank at the bottom.
Conscientiously applied motor bearing lubrication programs in a U.S. petrochemical plant and a Middle Eastern refinery, where regreasing of motor bearings with the correct lubricant was required twice a year, were compared to a refinery that practiced only occasional, and probably incorrect, periodic lubrication (Table 2) .5
Note that only 18 electric motor bearing replacements were required per 1,000 motors where a proper lubrication program was applied. In contrast, 156 motor bearings per 1,000 motors required replacement where only occasional lubrication was practiced.
IMPLEMENTING A PM PROGRAM
Important keys to implementing an effective preventive-maintenance program are the participation of all appropriate personnel, representation of all job functions, and provision for accounting for and reporting of results.
Some well-intended preventive maintenance programs are often doomed to failure because of the manner in which they are originated, developed, structured, implemented, or supported. The relevance of these programs, perhaps, has not been communicated to, or input may not have been solicited from, all parties affected by the program.
Personnel will more often support a program if they and their peers have participated in the program's development. Conversely, if key personnel are not involved in the planning stage, support can be marginal.
The package approach, where one group or person develops the entire program should be discouraged.
Instead, a development team should be used, composed of technical, operations, and maintenance personnel. Each member has critical input that can resolve a variety of problems facing the program. Ideally, the team should consist of:
- Site instrument engineer. An individual to lead the effort who is familiar with both the hardware configuration and with the proper design.
- Site instrument technician. One or two people who have worked in the plant or unit with applicable hands-on experience to guide them.
- Site operating specialist. An experienced operations person familiar with the equipment and one who knows the implications of its operation.
- Nonresident specialist. An experienced specialist from outside the plant who can serve as a source of new ideas, experience, and suggestions. Utilizing a nonresident specialist can avoid reinventing the wheel. The nonresident specialist should serve only as an advisor.
- Management sponsor. Although not a part of the working team, visible management sponsorship is a critical factor in the program becoming a success. The sponsor can help make available additional financial and personnel resources that may be needed to accomplish the goals of the program. A sincere commitment from management to implement the program must be more than mere words or memos. Support has to be visible and can be implemented in numerous ways.
For instance, support can be made visible by asking for regular semi-formal briefing sessions or status presentations.
The PM program objective should reflect which equipment, as a result of its failure, could cause a plant shutdown, severe economic loss, or a safety hazard. Each piece of equipment that falls into these categories should then be reviewed for proper hardware, proper design, proper installation, and its ability to be safely inspected, tested, and repaired.
At the same time, the organization should be reviewed for adequate experience, adequate skills, sufficient manpower, adequate documentation, and adequate financial support.
A potential impediment to the successful implementation of a sound PM program is the sometimes erroneous perception that a maintenance procedure could cause an inadvertent unit or plant shutdown. Because this is a valid concern, it should be addressed during the development stage of any program. There are other items that can affect reliability. These should also be addressed during the development stage.
Some of these are quality and reliability of utilities (particularly instrument air and electricity) and freezing, overheating, dirt, corrosion, general environment, and the presence of toxic or restrictive conditions.
Lists should be prepared, not only of the equipment to be maintained, but also of the frequency and nature of specific work to be undertaken on each item. Any special precautions required, specific approvals necessary, reporting functions, and detailed test procedures and equipment used should also be listed.
One extremely important feature of a program is to have one individual or position clearly accountable for its development and implementation. Another is to have an effective means to present the results and progress.
Only those directly involved in the implementation of the program require access to extensive details. However, the management sponsor and supporting organizations should be routinely presented with key statistics providing feedback on how the program is developing.
In some of the more successful programs, brief status presentations are made to top management on a monthly schedule. This gives visibility to the entire effort.
Most programs must be debugged when first implemented. Therefore, implementing it in a trial area can provide verification of the program's capabilities (and its problems) before it is implemented plant wide. For the trial area, it is prudent to select a more modern portion of the plant, especially one that has good equipment and documentation.
All of the program's features should be implemented in the trial area. In addition to verifying the capabilities of the program, the trial area will provide data that will be valuable in determining the magnitude of the effort required for the whole plant and any program modifications that might be needed.
EFFECTIVENESS SURVEYS
Periodic maintenance-effectiveness surveys are considered a highly suitable means of uncovering areas of vulnerability and areas where bottom-line maintenance cost savings can be realized. These surveys resemble machinery reliability audits which are aimed at identifying factors that can minimize forced machinery outages.
However, maintenance effectiveness surveys are far more comprehensive in both scope and detail than pure machinery audits. And unlike maintenance management studies which concentrate heavily on manpower and organizational matters, a maintenance effectiveness survey goes into the when, how, why, and what to do with various equipment and related hardware.
Surveys should be scheduled at least every 2 years, and should be conducted by personnel whose experience and continuing work exposure gives them access to state-of-the-art techniques which transcend both industry and national boundaries.
Maintenance-effectiveness surveys emphasize practical, implementable steps toward achieving plant-wide reliability and availability to the limit. They are an extremely effective way to identify inappropriate designs, inadequate equipment, poor installation, marginal applications, inadequate documentation, as well as repetitive problem areas.
Maintenance-effectiveness surveys also identify equipment-upgrade opportunities. They have been shown to shift the maintenance emphasis from unplanned to planned work.
BIBLIOGRAPHY
- Allen, J.L, "On-stream Purification of Lube Oil Lowers Plant Operating Expenses," Turbomachinery International, July/August 1989, pp. 34, 35, 46.
- Bloch, H.P., and Geitner, F.K., Machinery Failure Analysis and Troubleshooting, Gulf Publishing Co., Houston, 1983, pp. 196-205.
- Eschmann, Hasbargen, and Weigan, Ball and Roller Bearings, John Wiley & Sons, New York, 1985, p. 237.
- Bloch, H.P., and Rizzo, L.F., "Lubrication Strategies for Electric Motor Bearings in the Petrochemical and Refining Industry," Paper No. MC-84-10. National Petroleum Refiners Association Refinery and Petrochemical Plant Maintenance Conference, San Antonio, Feb. 14-17, 1984.
- Miannay, C.R., "Improve Bearing Life With Oil Mist Lubrication," Hydrocarbon Processing, May 1974, pp. 113-115.
Copyright 1990 Oil & Gas Journal. All Rights Reserved.
