29th December 2017, Kolkata
On 29th December 1978, F. Stanley Nowlan, Howard F. Heap, in their seminal work Reliability Centered Maintenance, revealed the fallacy of the two basic principles adopted by traditional PM (Preventive Maintenance) programs – a concept that started from World War II:
- A strong correlation exists between equipment age and failure rate. Older the equipment higher must be the failure rate.
- Individual component and equipment probability of failure can be determined statistically, and therefore components can be replaced or refurbished prior to failure.
However, the first person to reveal the fallacy was Waddington who conducted his research during World War II on British fighter planes. He found that failure rate of fighter planes always increased immediately upon time-based preventive maintenance, which for the fighter planes was scheduled after every 60 hours of operation or flying time.
By the 1980s, alternatives to traditional Preventive Maintenance (PM) programs began to migrate to the maintenance arena. While computer power first supported interval-based maintenance by specifying failure probabilities, continued advances in the 1990s began to change maintenance practices yet again. The development of affordable microprocessors and increased computer literacy in the workforce made it possible to improve upon interval-based maintenance techniques by distinguishing other equipment failure characteristics like a pattern of randomness exhibited by most failures. These included the precursors of failure, quantified equipment condition, and improved repair scheduling.
The emergence of new maintenance techniques called Condition Monitoring (CdM) or Condition-based Maintenance (CBM) supported the findings of Waddington, Nowlan and Heap.
Subsequently, industry emphasis on CBM increased, and the reliance upon PM decreased. However, CBM should not replace all time-based maintenance. Time-based or interval based maintenance is still appropriate for those failure cases, exhibiting a distinct time-based pattern (generally dominated by wear phenomena) where an abrasive, erosive, or corrosive wear takes place; or when material properties change due to fatigue, embrittlement, or similar processes. In short, PM (Time based or interval based maintenance) is still applicable when a clear correlation between age and functional reliability exists.
While many industrial organizations were expanding PM efforts to nearly all other assets, the airline industry, led by the efforts of Nowlan and Heap, took a different approach and developed a maintenance process based on system functions, the consequence of failure, and failure modes. Their work led to the development of Reliability-Centered Maintenance, first published on 29th December 1978 and sponsored by the Office of the Assistant Secretary of Defense (Manpower, Reserve Affairs, and Logistics). Additional independent studies confirmed their findings.
In 1982 the United States Navy expanded the scope of RCM beyond aircraft and addressed more down-to-earth equipment. These studies noted a difference existed between the perceived and intrinsic design life for the majority of equipment and components. For example, the intrinsic design life of anti-friction bearings is taken to be five years or two years. But as perceived in industries life of anti-friction bearings usually exhibit randomness over a large range. In most cases, bearings exhibit a life which either greatly exceeded the perceived or stated design life or fall short of the stated design life. Clearly in such cases, doing time directed interval-based preventive maintenance is neither effective (initiating unnecessarily forced outage) nor cost-effective.
The process of determining the difference between perceived and intrinsic design life is known as Age Exploration (AE). AE was used by the U.S. Submarine Force in the early 1970s to extend the time between periodic overhauls and to replace time-based tasks with condition-based tasks. The initial program was limited to Fleet Ballistic Missile submarines. The use of AE was expanded continually until it included all submarines, aircraft carriers, other major combatants, and ships of the Military Sealift Command. The Navy stated the requirements of RCM and Condition-based Monitoring as part of the design specifications.
Continual development of relatively affordable test equipment and computerized maintenance management software (CMMS like MIMIC developed by WM Engineering of the University of Manchester) during the1990s till date has made it possible to:
- Determine the actual condition of equipment without relying on traditional techniques which base the probability of failure on age and appearance instead of the actual condition of an equipment or item.
- Track and analyze equipment history as a means of determining failure patterns and life-cycle cost.
RCM has long been accepted by the aircraft industry, the spacecraft industry, the nuclear industry, and the Department of Defense (DoD), but is a relatively new way of approaching maintenance for the majority of facilities outside of these four areas. The benefits of an RCM approach far exceed those of any one type of maintenance program.
Fortunately, RCM was applied in India for a few Indian manufacturing Industries from 1990 onwards with relatively great success. I am particularly happy to have been involved in development and application of RCM in Indian industries, which has continually evolved in terms of techniques and method of application to meet contextual industrial needs.
I am also happy to report that RCM for industrial use has now reached a mature stage of its development, which can be replicated for any manufacturing industry.
I am of the opinion that this maturity would provide the necessary stepping stone to develop Industry 4.0 and develop meaningful IOT applications for manufacturing industries.
Wish RCM a very happy birthday!
Typical Symptoms: High 1x in the axial direction and 2x in the radial directions; at time 3 x is also present in severe cases (e.g. when coupled to coupling imbalance).
Reasons for misalignment:
- Thermal growth
- Movement of foundation
Types of misalignment:
- Parallel misalignment — we would find strong presence of 2x component in radial direction along with 1x in the axial direction. This is because two opposing forces act together at the coupling — both trying to align the shafts to each other.
- Angular misalignment — we would find strong presence of 1x component in the radial direction along with strong 2x in the axial direction. This is because angular misalignment produces a bending moment on both shafts.
- However, vibration patterns don’t change in very predictable patterns as described in points 1 and 2 above. This is because there is usually a mix of the two different types of misalignment. In addition foundation problem and stiffness (directional or variable) create further complexity in the situation.
- The 1x and 2x components would be strong in the radial directions (V and H) but these components would be in phase.
Usually we would find high 1x peak in the axial direction with small 2x and 3x peaks depending on the “linearity” of the vibration. There may be both 1x and 2x (at times accompanied by 3x) in the radial directions.
Time waveform in the axial direction would be dominated by sinusoidal 1x vibration
Phase: Motor and say Pump would be out of phase axially due to angular misalignment (across the coupling in the same direction).
Typical Symptoms: 1x radial (in Vertical and Horizontal directions)
Like eccentric pulleys, Eccentric gears generate strong 1x radial components, especially in the direction parallel to the gear.
They would also generate sidebands of the running speed of the eccentric gear around the GMF (gear mesh frequency). However, harmonics of GMF may also be generated (depends on the severity of the problem). Natural frequency might also be excited.
Time waveform: The waveform will have combination of 1x running speed of input and output shafts plus strong gear mesh vibration modulated by the running speed of the shaft having the eccentric gear.
Phase: Not applicable.
Typical Symptom: High 1x in the direction parallel to belts. Though 1x component can be found on both Vertical and Horizontal directions.
Instead of the typical Vertical and Horizontal directions it is best to choose the directions parallel and perpendicular to the belts.
The high 1x can be found on both sub-assemblies (e.g. the motor and fan). Since the motor and the fan would run at different speeds we would also find two distinct peaks on the signature corresponding to the motor and fan running speeds. Confirmation about which pulley is eccentric can be obtained by removing the belts and checking for the presence of high 1x on motor in the direction parallel to the belts.
Time waveform would be sinusoidal when viewed in velocity.
Phase: Phase reading taken parallel and perpendicular to belts will either be in phase or 180 degrees out of phase.
The inherent reliability of a system is determined by the system’s design. It means that the design of the system would determine the upper limit of reliability the system exhibits during operation. Suppose, for example, a system, with the best possible maintenance is able to achieve availability of say 90% we can say that this is the upper limit of the system’s capability that is determined by its design. A good “preventive maintenance” plan can never improve a systems inherent reliability. In other words, preventive maintenance, contrary to what many believe, cannot make a system “better”. It may, at best, only help realise the inherent reliability as determined by the physical design.
Hence the suggested process to “improve” the inherent reliability of a system, may be framed as follows: –
Understand the dynamics through tools like vibration analysis
Monitor changes and rate of change
Eliminate unnecessary maintenance tasks
Change the design of the system interactions to eliminate inherent “imperfections” and revise the maintenance plan.
In most cases, this would be the general approach.
Until we can effectively undertake some design changes (Design Out Maintenance – DOM) or take measures to eliminate inappropriate maintenance actions (Review of Equipment Maintenance – REM) it would not be possible to go beyond inherent reliability of an equipment, specially if it is undesirable in the business context. For example, a vertical pump of a power plant kept failing very frequently or had had to be stopped quite often when vibration shot beyond the trip limits. This behaviour of the system is determined by the design of the system. Unless the design (specifically the interactions between components) is corrected for improvement; the system (vertical pump) would continue to behave in that manner for all times. Likewise if the MTBF of a machine is say 90 days, it would not be possible to considerably improve the MTBF way beyond 90 days unless some undesirable interactions (which I call system “imperfections”) are corrected for improvement and a proper review of existing maintenance system is carried out. Such “imperfections” can be both physical and non-physical. Design features, most importantly, the interactions between physical/non-physical components are arguably the most important characteristic of a system that determine a system’s inherent reliability.
In addition, there are many physical design features that influence reliability like redundancy, component selection and the overall integration of various pieces of the system.
In the context of RCM, design extends far beyond the physical makeup of the system. There are a number of non-physical design features that can affect, sometimes profoundly, the inherent reliability of a system. Among these are operating procedures, errors in manufacturing, training and technical documentation. When a proper RCM analysis is conducted on a system or sub-system, there’s a good chance that the resulting maintenance actions will enable the system to achieve its inherent reliability as determined by its physical design features. However, if the inherent reliability is below user’s expectation or need then the design features are to be improved to achieve the desired level of inherent reliability.
Moreover, if unwarranted maintenance tasks are eliminated as it will greatly reduce the risk of suffering the Waddington Effect. There is also a good chance that if operating procedures, training, technical documentation and so forth are found to negatively impact inherent reliability, these issues will be identified and corrected. As evidenced by the Waddington Effect. In virtually every case, less than optimal, non-physical design features almost always have a negative impact on inherent reliability. Therefore, in RCM analysis a through review of existing maintenance plan (REM) along with DOM is necessary to improve inherent reliability of a system.
In brief, right amount of Condition Based Maintenance (CBM) tasks, Scheduled Inspections (which is a part of CBM activity) REM and DOM would not only help us realise the inherent reliability as determined by the physical design but also improve it, if the original inherent reliability is below business expectation.
General Symptom: 2Lf (Lf = Line frequency)
Stator problems would create high vibration at 2Lf. Stator eccentricity produces uneven stationary air gap between the rotor and stator that produces a very directional source of vibration.
Soft foot is often the cause of eccentric stator.
Other key indicators:
- 2Lf peak would be comparably high
- For a 2 pole motor this peak would be close to 2N (N= running speed). Would need sufficient resolution to separate them
- A spectrum may reveal beating — 2Lf and 2N peaks may appear to rise and fall if we don’t have sufficient resolution to separate them.
- Time waveform — a combination of 2N and 2Lf would reveal a beat type pattern if the time period covers more than a few seconds. If the time period isn’t long enough, then we would see a wobble or take on the classic M or W shapes due to combination of 1N, 2N and 2Lf.
- Thermal images would reveal heat bands in the direction perpendicular to the direction of high vibration
- Vibration would be highest at the point where the stator is closest to the rotor. Move the accelerometer around the motor housing to see if the peak is high in one or two locations.
Symptom: Pole pass sidebands around 1x N (N=running speed) and 2xLf (Lf = line frequency)
Eccentric rotors will produce a rotating variable air gap between the rotor and the stator which induces a pulsating source of vibration. We would see 2xLf. However, there will also be pole pass sidebands around the 2xLv and 1xN peaks. 1xN is expected to be high.
Note: Pole pass frequency is the slip frequency times the number of poles. The slip frequency is the difference (in terms of frequency) between the actual RPM and the synchronous speed.
Presence of pole pass sidebands around 1N and 2Lf is the key indicator of this fault. One needs sufficient resolution to see those sidebands. Else we would either miss them altogether or mistake them for resonance (a broadening of the base of the peak).
Waveform: Time waveform that covers many seconds of time will reveal the pole pass frequency modulation. Due to lack of impacting the waveform will smooth and will be a combination of the 1N and 2Lf frequencies of vibration.
Phase: Not applicable for this fault unless eccentric forces are high in magnitude.
General symptom: 1x radial (Vertical and Horizontal direction of horizontal machines)
Usually a rotor bow in a motor looks like a static imbalance. Broken bars and loose connections (at motor terminals and at MCC) cause motors to heat up (localized) owing to uneven current flow through the phases causing rotor bow — uneven weight distribution around the rotor’s centreline. Hence we see high amplitude peak at 1x running speed in the radial and horizontal directions.
Localized overheating can be seen on the motor body through infrared thermal imaging.
The effect of can also be seen on the rotating air gap — a high peak at 2xLf with pole pass sidebands around 1x and 2x peaks. The 2x peak often comes up when the effect is more severe.
The time waveform would be sinusoidal when viewed in velocity.
Phase: expect 90 degree shift between vertical and horizontal axes. The inner race will move in and out once per revolution with a bent shaft
In a spectrum, if the entire noise floor is raised, it is possible that we have a situation of extreme bearing wear.
If the noise is biased towards the higher frequencies in the spectrum then we may have process or flow problem like possible cavitation, which may be further confirmed by high acceleration measurement (or filtered acceleration measurement) on the pump body on the delivery side (since high frequency waves are always localized).
Smaller “humps” may be due to resonance (possibly excited by anti-friction bearing damage, cavitation, looseness, rubs or impacts) or closely spaced sidebands arising from other defects. A high resolution measurement (or graphical zoom and a log scale) may reveal whether the source is problems that exhibit sidebands or a problem of resonance. If machine speed can be changed, (for e.g.motor connected to VFD drives) the resonant frequency would not move – but the other peaks would. Sidebands will typically be symmetrical around a dominant peak – e.g. 1X, 2X, 2x LF (100 or 120 Hz) etc indicating different faults.
Interestingly, the time waveform would reveal the reason as to why the noise floor has been raised.
We would see signs of looseness, severe bearing wear, rubs, and other sources of impacts in the time waveform. We must make sure that there are 5 – 10 seconds of time waveform if we suspect an intermittent rub (e.g. white metal bearings of vertical pumps or loose electrical connection of motor terminals) or if we suspect flow turbulence or cavitation.
If the time waveform looks normal (making sure there is a high Fmax (following Niquist criteria) and we view the waveform in units of acceleration then increase the resolution in the spectrum to 3200 lines or higher in case we are seeing a family of sidebands (like the sidebands we find around gear mesh frequency or rotor bars).
But if a natural frequency is being excited (necessary condition for resonance) then we have to perform a bump/impact test or a run-up/coast down test to confirm the situation.