asset maintenance

In history, the evolution of maintenance has a few stages going from Reactive maintenance to Preventive maintenance to Predictive maintenence.

1. Mostly reactive fix when broken, 

back in the old time 1900s, machines were slow and over-engineered, and it's fixed when broken

2. mostly overhaul, 

about world war2 time, a lot of pressure on the factories to keep things going. reliability is so important.

the philosophy changed from "fix when broken" to "prevent it before broken"

overhaul is the common practice, so issue was removed before failure

3. mostly condition based, fix "on condition"

1960s, even more pressure on the quality management

Monitory the conditions of machines and fix issue based on condition.

4. proactive maintennance, condition and risk based

get rid of any potential fault we could have, rather than monitoring faults that we do have.

Preventive maintenance

The aim is to reduce breakdowns through: interval maintainence, operational/visual/automated check, servicing, overhaul/restore/discard if needed

Above is an example of asset's probability of failure.

In the begining of life, the probability is low. When coming close to it's designed life span, the probability increases dramatically. that is entering the wear-out zone. Preventive maintenance typically kicks in when the failure probability is higher than our comfort zone.

However, this approach requires the Failture statistics available, which may be doable based on previous failure data. 

The preventive maintenance will follow the failure profile, and assuming all equipments follow the same probability distribution, which may not be true.

Other possible failture probability distributions

Here is a list of possible failture profiles based on a investigation for the airline industry decades ago.

The left 3 are called AGE-weighted profiles, because they all get higher probability of failture as aging at the end.

The right 3 are called Random profiles, because the probability doesn't increase after stablizing.

Back then,  an investigation into United Airlines maintenace unveiled that 89% of faults are random profiles. More often maintenance didn't help but made reliability worse. Every time a maintence is done, it reset the conditions back to the start, so for 'Initial break in' and 'Infrant mortality' profiles, the failure probability goes up.

Two important discoveries:

1. Scheduled overhaul has little effect on the overall reliability of a complex item unless it has a dominant failure mode.

simple item's failture can be relatively easily predicted based on e.g. the 'wear out' or 'fatigue' profiles, but for complex items (e.g. a pump, hydraulic) there are a mix of different failture modes and schedule overhaul helps little.

2. There are many items where there is no effective form of scheduled maintenance.

e.g. in the airline industry , 89% of equipment has no 'end of life', then no need for a scheduled maintenance at the end of it's life to revive.

Avoid unneeded maintenance - restarting infrant mortality failtures

Monitor and condition of processes and equipment to detect and prevent onset of failure.

Preditive maintenance

Many ways:

Visual inspection

vibration monitoring

ultrasonic testing

lubricant analysis

temperature - termography monitoring

condition monitoring

non destructive testing

Most importantly: 

  don't forget abou the operator! Operator should care and monitor the conditions. Through look listen, feel and smell, operator can alreay detects a large proportion of defects.

to continue: