Determine inspection intervals 2

[netherhay]

I have a mechanical component with a know MTBF, how can I determine what the inspection interval should be to give 90% and 95% confidence intervals to ensure a 98% availability?

Is this the same for systems and individual components?

Similar advice regarding electrical compenents/systems would be appreciated.

 

[IRstuff]

Is there a difference in MTTR between a scheduled replacement vs. unscheduled?

What is the actual variance in failure rate?

Does your inspection guarantee identification of a potential failure?

TTFN

 

[netherhay]

There is no difference in the repair times between scheduled and unscheduled replacement of failed component.

As I am looking at 'new to service' equipment, I only have theoretical figures to go on; hence the reliance on NPRD to derive MTBF values - unfortunately I have no actual data.

We must assume that inspection will identify potential failure. I don't see how we can specify any intervals unless we assume this - the only alternative would be to inspect before every use which is not viable in this case.

 

[sdl]

Netherhay:

There are a couple of things to look at here:

1. Availability is calculated by MTTF/MTTF+MTTR). For this you will need to know what the MTTR values the manufacturer (or the source for your data) has used. Usually a larger valve would be roughly 20-24 hours. This should also include the time from when the failure is noticed until the process is back online.

2. Test intervals are usually determined by calculating the probability of failure on demand. Test intervals are used to gain "credit" in the calculation, i.e. the more often you test, the more chances you have of finding covert failures.

My assumptions are based on your application being safety related. You could employ the same methods and calculations to non safety applications. Reliability and Availability are two different metrics.

 

[Drazen]

I would add to fine sdl summary - you should find yourself Mean Time To Repair for the most important components, no matter whether you have manufacturers data or not: local conditions, organizational features, skill level of your employees must be taken into account by proper maintenance pre-planning. Of course, you should leave door open to furhter adjustments of any assumption for MTTR, even if you will have to recalculate your benchmarks several years later - maintenance is very complex activity and it is allways necessary to combine educated guess with field feedback.

Regarding inspection interval, I remember basic rule of thumb which states that inspection interval is (Tinc+MTTR)/2, where "Tinc" is time period from first notable sign of failure to occurence of failure. This gives you opportunity to note incipient failure sign and do necessary maintenance activity on time even if such sign occurs one second after your previous inspection (worst case scenario).