Reliability Engineering for Maintenance

Hi guys,
I would like to ask about using reliability to determine maintenance interval.

Assume that I have a number of one failure mode of a set of identical equipments over a period of time.

For example hot electrode failure on spot welding guns over a three year is 19 times. Can I calculate the reliability for those guns and use it to determine maintenance interval. Such as if I accept the reliability is 95% (only 5% of the guns will fail over the period t) then I calculate the period t is 147 days which mean I would carry out maintenance activities at the 147 days interval?

Is it ok to make an assumption that the failure rate is constant, since what I have is the interval censored data which only show the failure over a period of time. There are no info about the time to failure so that I could use Weibull.

Thanks in advance

May we know what is the failure mode? Is it a uniform degradation over time?

Can you tabulate the failures with their failure intervals?

If you are an industrial trainee from a uni, perhaps refer to your lecturer again about this.

Here is a writeup in the internet regarding the limitation of MTBF:
http://www.maintenancetechnolo...-frequency-is-wrong/

Perhaps you want to read this article if relevant:

http://www.weibull.com/hotwire...113/hottopics113.htm

Hi,
Thanks for response, I already read these articles. In my situation, I have no cost record related to PM action, what I have is a data set like interval censored data (as described here http://reliawiki.org/index.php..._Data_Classification
I only know how many failure appear and theirs time over a specific period. Outside the period I have no information.

The failure mode here would be bad welding point caused by hot electrode at the resistance spot welding gun. This can be fix by cleaning the cooling pipe which used to lower the temperature at the electrode. What I need now is to determine how often should I clean the pipe. Based on limited data, what I can think is to use reliability to have a rough estimation. Based on how many percent of equipment fail would be acceptable. For example if I accept 5% of equipment will fail (reliability 95%) then the estimated period for maximum 5% fail is T, then I will clean the pipe every T time