Hi, I'm doing a small case study about preventive maintenance. Suppose the failure modes of a machine follow weibull distribution,the preventive maintenance is carried out based on the failure rate threshold.
So I did a very small case (two machines, one buffer). Attached is one unfinished model, with basic failure mode configured.
There are two failure modes in the model (see method failuresetup), failure 1 is the failure mode after repair, i.e., the failure mode after one failure is failure 1; and failure 2 is the failure mode after preventive maintenance, i.e., the failure mode after one preventive maintenance is failure 2.
Who can help me to realize the preventive maintenance of both machines?
Solved! Go to Solution.
only some remarks about your approach:
The basic idea of preventive maintenance is to change from stochastic failures partially to controlled downtimes. So you perform e.g. once per day, once per week maintenance to prevent failure-events during the operational time. You can find a lot of literature about it.
To model this means: add an additional failure profile to the machines with regular intervals and constant MTTR (e.g. one hour maintenance with a distance of 24 hours) and increase the availability of the "standard" failure profile. Normally you have to change the working-times of the machines, so that the planned downtimes are happens during the operational times of the machines (also on weekends).
I was trying to model the preventive mantenance based on the failure rate of a machine, say if the failure rate of a machine rised to a threshold value, the maintenance was carried out. According to this rule, the machine's MTTF follows a weibull distribution, and the threshold value equals to 2*beta (parameters of the weibull distribution). Two failure profiles are input into the model.
Here is what I did: I active failure1 and deactive failure 2 at the initial stage, then monitor the real time after MUout events. If the time this last fail/maintenance is greater than a value corespongding to the threshold, the machine begins a pause event (representing the maintenance event), deactive failure profile 1 and active failure profile 2, at the same time, erasefailureevent of the machine. Every time the machine is failed, failure profile 1 is activated and the other is deactived, if the last failure is caused by failure profile 2, erase failure events, otherwise, I don't change the failure profile's active state.
I encountered a problem when I was trying to simulate this: after the simulation runs for a while, the protion of pause event "eats up" all the portion of the failure events, why would this happen? Is it because the erasefailureevents method call?
Pause and failed can overlap each other see help:
When a station fails, and a pause takes place during the failure, the failure time of the station will still be consumed, although the station is paused.
I realized that. So in my original model, before each pause, one command line "?.endfailure;" is called. But the statistics overlapping is not known until now. Thanks.
Is there a detailed discreption of the different states of machine? Sometimes I really feel confused about different machine states and the statistics of them.
Thanks for your help.
If I understood correctly, you suggested to configure preventive maintenance as a failure, which means it will be charted as failed (red).
Is there a way to chart preventive maintenance as planned (or paused - dark blue)?
A workaround is to configure it as a shift.
The problem is we can define daily or weekly tasks, but not monthly or yearly ones.
to be correct you model the repair-times, if an error accours (red LED). Maintenance you can model in the same way as an additional failure profile. In the failure control you can read the name of the failure profile for creating a more detailed statistics.