Several MTBF/MTTR for 1 processor ?

HI All..

I tried develop simple model, but i have problem in it.
I have 1 processor which has 2 kinds of maintenance.
Both of them use processing time as a reference.

Suppose:
Maintenance A: MTBF=5(constant) MTTR=5(constant)
Maintenance B: MTBF=8(constant) MTTR=6(constant)

In logic, it should be:
time 0-5 (processing)=>time 5-10 (Maintenance A)=>Time 10-13(processing)=> time 13-19(Maintenance B)

To deal with this, i combined "Experimenter" and "MTBF MTTR"
But the result, not as in logic. I think, i should set something in my model, but i don't know where..

Need help, to solve this problem
Thanks in advance
Ankus

Maybe I am wrong with my idea, but why not using a MultiProcessor with four processes? Give the processes as process time a uniform distribution. The different state of the multiprocessor then shows than some statistics about the different processes.

I haven't looked into the model but why not use 2 mtbf's 1 for the first maintenance period and the other for the second period and then select that they only work on the state 'processing'
The first has firstmtbf of 5 and mtbf of 19
The second has firstmtbf of 13 and mtbf of 19

As far as I can tell everything is constant so there should be no probl
em there.

Steven

Hi Tom..

If we treat like that the statistic should be:
Suppose, run time =19
Processing1 (5, 26.31%)
Maintenance1 (5, 26.31%)
Processing2 (3, 15.79%)
Maintenance2 (6, 31.58%)

but your model:
Processing1 (4.5, 23.9%)
Maintenance1 (6.6, 34.8%)
Processing2 (5, 26.1%)
Maintenance2 (2.9, 15.3%)

Moreover, if i have many processor and if each processor has 10-15 kinds of maintenance, it could be confusing if i build like your, since i should add many processes in Multiprocessor

Originally Posted by Steven Hamoen

I haven't looked into the model but why not use 2 mtbf's 1 for the first maintenance period and the other for the second period and then select that they only work on the state 'processing'
The first has firstmtbf of 5 and mtbf of 19
The second has firstmtbf of 13 and mtbf of 19

As far as I can tell everything is constant so there should be no probl
em there.

Steven

Yes, i used 2 mtbf's in my model.

If i have 10-15 kinds of maintenance, and i don't want to calculate firstmtbf like that, and assume we don't want arrange our MTBF first.

I still don't know....

Ankus,

Did you have a look into my model or read my post?
I wrote that I used uniform distribution in my model for the process times, because that’s what I understood. Anyway, if you change these to constant times (by expression) then you get exactly the statistics you are looking for.
I agree that using the Multiprocessor might be some effort to build the model if you have lots of maintenance processes. But at least it seems to be a possible solution.
I tried to help you and give you some hints and ideas. Sometimes they are not a solution for the given task, but hopefully still helpful, because they may show a direction which is not a solution and therefore not the way to go.

I hope you will find a solution for your task and if you are successful, please let us know how you did it.

Originally Posted by Tom David

Ankus,

Did you have a look into my model or read my post?
I wrote that I used uniform distribution in my model for the process times, because that’s what I understood. Anyway, if you change these to constant times (by expression) then you get exactly the statistics you are looking for.
---
I hope you will find a solution for your task and if you are successful, please let us know how you did it.

Yes, I did
Even we change into constant times, the result still different.

I think the problem is that the next maintenance depend on the previous maintenance.

If we have 2 and the run time is short, we can consider as Steven Hamoen said. But if we have several maintenance, it's become problem since we don't want to arrange our schedule first in determining first mtbf for next maintenance.

SO, i am still trying...

What I find very strange is, that you do not get the exact expected times if you switch in my little model to constant times. I let one product in the system and it gives the exect expected times (5,5,3,6). I attached the model to show the times (one product arrives at time 0 and a second one at 60 and the statistics makes absolute sense in my eyes.)
Maybe if this is not solved, that’s why you get strange results.

Anyway, you said that my direction is not the right way to solve your problem and also Steven’s way seems not to be a nice solution. So I guess you need to find another way and maybe someone else has some hints for you.
I wish you good success.

Originally Posted by Tom David

So I guess you need to find another way and maybe someone else has some hints for you.
I wish you good success.

Thank you Tom, for your attention.
I am waiting for the others idea..

I am still stuck on this problem ...

Any ideas..??

Ankus,

What exactly are you looking for? Both Tom and Steven gave excelent examples for solving the problem but it hasn't seemed to answer your question. The only conclusion that I can draw is that you need to be more specific about your question and about why the answers given don't work. I think that this would allow people to be more capable of helping out.

Good Luck,
Brandon

Thanks Brandon for your response..

Actually I have 5 processor...and for each processor has 10 different kinds of maintenance.

I want to schedule maintenance for those processor.
So in this case i want to evaluate the impact of maintenance schedule...

To deal with, i collect all mttb/mttr for different maintenance task of processor..

I am trying to automate model so that it can read logic of mttb/mttr, without arrange manually.

So in this case, is it possible if we just input exact mttb/mttr for those processor?? since in the next process i will also update this mttb/mttr automatically

Ankus,

Is it really a schedule that you are after? By schedule I mean that the down times always occur in a distinct sequence at specific times? Or is it that you have 10 different randomly occuring breakdowns that have no effect on each other?

Hi Ankus,

I attache a model, which uses three different constant values for mtbf. The change is controlled by a global variable, that is incrementated with every mtbf event. The model behaves similar to the first model by Tom.

Jörg

PS: I change the attachment file

Hi all...
Finally i found trick for my problem...
I added simple code in fistmtbf, mtbf, mttr and resume function.

so, mtbf will dynamically change, but the actual mtbf still the same as reference.

Thank you all, i appreciate your response and comment, it's make me feel "I am not alone"..and trigger some ideas for me..

Hi Ankus

Could please you submit the last attachment again?

Thank you