MTBF can't breakdown individually

I built model and run it already getting the results I want, but when I started to add MTBF problems started facing me


I started by adding MTBF for one machine only (machine 1) in my model Also I got the result i want which was common sense with total breakdown around 2.2 hours from total of 24 hours, since the MTBF was expo (0, 2.07 hrs,1 ) and MTTR is Uniform ( 8,20 mins,1) Till here everything was perfect

But when I added 3 more members in the same MTBF Memebers tab in my model (machine 2,3 & 4 ) with the same values of MTBF and MTTR and from the breakdowns tab checked the ( break down members individually ) it started breaking them alone but all the 4 members every MTBF resulting of total breakdowns around 10 hrs, i checked this from the state bar

What I want is to let the model breakdown 1 of the 4 machines randomly alone each MTBF, one and only one not all together each MTBF , please can you help in solving this issue as soon as possible and guide me how can I do it ? because I have more than one model and I want to apply it for all the models

regards
Hassan Elhajj

From this and your other post on the forum, I think it might be worth saying that the MTBF/MTTR is a profile characterizing the unplanned nature of a loss per machine/member. It does not define an overall system/group loss.

This sounds like a problem of using aggregated summary data to model individual machine losses. So if you're saying that you want to lose 2.2 hours across 4 machines in 24 hours, rather than per machine, then you need to multiply the TBF frequency for the system by 4 or divide the repair time for the system by 4 - whichever seems more sensible to you - or come up with non-aggregated values.

The statement that you had one machine behaving 'perfectly' might be misleading if in fact it had 4 times the loss that it should.

Originally Posted by Jason Lightfoot

From this and your other post on the forum, I think it might be worth saying that the MTBF/MTTR is a profile characterizing the unplanned nature of a loss per machine/member. It does not define an overall system/group loss.

This sounds like a problem of using aggregated summary data to model individual machine losses. So if you're saying that you want to lose 2.2 hours across 4 machines in 24 hours, rather than per machine, then you need to multiply the TBF frequency for the system by 4 or divide the repair time for the system by 4 - whichever seems more sensible to you - or come up with non-aggregated values.

The statement that you had one machine behaving 'perfectly' might be misleading if in fact it had 4 times the loss that it should.

well jason It seems you didn't get my point I want the MTBF to be 2.07 hrs but i want to apply it for the 4 machines that each time it brakes one machine randomly,
even if i multiply the TBF by four it will fail all the machines at once or the whole 4 machines at the same time.
I need to have one break at a time from one of the four machines. NOT all Together Neither ALL ALONE, I need ONE failure random from the 4 machines

statement that you had one machine behaving 'perfectly' might be misleading if in fact it had 4 times the loss that it should.


why if in 24 hours i have around 10 failures ( 24hrs/2.2 = 10.9 )(MTBF expo2.2 hrs)
then 10 failures will get breakdown time with
minimum 8*10 = 80 mins = 1.3 hrs
Maximum 20*10= 200 mins = 3.3 hrs

which is the interval of fixing all the breakdowns generated in excel (breakdown 2hrs and some mins ) is in between this range.

so why I had 4 times the loss

There is an option to break members individually on the MTBF, which I thought had been mentioned in the other post. You're right I didn't find it obvious from either of your posts what you wanted - which I why I tried to clarify things.

What you described as the desired outcome seemed to contradict to me - so yes - sorry if I've misunderstood.

no worries but i have to know how to let the model pick one machine at a time for the failure not all the for
do you know who may get me the answer ?




regards
Hassan elhajj

What you're describing is the default option when you create a new MTBF/MTTR. If you open the event list at reset you should see the breakdown events for all 4 processors at different times.

Try creating a new one or post your model and we'll take a look.

Attached is a model to show how each processor is around 10% (2.4 hours) in breakdown.

Dear Jason

please find attached model
again as an explanation of what i want is to breakdown one machine every MTBF randomly
not all the 4 at a time, nor the 4 every MTBF,
Only one machine Every MTBF between those machines




regards,
Hassan El-hajj

In your model the machines break down individually. Look for the state in the quick properties tab, and for the yellow square drawn at the base of the object to see when they are in breakdown.

The other machines stop because they are blocked. If it's a tightly coupled machine group (like the monoblock) - then they should all stop. If they're not tightly coupled you can uncouple them and stop them being blocked by providing some buffers or accumulation between the machines.

Originally Posted by Jason Lightfoot

In your model the machines break down individually. Look for the state in the quick properties tab, and for the yellow square drawn at the base of the object to see when they are in breakdown.

The other machines stop because they are blocked. If it's a tightly coupled machine group (like the monoblock) - then they should all stop. If they're not tightly coupled you can uncouple them and stop them being blocked by providing some buffers or accumulation between the machines.

regarding the machines I know that they break down individually but what happens that all the 4 machines breakdown after each other Every MTBF
Again to Explain What I want is to breakdown (ONE) machine Each (MTBF) not the
4 machines Break-downing each MTBF

Regarding that the machines will be blocked, I know that and actually I want it to be blocked when any machine is in a breakdown

Hello Hassan,

probably it isn't clear what you want. Do you want, that a group of four machine act as one machine when it comes to a breakdown. If a machine breaks down, the nex break down in the group occurs after the mean time before failure. if that is what you want to model, you can add just a new object which initiate the breakdowns and the repairs. In the trigger On Breakdown you dice which processor breaks. You can build a global table which contains the pointers to the processors as numbers: tonum(node("[path to the processor in the model tree]",model())). Then you choose a row by a duniform distribution. This processor stops. You store the selected row or the stopped processor in a label, because you use this data to resume the right processor on the On Repair Trigger.
If you use a MTBF MTTR and add the processors individually to the tool or you build different tools for each processors, then the events happen independently.
Jörg

instead of using a global table to store the pointers, take a look at the global variables and a treenode array