Tool frequency change - Data Analysis needed

[nagumuthu]

Hi,

I am using 10 tool at 10 locations [ each location one tool ] in a machine. These tools fail at different period. One tool fails after 1 month. another tool fails after 2 months.

I need to arrive at a frequency for changing this tool at definite interval.

Whenever I change the tool, I have recorded the date of changing this tool.

Can anyone help me to advice, which method I can use, in Minitab, to arrive at the frequency of change.


For your information, above case is in one machine. I have 5 machines facing same situation as above.

Your help and advise is greatly appreciated.

Best Wishes
MM

 

[kuyakut]

Are they running at the same brand and model of machine on each location?
Do they run at same production loading - ie 24 hours run, they go for same frequency of mantenance?

 

[nagumuthu]

Yes. All the machines are in production 24 X 7.

 

[nagumuthu]

Brand,model, maintenance all the same for all machines

 

[Alpine]

Is the material the same? some materials cause more wear than others. Do all the jobs use this tool? Dependent on the job loading on each machine, you will get variation on wear as well.

 

[Miner]

This will be a type of reliability analysis. You will want to analyze this for each unique tool/application.

 

[nagumuthu]

Tool Material is the same.

Each tool machine same jobs. Only location on machine is different.

 

[nagumuthu]

All your threads hint me to think about MTBF mean time between failure.

How to use mini tab for MTBF

 

[Miner]

All your threads hint me to think about MTBF mean time between failure.

How to use mini tab for MTBF

No MTBF! Not now, not ever!

Seriously, MTBF is a very flawed metric. For your application, I recommend using a metric like L5 or L10. This is the time (or cycles or ops) at which 5 or 10% of the tools would fail.

In Minitab, this would be the parametric distribution analysis.

 

[bobdoering]

I recommend using X hi/lo-R charting methodology on the dimensions the tools are used on. It is the only method I am aware of that uses the data of each tool in conjunction with the specific lot of material it is being used on (rather than an estimate from other tools and other lots - fraught with sampling error) to determine if the tool needs changes before it breaks.

Since tool wear is a continuum from cutting to rubbing, as the tool wears more rubbing starts to occur. This transfers the vibration of the machine that causes lobing to increase the lobing effect. This will be seen as an increase in roundness error seen in the R chart. From past experience, a level of roundness can be used as a "control limit", whose reaction plan is to replace the tool. Generally, there is a dramatic increase in the roundness error (non-linear) that is a clue you are no longer cutting cleanly. Exceeding the control limit will cause the tool to enter a range of rubbing that will overheat the tool until it is destroyed.