Hi,

I am intending to implement attribute control chart in my process but my Master Black Belt says that it's not appropriate because I still allow defect to be produced in my process.

My intention of implementing attribute control chart is to capture the defect first and prevent it from escaping to customer. At the same time, when a defect is detected, we can stop the production and ask the process owner to solve the issue completely before we can continue production. Most of the time, defect is undetected and flow to our customer.

Is my intention correct? Any comments on my MBB's remark?

Suggestions are greatly appreciated.

Regards.

Hi,

I am not sure whether it is good to use an attribute control chart, since you can not see developements in your measured values. You either get a go or nogo. If you get a nogo you have to look at the process immediately. So the consequence would be to measure 100 % of the parts or if you measure less, to sort (measure) all parts between the last and the current measurement. You might have a big number of nonconforming parts.

But you can read about attribute control charts in the AIAG manual for SPC.

Regards,
Kira_n

Hi,

Thanks for the reply. Actually, most of our defects are related to cosmetic and performing 100% sorting of parts on that hour's of production will be more appropriate because we can't afford to sort through 100% at the end of the whole production.

Do you have the link to the AIAG manual for SPC? Appreciate if you can provide the link.

Regards.

Hi,

I am not sure whether it is good to use an attribute control chart, since you can not see developements in your measured values. You either get a go or nogo. If you get a nogo you have to look at the process immediately. So the consequence would be to measure 100 % of the parts or if you measure less, to sort (measure) all parts between the last and the current measurement. You might have a big number of nonconforming parts.

But you can read about attribute control charts in the AIAG manual for SPC.

Regards,
Kira_n

Hi,

I am intending to implement attribute control chart in my process but my Master Black Belt says that it's not appropriate because I still allow defect to be produced in my process.

My intention of implementing attribute control chart is to capture the defect first and prevent it from escaping to customer. At the same time, when a defect is detected, we can stop the production and ask the process owner to solve the issue completely before we can continue production. Most of the time, defect is undetected and flow to our customer.

Is my intention correct? Any comments on my MBB's remark?

Suggestions are greatly appreciated.

Regards.
Your intentions are correct.Go ahead with Attribute Control chart.But brfore that train operators who are filling the charts on machine.I presume you alreahy have the excel work sheet for attribute control chart
V.J.Brahmaiah

Hi,

I am intending to implement attribute control chart in my process but my Master Black Belt says that it's not appropriate because I still allow defect to be produced in my process.

My intention of implementing attribute control chart is to capture the defect first and prevent it from escaping to customer. At the same time, when a defect is detected, we can stop the production and ask the process owner to solve the issue completely before we can continue production. Most of the time, defect is undetected and flow to our customer.

Is my intention correct? Any comments on my MBB's remark?

Suggestions are greatly appreciated.

Regards.

Before you prevent defects you must understand what type of defects your process may produce. Ideally this would be done when the process is designed, through FMEA, so that problems can be prevented by engineering solutions.
If, however, your process is already in place, attribute charts can be useful to record the type and quantity of defects found in order to target improvement. They can also provide a baseline to measure any improvement from actions taken.
Ask your MBB, why would you not measure your process?

Hi,

Thanks for your reply.

Regards.


Before you prevent defects you must understand what type of defects your process may produce. Ideally this would be done when the process is designed, through FMEA, so that problems can be prevented by engineering solutions.
If, however, your process is already in place, attribute charts can be useful to record the type and quantity of defects found in order to target improvement. They can also provide a baseline to measure any improvement from actions taken.
Ask your MBB, why would you not measure your process?

You are welcome.

Hi,

Thanks for the reply. Actually, most of our defects are related to cosmetic and performing 100% sorting of parts on that hour's of production will be more appropriate because we can't afford to sort through 100% at the end of the whole production.

Do you have the link to the AIAG manual for SPC? Appreciate if you can provide the link.

Regards.
What you are looking for and what an attribute control chart will provide are different.

An attribute control chart will only distinguish between the level of defects/defectives that is common to the system and that level of defects/defectives that is due to special causes. Your operators can only respond to those special causes by investigating and correcting. The level of defects common to the system is the responsibility of management to change the system that is generating the defects.

You will only be fooling yourself if you think that zero cosmetic defects were produced between two successive checks.

The AIAG SPC manual is available for purchase through www . AIAG . org. It is copyrighted and cannot be posted here, and would not be found online for linking.

As with any SPC implementation, you need to determine the dependent variable(s) - here a variety of cosmetic defects - and the independent variable(s) that contribute to them. It is nice if you can ponder your process and prognosticate them in your FMEA development - but often reality throws in a few unexpected variables. Charting the frequency that the dependent variable occurs will give a the scope of the problem. It will also give you feedback on if any of your independent controls are truly the influence and have been successful.

You also need to develop a mathematical model for each of the independent variables that makes sense. Many people take a sample, dump it into a program and expect to magically have it discern the distribution. That is a key failure in implementation - primarily because the program has no idea what sampling errors, measurement errors, etc., you have committed. Unfortunately, you need to think through that - and see if the data supports the theory.

Once you have collected data on the dependent variable, and started to develop your list of independent variables to control and the mathematical models for each, then you can determine the appropriate charting methodology.

So, yes, as a full implementation your MBB is correct, a variable chart will be ideal, but - as the MBB should recognize - you have several steps to complete before implementing one.

I'm not usually a name dropper, but what the heck: Deming said "Plot the Bloody Dots". Ellis Ott said: "Plot your data and think about it".

That's why you should plot your defect rate. as Bob said, the patterns may yield clues as to cause. If nothing else they will give you your baseline and enable calcualtion of the appropriate sample size for future experiments or determining if a change made a difference... by the way, your MBB should know this.

Since you are dealing with cosmetic type defects, I might also recommend the use of a "measles chart". make a picture of the thing and have the inspectors make a mark on the picture corresponding to the location of hte defect(s). Depending on the occurence rate, you may use a different color marker for different peiods of time or shifts...or you may need to change over the picture every 2 hours or every shift. The location of cosmetic defects can be a real clue as to their source...

As an aside I wouldn't recommend the AIAG manual. There are many wonderful insightful books available that deal with the theoretically perfect and the real world. The three authors I would recommend for a start: Donald Wheeler, Wise &Fair (Innovative Control Charting) and Grant & Leavenworth.

Great responses so far. :applause:

I'd like to add Statistics for Six Sigma Made Easy by Brussee to Bev's good reading suggestions - I haven't seen the books she is recommending, but I know myself well enough to gravitate to books using small words! :D

As an aside I wouldn't recommend the AIAG manual. There are many wonderful insightful books available that deal with the theoretically perfect and the real world. The three authors I would recommend for a start: Donald Wheeler, Wise &Fair (Innovative Control Charting) and Grant & Leavenworth.

I have the privilege of developing a medium-level SPC class, and I started looking for an appropriate text. Guess what...it doesn't exist. I have to pick and choose pieces of each - although I use Gerald Smith's as a basis, solely because it covers more of the scope of the course than others I had surveyed - not because it approaches perfect.

Fact is, SPC is not done being developed, and it still has one particularly difficult aspect that is very, very difficult to teach in a book....

....you have to think.

It is not plug and chug. Sorry.