Yield Monitoring

[JosephERG]

Hi All,

Good day!!!!!

I would like to inquire, especially for those in a job-shop business, if you can share some tips on how to monitor process yields.

In our company, process yields are determined by getting the average yield for each process (of a particular product type), within a 6-month (or more)period. These values are then considered as the "cut-off yields" for each process.

We use these "cut-off yields" to determine whether a process is low yield or not. For example, for a product with a Lot Size of 5 pcs, the results are:

Process In Out Actual Yield Cut-Off Yield
A 5 5 100% 100%
B 5 4 80% 75%
C 4 3 75% 80%
D 3 2 67% 70%
E 2 2 100% 100%

Based from above example, processes C and D are considered low yield already. I have doubts about the validity of this method because the lot size has influence over the actual yield. If we increase the lot size, probably the yields for each process will be above the cut-off.

Are there other factors to consider when monitoring process yields? I hope I can hear from you guys soon.

Regards......

 

[MikeHalloran]

With a lot size that small, can you afford a yield less than 100 pct?



Mike Halloran
Pembroke Pines, FL, USA

 

[JosephERG]

Hello Mike,

Good Day!!!!

That is actually my dilemma right now.
Due to the small size of our lots, it is almost inevitable that our target yield should be 100% for each process even though the average yield (within 6 months) is lesser.

I believe that there is a better way of monitoring the yield, that is why I am interested to know how other job-shops are doing it.

Does the yield have to be monitored by lot?


Best Regards.......

 

[MikeHalloran]

I'm not convinced that any sort of statistical analysis or delayed monitoring or other 'office work' will help.

Old machinists develop a 'sense' for what a particular machine will do on a particular workpiece with a particular sort of tooling. ... by producing scrap while they are young machinists, or by stealthily recovering parts that might have been scrap by means of some undocumented additional operation or ad hoc tooling change. ... or by initiating a design change that redefines the threshold between scrap and acceptable stuff.

Old machinists also eventually become conditioned to not actively support systems or programs for yield improvement, because they're >>already working toward 100 pct yield all the time<<, and not getting recognition for doing it, while the young whippersnappers who come and go with the tide, often get credit for improvements that are either statistical happenstance, or direct results of changes implemented by the shop floor people alone.

The only 'program' that I've ever seen actually work is a one-on-one application of the Hawthorne Effect. I.e., get out on the floor where the work is done, talk to the people who are doing it, and let them tell you where the process capability is thin, and especially, let them help you make it better. You'll suffer some abuse until they really believe you're sincere and honest and fair, and seriously working on what you say you're working on.




Mike Halloran
Pembroke Pines, FL, USA

 

[JosephERG]

Hello Mike,

Good Day!!!

Everything that you said applies in our organization as well. I also believe that quality is more than just continuous improvement, it is "innate exellence".
However, management doesn't wanna listen to that. They always demand figures.

Do you use any quality metrics?


Best Regards.........

 

[MikeHalloran]

(
I was speaking of another place that exists no more.

Right now, I work with a shop full of individual craftsmen.
We had a QC department, but all that's left of it are procedures that were too cumbersome to follow even before the company shrank.
)

All I can think of is to feed back whatever data you're gleaning from production records to the operators and setup people responsible for the machines.

The shop people won't take you seriously unless you allow them to spend some money on maintenance and small enhancements, at their discretion. If that money is (sensibly) limited, they may develop their own metrics for allocating it.



Mike Halloran
Pembroke Pines, FL, USA

 

[JosephERG]

Thank you very much Mike for the valuable insights.....