E
equesnel
Everyone,
I am looking for some assistance on this one, Rolled Throughput Yield vs. Normalized Yield.
We have a customer that insists on First Pass Yield metrics for our respective meter assembly processes which consist of a 12 Station Process. Out of these individual stations we compile various defects some station performance related & some meter form, fit, function related.
Issue is they are all lumped together in the process tracking database. We will have to filter data for the meter specific defects - not too big a challenge. The customer only needs to be concerned with the meter specific information, as these are issues that will be seen potentially by them.
I have gone a little divergent to frame the current situation.
Customer visits for an audit observes series of meters failing, asks what is the First Pass Yield for the station - stats are pulled and it appears that is is performing below goal. However, the optical camera needed adjusting after this values meet and exceed target.
Customer asks for the Rolled Throughput Yield for the Entire Process:
Rolled Throughput Yield (RTY) = Station 1 * Station 2 * .... Station *12
We report the RTY yield - Let's say for discussion it was a theoretical number such as 87% - they say Automation Process at Risk in meeting Customers demands of 95%.
Is it just me that sees how difficult this is to achieve - .995 ^ 12 = .9416 not quite good enough. If all the stations performed at a FPY = 99.5%, we would still miss the mark.
This is why I like the concept of normalized yield:
Normalized Yield = nth Root (Rolled Throughput Yield) = .87 ^ .0833 = .988
They would be happy with 98.8%, but how do I explain the Theory behind this. This is the expected FPY for stations in the process, which I think they really want vs. the chance of a single meter getting through the process defect free.
Rework is a fact of life for a new start up process until the learning curve gets firmly established.
HELP WITH THE THEORY BEHIND THIS, SO I CAN EXPLAIN NORMALIZED YIELD EFFECTIVELY TO OUR CUSTOMERS AND SELL THEM ON IT!!!!!
Any feedback on this will be greatly appreciated!!!!
Thanks.
Eric Quesnel
QC Manager - Itron
I am looking for some assistance on this one, Rolled Throughput Yield vs. Normalized Yield.
We have a customer that insists on First Pass Yield metrics for our respective meter assembly processes which consist of a 12 Station Process. Out of these individual stations we compile various defects some station performance related & some meter form, fit, function related.
Issue is they are all lumped together in the process tracking database. We will have to filter data for the meter specific defects - not too big a challenge. The customer only needs to be concerned with the meter specific information, as these are issues that will be seen potentially by them.
I have gone a little divergent to frame the current situation.
Customer visits for an audit observes series of meters failing, asks what is the First Pass Yield for the station - stats are pulled and it appears that is is performing below goal. However, the optical camera needed adjusting after this values meet and exceed target.
Customer asks for the Rolled Throughput Yield for the Entire Process:
Rolled Throughput Yield (RTY) = Station 1 * Station 2 * .... Station *12
We report the RTY yield - Let's say for discussion it was a theoretical number such as 87% - they say Automation Process at Risk in meeting Customers demands of 95%.
Is it just me that sees how difficult this is to achieve - .995 ^ 12 = .9416 not quite good enough. If all the stations performed at a FPY = 99.5%, we would still miss the mark.
This is why I like the concept of normalized yield:
Normalized Yield = nth Root (Rolled Throughput Yield) = .87 ^ .0833 = .988
They would be happy with 98.8%, but how do I explain the Theory behind this. This is the expected FPY for stations in the process, which I think they really want vs. the chance of a single meter getting through the process defect free.
Rework is a fact of life for a new start up process until the learning curve gets firmly established.
HELP WITH THE THEORY BEHIND THIS, SO I CAN EXPLAIN NORMALIZED YIELD EFFECTIVELY TO OUR CUSTOMERS AND SELL THEM ON IT!!!!!
Any feedback on this will be greatly appreciated!!!!
Thanks.
Eric Quesnel
QC Manager - Itron