First Pass Yield (FPY) of a Manufacturing Site?

[tariq18]

Hello experts,
May be its a stupid question but I was wondering what's the best approach to calculate first pass yield of a manufacturing site if you have more than one production line producing different range of products?

Option-1: Calculate no. of batches produced at all the lines and check total failures to calculate FPY.
Option-2: Calculate FPY of each line then take average? .......wait a sec........ should we take average here or multiply FPY of each line to get total FPY(as what we do to calculate Rolled Throughput Yield RTY) , its really confusing isn't it?

Would love to hear feedback fromeveryone!

Thanks,

 

[Miner]

When you aggregate data, you lose actionable information. You are better off analyzing yield at the detail level, whether that is by line, by product, or both, so you can identify and target the areas with the lowest yield for improvement.

 

[Zero_yield]

Depends heavily on what you're trying to do with the information. We calculate several different metrics that are used for several different purposes.

One metric is "right the first time", which is essentially number batches not attached to a nonconformance divided by the total number of batches, as a percentage. It's calculated for each production line at a very high level, and is usually 99%+. It's mostly there as an early-warning system for spikes in nonconformances.

We have other metrics that get deep in the weeds of the exact breakdown of rejects by every product family and production line.

Ultimately, I think it's best to have a mix of high level and detailed breakdowns - it's easy to have too high an overview and miss details, and it's easy to miss the forest for the trees. Just keep in mind what the goal of each item tracked is.

 

[tariq18]

Depends heavily on what you're trying to do with the information. We calculate several different metrics that are used for several different purposes.

One metric is "right the first time", which is essentially number batches not attached to a nonconformance divided by the total number of batches, as a percentage. It's calculated for each production line at a very high level, and is usually 99%+. It's mostly there as an early-warning system for spikes in nonconformances.

We have other metrics that get deep in the weeds of the exact breakdown of rejects by every product family and production line.

Ultimately, I think it's best to have a mix of high level and detailed breakdowns - it's easy to have too high an overview and miss details, and it's easy to miss the forest for the trees. Just keep in mind what the goal of each item tracked is.

As far as I understand FPY and Right first time(RFT) is same thing. we also do RFT and its always 99%+ but the problem with this high level RFT is missing details as you mentioned, and management is not aware of real issues on the floor. another problem with RFT/FPY is it does not count reworks/scraps.

But anyways thanks for the reply, my second question was if I record RFT for each line, should we then calculate mean/avg. RFT of all lines combined or we should just multiply them all with each other as what we do in case of rolled throughput yield.?

 

[Mike S.]

What if Line 1 has a first pass yield (however you measure that) of 90% and Line 2 has a yield of 99%. You'd wanna focus on Line 1, right? Well....maybe not. Maybe actual losses to the company are higher on the 1% from Line 2.

So why are you measuring FPY? Clearly state the goal of measuring FPY. Then collect the data in a way to help you achieve that goal. There is no reason to collect data unless you plan to take action as a result. So many times I see data collected and no one pays any attention to it.

I really enjoyed watching a Plant Manager go on and on for 15 minutes telling a customer about all the data (many dozens of pages posted on carpeted wall panels) they are collecting for each line and how the andon lights tell anyone walking by if things are going well for that line or not, etc. etc. ad nauseum. The customer patiently waited until the guy was done and asked why the light was green for Line 1 when the data showed they had a lot of defects for the last 2 weeks with no apparent change and no actions being taken.

After the PM got the stunned look off of his face he babbled about manpower shortages and how people from Line 1 had to be moved to another line to help out there, etc. It was not a good look.

 

[Zero_yield]

I actually had not heard of a Rolled Throughput Yield before. It does sound like a useful tool to meet your situation (i.e., trying to explain to management how much rework and waste is going on).

 

[Bev D]

…management is not aware of real issues on the floor. another problem with RFT/FPY is it does not count reworks/scraps.

First and most importantly is that FTY (and RFT) is SUPPOSED to include reworks and scraps. Period. The only reason to exclude them is to fool yourself that things are better than they are. The ENTIRE POINT of FTY is to include rework as well as scrap. (I hope this was a misstatement but it doesn’t feel that way)

Second FTY by itself is pretty meaningless and shouldn’t be ‘averaged’. It was intended that the FTYs of each step are MULTIPLIED together. (This is based in the rules of conditional probability. What is the probability that any given unit of product passes they the entire process without a single defect…). This metric is Rolled Throughput Yield or RTY. The individual step FTY is valuable in telling you which steps are the most troubling but averaging them hides this information.

Third: RTY is intended to be calculated for each unique product manufacturing line. Aggregating RTY (or the meaningless FTYs) into some large overall metric is not informative or actionable. It may be mathematically interesting but it does nothing to improve the quality of your processes and products.

A great simple article on RTY is: “Yield the Right Way” by Thomas Pyzdek. It is free…

If you want to geek out on math try “Statistical Quality Control of a Multistep Production Process Using Total Process Yield” by Spencer Graves (Quality Engineering, 1998). This is not a free article, has limited utility for real life quality engineering (just too complicated) but is interesting and informative.

 

[tariq18]

First and most importantly is that FTY (and RFT) is SUPPOSED to include reworks and scraps. Period. The only reason to exclude them is to fool yourself that things are better than they are. The ENTIRE POINT of FTY is to include rework as well as scrap. (I hope this was a misstatement but it doesn’t feel that way)

Second FTY by itself is pretty meaningless and shouldn’t be ‘averaged’. It was intended that the FTYs of each step are MULTIPLIED together. (This is based in the rules of conditional probability. What is the probability that any given unit of product passes they the entire process without a single defect…). This metric is Rolled Throughput Yield or RTY. The individual step FTY is valuable in telling you which steps are the most troubling but averaging them hides this information.

Third: RTY is intended to be calculated for each unique product manufacturing line. Aggregating RTY (or the meaningless FTYs) into some large overall metric is not informative or actionable. It may be mathematically interesting but it does nothing to improve the quality of your processes and products.

Well, we have 4 to 5 single step production lines and one multistep (for which I am planning to use RTY as exactly what you mentioned), & for single step production lines, FTY is enough. My only question to you is how do we include reworks and scraps in FTY?

My question why I start the thread was, how do I calculate then overall FTY of this manufacturing site? What's the best approach?

Anyways, thanks for the links, its quite useful.

 

[Bev D]

I’m not sure wha tyou are including in your FTY calculation besides rework and scrap…but here is what you do for a single step process
Say you start 100 units. 4 units are rejected. 1 is used as is, 2 are reworked and pass and 1 is scrapped. FTY = 100-4/100 = 96%.

As for a manufacturing site FTY, I still say don’t do it. Every metric needs a purpose. Every metric should be actionable (even if it means you have to break down the elements of the metric which is typical). What is the purpose of a site wide FTY? I think there is something critical that you are missing here so really think about it…If we know what you are trying to understand with a site wide FTY we can help you get to a useful metric that satisfies that need. I don’t see any useful purpose for a site wide FTY. At all. Ever.

 

[tariq18]

I’m not sure wha tyou are including in your FTY calculation besides rework and scrap…but here is what you do for a single step process
Say you start 100 units. 4 units are rejected. 1 is used as is, 2 are reworked and pass and 1 is scrapped. FTY = 100-4/100 = 96%.

As for a manufacturing site FTY, I still say don’t do it. Every metric needs a purpose. Every metric should be actionable (even if it means you have to break down the elements of the metric which is typical). What is the purpose of a site wide FTY? I think there is something critical that you are missing here so really think about it…If we know what you are trying to understand with a site wide FTY we can help you get to a useful metric that satisfies that need. I don’t see any useful purpose for a site wide FTY. At all. Ever.

This is exactly how I calculate FTY for single step process. 2 units reworked here are not included in your FTY. This is what I meant before.
So those units needed rework, have to go though process of internal NCRs etc.

Here I have another question, when you rework those 4 units, do you again consider them into your FTY stats?

For site FTY, we have more than 3 sites across the country and senior management has set some targets for each site, that's why we need site wide FTYs.