How to determine target yields

[JosephERG]

Hi All,

I would like to inquire if any of you guys, especially
those in the fabrication business, can provide information on how to determine the "target yield".

In our company, QC engineers are required to set the target yield for each process involved in the manufacture
of our products. This value is used to monitor the process performance (whether it is low yield or not).

Currently, the target yield for each process is determined by getting their average within the previous 6 months (or more). Is our method correct?

Please advise. Thank you very much.

 

[tripleZ]

Our target yield is set by the applications engineer quoting the job. If they quote 3% scrap, then that is the minimum level we should be seeing on a monthly basis. The disconnect comes in where profitability targets indicate that scrap must in fact be far lower as a percentage of sales than 3%.

These levels are monitored on a monthly basis via QOS metrics meetings. Individually, process engineers will be monitoring the operations on a daily basis via scrap hash sheets or similar charting methods.

IMO the target yield should often be based on the dimensional tolerancing of the part, the part's geometry and how it is fixtured/manufactured, and the statistical machining capability of both your machine and your process. In actuality, the guys quoting often have a preset number that they use and which doesn't change often.

 

[JosephERG]

Hello tripleZ,

Thank you very much for the valuable information.

I would just like to clarify something. Is this "preset number" based from historical data or from actual process study?

In our case, the order qty of the products we manufacture ranges only from 3 to 10 pieces. It is very rare that we get orders past this range. And also, our products are not repeatedly ordered. In a span of one year, only a handful gets re-ordered. Therefore, it is difficult to acquire sufficient data.

In addition, new products are supposed to undergo quality qualifications so that we can determine its target yield. But considering the amount of time it would take to perform such activtiy, it is quite impractical to apply it in our situation. Most of our customer demand fast delivery.

I would assume that you are also in a fabrication business.
If so, how do you come up with your target yields without necessarily increasing lead time?

Please advise. Thank you.

 

[tripleZ]

If you only make 3-10 parts at a time, and your runs only occur several times per year, then we're talking an estimated annual quantity of say, 12-40 parts? At that level, I'd assume your target yields would need to be relatively high in order to maintain profitability...say 95% or higher. So your scrap per year is in the magnitude of 1-2 parts per product line max.

My statement of 3% was for a mass customized business where lot sizes vary from 5,000 to 125,000 parts at a time. When I managed production of a job shop we made one-of-a-kind toolsets. There were no multiples, so the obvious scrap goal was 0%. That's not to say that scrap didn't happen, but you certainly looked to keep the amount below a certain percentage of sales. You can look at historical data, but often this value is set by accountants or managers looking to manage a budget. It's irrespective of the job difficulty.

For job shop work, we didn't put time into evaluating the potential target yield. What we did to was try to design the manufacturing process around obtaining 100% yield. The first task for every new job was for us to determine where employees in various departments could generate scrap. We would then construct our shop instructions to take these factors into consideration so that each operator knew the potential pitfalls and where to be careful. Supervisors were responsible for following up with each task their employees covered.

This doesn't get you your target yield value, so this may not satisfy your pricing department. But as a small shop, we had limited resources, so we deployed them where they were deemed to be most effective. I hope that helps to clarify my position. It sounds like your shop structure is far different.

 

[JosephERG]

Hello Triple Z,

Thank you very much for your valuable insights.

We do manufacture 50-70 parts(more or less) annually, but this is only for one product type. We manufacture several
product types (each of them is distinct from one another), but in small quantities. If we talk about the total output
of the company per year, its way above 50-70 parts. But if
we talk about annual output per product type, then the
qty is within this range.

Actually, what you explained is quite similar to what
we are doing right now. I'm glad we basically have the
same approach.

I am interested to know how you do your scheduling. Do
you factor in the target yield in your schedule? In our
case, the target yield is used to determine the material
allowance (to cover up for yield loss).

If the target yield is relatively high, the material
allowance would be few also. In case, the accumulated
defective parts is more than the acceptable, the total
output could possibly be less than the customer's ordered
qty. Were you able to experience this situation also?


Please advise. Thank you.

 

[tripleZ]

My current facility is the mass customized one where lot sizes are in the 5,000 - 125,000 range, depending on the product line. We do figure in our scrap quantity when scheduling. So, yes, we will build larger lot sizes than ordered. But we have other factors we consider when scheduling. For instance, with some customers we have a 2%-5% overage/underage clause in the contract, meaning I can ship a final quantity within that range of their order quantity and be considered good. That's fairly common for the larger volume customers. Basically target your yield for 97% and run 3% higher than your order qty. If you have less than 3% scrap, you can still ship everything if the clause is active. If you have 6% scrap, you should still be able to make your quantity to the customer.

The other thing you have to consider is that in our facility, we work to forecasts that are anything but accurate, especially for various automotive customers these days. You develop a relationship with customers over time such that you learn to anticipate their patterns. I have several customers that always understate the forecast, then drop in orders at the last minute and expect immediate turnaround on what is normally a six-week lead time. So we build inventory ahead of time with this expectation. Does that mean our inventory turns are less and that we pay a higher holding cost? Yes! However, if I didn't do this, I would be shutting assembly lines down at which point my customers would be less than pleased. Is it an optimal situation? No. But we charge our customers for holding extra parts by having a higher piece price. So in the end, they pay a higher price for better service. My customers with accurate forecasts who accept the concept of lead time pay lower piece prices because they are willing to do some of the leg work. In the end, it's an accounting game.

Now, if your rate of defective parts is higher than your target yield, you hope that you have that excess inventory on the shelf to cover your order. If you do, you have just the added cost of maybe a break-in setup to get your safety stock back up to the desired levels. If you end up with a severe defect, sometimes you miss the delivery. How much safety stock you carry will depend on your end user. If you're tier 1 automotive, you'll probably run a higher level of safety stock because the fines for shutting down a line are high. If your customer base is different, then maybe you sacrifice on-time delivery for lower carrying costs.

So it's all a give and take relationship in our business. Optimally we would love to have a push-pull kanban style system with proper lead times built in. That would be both lean and cost effective. But when everybody is trying to push the holding costs of inventory to their supply base, you end up with the excess inventory situation.