In case of a quality problem from the customer or internal, one of the most challenging steps is “finding the dirty batch,” which is in the early stages of 8D Problem Solving studies.

For example, we have a few broken parts that were notified by the customer, but what if there are more defective parts somewhere? Even thinking about it can trigger some anxiety. How can we be sure and give an answer to the customer that “Don’t worry, all is okay”?

Wish it could be that easy, but not always!

Checking Every Part

We should remove uncertainties through our aim, which is identifying the exact defective batch. However, this depends on various factors, such as failure mode, production process type, production volume, and more.

Should we check every single part in case of a Quality Claim?

This might seem like the most secure option. Because if we check all the parts, then we can ensure that there will be no surprises later. If the production volume is low and there are no different far locations, this can be manageable, even the best method in such cases. Why take risks if it is easy and cost-effective? Absolutely the best method at that point.

But think of it this way: what if the production volume is too high, parts are shipped to many different locations, and there are various intermediate storage warehouses? How can we check every single part? Just imagine that, these kinds of sorting/control activities can even exceed the product’s annual net profit. It’s literally a nightmare, isn’t it?

Risk Assessment

We should absolutely do a risk assessment using all available data.

For example, let’s say your customer submitted a quality claim stating that 10 parts were detected as broken, and these parts are also being shipped in high volumes to other locations. Before making a decision about controlling the parts in those locations, a risk assessment helps to identify the dirty batch. Let’s make an example of that:

Imagine we received a quality problem notification from the customer, 10 plastic parts broken:

• One of the first things to do is checking the traceability data of those parts.
• So, we’ve checked the traceability of 10 broken plastic parts and seen that they were produced in 2nd mold. This is strong evidence that something is wrong with that specific mold.
• Even now, we just reduced the quantity to be checked by half!
• Let’s go one more step and make a table showing which batches were sent to which location and at what time.
• Now, we also see that the parts produced in 2nd mold were only sent to one location. So, we can just check the parts in that location.

Conclusion

While dirty batch identification can be a bit painful depending on the case, it is worth to avoid wasting money and effort.

Dirty batch analysis is an unwritten obligation that needs to be done, but you need to find out how! Because there are numerous possibilities depending on the process, customer, product complexity, and whether there is a sub-assembly, it gets more complicated. How can you generate a formula for something that varies in difficulty at every organization?

While the formula is not possible, I believe that the experience absolutely helps you do it better and faster the next time.

If you want to learn more, you can check out my Udemy course.