We’re setting up specifications for our product and we need to figure out the incoming and final inspections. How does quality assurance perform inspections on lots? What level of inspection is needed for which features of the product? We’ll explore some quality assurance methods and the designers input into them after this brief introduction.
Hello and welcome to Quality during Design, the place to use quality thinking to create products others love for less. My name is Dianna. I’m a senior level quality professional and engineer with over 20 years of experience in manufacturing and design. Listen-in and then join the conversation at qualityduringdesign.com.
If you’re designing products, you’re likely defining specifications. These are the drawings, lists of materials, and other things needed for someone to be able to make our product. Forefront in most of our minds is communicating to manufacturers and suppliers. But we need to remember our QA friends, too. As far as inspections, there are really three options: don’t inspect (rely on other controls or measures), 100% Inspect every part, or take a sample of a lot. If we’re developing products made in batches or lots, QA is likely going to sample part of that lot. They take a few parts from the batch and inspect it, and then they make a decision about the acceptability of the whole batch based on that sample. Do these products meet our spec? This is called acceptance sampling.
The most obvious case of when we use acceptance sampling is when our test to specification is a destructive test. I mean, we can’t wreck all of our product to see if it meets specs. It’s also used when 100% inspection is too expensive or too difficult; someone examining thousands of parts can get tired and miss defects anyway. Near you. We could also use acceptance sampling when we want the suppliers to be responsible for the quality of their lot. If we reject the lot, they may be forced to screen that lot for other defects, which is costly for them. This is all set up as part of the supplier relationship.
With sampling, yes there is risk in making an error: rejecting lots that actually conform to our specs and accepting lots of product that don’t meet our specs. This is where some statistics get involved. Think of it as a sliding scale: on one end is no inspection, on the other is 100% inspection, and in between there are various sampling recipes, plans and systems that can be set up. You can talk to your friendly neighborhood Quality Engineer about things like OC curves (which is short for operating characteristic curve), distributions, producer’s and consumer’s risk, single versus double sampling plans, Dodge-Romig tables…There’s a whole branch of quality studies just associated with sampling, but lots of producers use it and it’s a proven system that has been working for about 100 years.
Today, we are design engineers deciding how to characterize the features of our product. Our goal is to set up inspections for parts that line up with what’s important. What we need to decide is this: identifying what to inspect, defining how critical it is to the design, and helping to determine how it’s physically going to get inspected. Today I’m going to use my FMEA (failure mode effects analysis). This is a table that organizes information about potential failures and their effects (to the product, the people that use it, and the environment) and what could cause the failures to happen. This table also helps us analyze what we’re doing to control these problems. Are we doing things to prevent or detect the problems so that they don’t happen?
The first thing I consider is what features should Quality Assurance inspect? Here’s a thought process I would take:
- If a design feature is wrong, does it cause failure?
- What kind of failure does it cause?
- Is it something that has an effect that’s really bad?
If yes to all of those questions, then I’d consider it to be a critical feature. Now, what types of controls do we have in place? Should we add a control to this design feature that detects if there’s a problem? If so, we can use QA inspection as a way to control the cause of the failure. We’re using QA inspection to detect problems.
Another thing I consider is what level of assurance do we need about the quality of the product? Or in other words, how critical is it? I would again refer to my FMEA. How bad can things potentially get if this design feature isn’t right? We can base this on a lot of things:
- Is this a piece of a system that if it fails, the whole system becomes unusable?
- If this piece is wrong, does it affect any decisions that the users are making when using the product?
- Or does the product become hazardous to people or the environment?
From this we could start to get a better understanding of how much risk we’re willing to take in the sample for inspection. Your producer will likely have a documented sampling system associated with a level of criticality, like critical, major, and minor. And you can associate the level of inspection of your particular part feature to one of those levels. Critical features will have more stringent sampling requirements than the feature that is minor. If using the FMEA to guide us, we’ll likely get a mix of critical, major, and minor inspection levels for the features were inspecting. I recommended to assign them to be true to what it actually is. We don’t want to label everything as critical just because that’s the safest option. That’s putting unnecessary burden on the sampling scheme.
Another thing I consider is:
- How are they going to inspect these features?
- What tools are they going to use?
- When are we checking our parts: while it’s being made or after it’s all done?
This is where we may want to talk to our QA friends about the tools available to them. If we start thinking about this early in the design process, we could create our design to be able to be inspected. Sometimes what we think is a critical feature that we want to inspect is too difficult to reliably inspect. Or if we would have created our spec a little differently, then instead of having to use an expensive coordinate measuring machine, QA could use a handheld caliper to measure the feature. Another example is designing to allow for in-process inspections. Can we easily interrupt the assembly to perform an inspection or test, or does it become difficult to handle the half-assembled product so test becomes difficult? These are things we can consider as we’re designing.
The last thing I likely think about is who is going to inspect it: our in-house QA department, or are we relying on our suppliers to report the results of their QA inspection? This is where a conversation with someone in supplier quality management may help. They may understand the capabilities of the suppliers and what types of inspection agreements are possible to set up.
These can all be tricky questions, but ones we need to consider to fully develop our specs.
What can we do with what we’ve talked about today? I’ve demonstrated the use of an FMEA to help guide decisions about inspections. If you have an FMEA, pull it out and take a look at it. See if you can use it like I do. If you don’t have an FMEA, then you can still think through a similar thought process. Base inspection levels from how critical that feature is to the design. Finally, as we’re designing, we can think about how we’re going to be testing and inspecting our product against our specs. We may need to adjust our designs on paper to be able to inspect the parts in hand.
Please visit this podcast blog and others at qualityduringdesign.com. Subscribe to the weekly newsletter to keep in touch. If you like this podcast or have a suggestion for an upcoming episode, let me know. You can find me at qualityduringdesign.com on Linked-In, or you could leave me a voicemail at 484-341-0238. This has been a production of Denney Enterprises. Thanks for listening!