Using Statistical Process Control to Reduce Inspection Costs

Welcome to the Manufacturer’s Edge video series. My name is Karen I am the manager of Quality Systems here with Manufacturer’s Edge. Today I’ll be talking about SPC or Statistical Process Control and how it reduces costs and improves quality for manufacturers. I’ll be providing a brief introduction to SPC and why it can be such a powerful tool and then we’ll show an example of SPC in action and finally we’ll talk briefly about how to get started with SPC and show you that it doesn’t have to be complicated or expensive okay so let’s talk a little bit about what SPC is.

So SPC stands for Statistical Process Control and it’s a preventive tool which is what really makes it so powerful um so we use it to predict process performance and identify unpredictable variation so depending on the capability of your manufacturing process. SPC can often alert you to process issues prior to making any non-conforming performing material so prior to you needing to do any sorts or dispositioning or anything like that. So functionally SPC requires that you take the measurements that you’re probably already measuring, plot them, and apply some calculations to alert you when statistically unlikely conditions occur. So, whenever things are running normally you don’t expect statistically unlikely conditions to occur um so when you do detect something unlikely that happens it probably indicates that things are no longer running normally. Something has changed in the process and this is when it needs to be investigated so that you don’t produce bad material for your customers um and ultimately SPC also indicates to operators when adjustments are needed so operators sometimes have a tendency to adjust things um and often this is to the detriment of the process we call it tweaking sometimes um because they don’t know exactly when the variation is just the process doing its thing and being normal and when it’s an issue. Sometimes we’re over-adjusting our processes and so operators are the primary users of SPC but most roles in an organization will likely benefit from using SPC again because we’re predicting the process performance so we can use it to make inferences about the entire population of product. What I mean is while you may not be measuring every product you can still get an idea for how the products that you aren’t measuring would perform and what would look.

I think the best way to see how SPC works in practice is in an example. Here I’ve got a process where we’re sampling product over time we’re measuring. We happen to be measuring diameters and we’ve got a target of one just for simplicity. Alright so I’m going to go ahead and take a take a measurement right so we see it’s not exactly one but we know there’s always very building process so probably not an issue and we will continue to take samples. So over time, obviously, their measurements aren’t exactly one um but you can see that over time we get some idea of how the process behaves based on the samples that we’ve taken so what we can’t predict the next value we could kind of estimate that of the bounds where we think it would fall. I don’t expect my next measurement to be a 0.2 and I don’t expect it to be a 1.8 um so this is kind of the first thing that SPC does is uh is calculate what we call control limits of the process. Or this is where we expect the process to operate under normal conditions.

So here you can see I’ve added some control limits here so we’ve added a lower control limit and an upper control unit so then if you take a sample and it’s outside of those control limits SPC would alert the operator that there’s probably an issue there. This is statistically unlikely and something has probably happened in the process and so this is some very special cause variation has been introduced and the operator should go figure out what went wrong and try to fix the issue. Similarly, right we might get something like this where we get a pattern. It’s pretty unlikely that you happen to measure five in a row and they’d all be they’d all be increasing so it’s still in the control limits but again it’s pretty statistically unlikely and this is this is where um what we call run rules would come into play so again SPC would go back statistically unlikely there’s something has changed in the process and that needs to be to be addressed foreign so and remember too that we are likely just sampling product here. So in this case in the first case when we had the high value there’s there may be product that we didn’t measure that’s even higher um this becomes an issue when this the product the measurements are so high or so low that they’re not conforming or they’re out of spec and that’s when you end up having a whole product scrap rework or disposition. And this is where it can really be costly to manufacturers um or you know worst case scenario you send bad product to a customer right which is extremely costly in most cases. Okay so some of the benefits of SPC so as we saw SPC predicts process performance and ultimately that means we can catch issues faster in many cases this is before they become a non-conformance and affect the customer. So predicting process formats performance why is this helpful well data costs money imagine a line where you’re producing a product at a lot of product at a high rate of speed you can’t measure every part off the line but again SPC can predict what your entire population or all those other parts look like. Um and even if you don’t have a process where you’re producing a lot of parts consider your suppliers if you can get SPC data from your suppliers you can have more confidence in what they’re giving to you, and you can reduce or eliminate the need for incoming parts inspection as well other you know other benefits obviously catching issues before but earlier. You know reducing smaller holds reduce scrap reduction in overtime and sorting costs and again improved overall customer satisfaction and retention and reduction in time in the time it takes to train new Personnel so we said you know um it reduces the amount of adjustments you have to make so it indicates to operators exactly when they should make an adjustment and when they shouldn’t make an adjustment so it kind of takes some of that some of that out as well.

Karen Zrebiec
Manager of Quality Systems
Manufacturer’s Edge

Karen has 14 years of experience in quality and process engineering in manufacturing organizations. Before coming to Manufacturer’s Edge, she spent eight years working for two high-volume manufacturing companies producing metal packaging solutions. Prior to that, Karen worked in printer supply development and nuclear fuel manufacturing.

Karen’s focus has been on data-based decision-making and problem-solving. She has extensive experience training six sigma methodologies across all levels of the organization both domestically and internationally. She also has significant experience developing quality management systems and auditing against quality standards including ISO9001 and Safe Quality Food (SQF). Karen has a proven track record of leading cross-functional teams to drive process improvements, increase asset utilization, and improve customer satisfaction.

AREAS OF EXPERTISE INCLUDE:

• Statistical Process Control
• Data Analysis
• Process Based Auditing
• Gauge R&R Studies
• Problem Solving

EDUCATION/PROFESSIONAL CERTIFICATIONS:

• Master of Engineering – Engineering Management, University of Colorado Boulder
• Bachelor of Science – Engineering Physics, University of Colorado Boulder
• Certified Lean Master – APICS
• Six Sigma Master Black Belt
• ISO 9001 Lead Auditor
• Certified Practitioner Safe Quality Food
• Certified Hazard Analysis and Critical Control Point (HACCP) Practitioner
• Certified Quality Engineer (CQE) – American Society of Quality (ASQ)
• Certified Quality Auditor (CQA) – American Society of Quality (ASQ)
• Certified Manager of Quality/Organizational Excellence (CMQ/OE) (ASQ)