Make the Zero-Defect Factory a Reality for Manufacturing with IIoT and Analytics

A plant manager’s dream come true: machines at optimum efficiency with planned maintenance and zero defects. Is it too good to be true?

In a recent blog post “SPC 2.0: How IoT Predictive Analytics Gives You Psychic Powers” I outline how SPC is a great tool for looking back at historical information to make changes in manufacturing, but if you want to get ahead, IIoT is needed to take the first steps into the world of predictive analytics to begin looking into the future of manufacturing and taking action in the present.

Today, I want to take that one step further and answer the question “Is a zero-defect factory too good to be true for manufacturing?”

There is a dramatic shift taking place: a move from statistical process control (SPC) usually used to reduce defects in manufacturing to a continuous improvement approach. With the SPC approach, methodologies such as Lean Six Sigma are based on batch and offline analysis of data. Here is the challenge: the data is immediately outdated – real-time results are not possible, so plant managers never have a truly accurate, and updated situation of the plant. See the risk?

Oh, and one other big problem with SPC and Lean Six Sigma: the number of false alarms in early detection of problems. If you set the sensitivity too high, it is “crying wolf” a lot that something is wrong when in fact, nothing is. Yet, if you don’t set the sensitivity high enough, defects get through. Another challenge.

So SPC gets you part of the way to a zero-defect factory. Many manufacturers are happy with SPC and have given up on the promise of zero-defect – they have resigned themselves that defects are just part of the manufacturing process. I’m here to say – not necessarily.

Update Your SPC with IIoT

Now, if you take all the excellent things that Lean Six Sigma does offer and combine SPC with the application of IIoT tools such as runtime advanced analytics and the right hardware (which must be able to collect enough data from the factory’s now-connected data points), it is possible to run a continuous analysis of the industrial assets and detect much earlier and more precise quality and process defects of a single machine or an entire plant floor.

We’ve just moved from SPC looking at “past data only” to continuous improvement looking at the entire data set in real-time, with predictive analytics thrown in as well. Past, present and future. Powerful.

A side-by-side comparison of traditional statistic methods/SPC vs. IIoT data analysis shows process defect detection. See the trend in IIoT? Very clear – and it can be corrected before it hits the threshold. Unfortunately, with SPC alone, it’s often too late by the time the defects are recognized and the threshold is passed.

IBM has studied the impact of predictive analytics vs. traditional statistical methods. The IBM Quality Early Warning System (QEWS) allows for predictive analytics with some variables controlled by the plant manager, including:

  • the frequency at which measurement data are captured
  • the frequency at which measurement data are analyzed
  • the magnitude of shifts in measurement data
  • the period over which shifts in measurement data occur

The results are impressive: for subtle or slowly-developing problems, QEWS can alert 25 to 35 data points earlier than a traditional SPC chart, without any increase in the rate of false alarms. SPC can’t compete on that level.

Other companies have first-hand testimonials – this short video (Smart Manufacturing and Zero-Defect Factory) offers several case studies on the effectiveness of IIoT and predictive analytics.

Dan Zhang penned an interesting article earlier this year that was published in Dataversity. “Smart Factories and the Industrial Internet of Things” gives several examples, mostly from the automotive world, where IIoT has been implemented and he describes the evolution and importance of smart manufacturing. I am beginning to think that the term “smart” to describe everything from our homes to our cars to our lightbulbs is now overused, so I hesitate to call the manufacturing “smart.” But, Zhang is right and it is worth a read: factories are moving from “Lean to Smart” – they certainly have intelligence built in well beyond traditional SPC and Lean Six Sigma.

Do We Still Need Lean Six Sigma?

In a word: YES. SPC is still incredibly valuable and does an amazing job collecting the historic data in most cases. What IIoT does is complement SPC and Lean Six Sigma to give a bigger picture.

Let me give you an example:

With SPC: It’s like you are driving a car but you are only able to look in the rearview mirror while driving. All your windows are blacked out and you have no other mirrors. You hope the road is straight, that nothing is ahead or on the sides of you and just have faith in the data you’ve collected. If you see a change in road conditions behind you, you can adjust but you are already in the situation.

With SPC + IIoT, now your car has all the windows clean so you can see. All its mirrors around the car are clear, the backup cameras are in place and you have a screen on your dash showing you what the road ahead of you looks like before you get there so you can adjust accordingly. You even have gauges telling you about efficiency and that you have scheduled maintenance coming up, so your car never breaks down. Ever.

Do you still need your rearview mirror? Of course! YES! But it is a part of a much larger solution!

Further, Durgendra Singh has a great piece which outlines how IoT can Help Lean Six Sigma on Pulse, including:

  1. Data integrity
  2. Identifying problems/defects
  3. Finding root causes
  4. Solution monitoring

In closing, IIoT will help manufacturers improve their SPC approach to reach a zero-defect factory goal.

At Wi-NEXT, we strongly believes that to build the factory of the future it is necessary to know the reality, the language, the approach and the practical needs of the current market.

If you start from SPC and the Lean Six Sigma approach to improve the control capability with IIoT, is not disruptive, but instead a really concrete way to move toward to continuous improvement by leveraging all the capabilities that a digital factory can give to manufacturers.

For more information on how to implement a IIoT or predictive analytics solutions, please contact Wi-NEXT.