Skip to Content

Time to upgrade your shelves: Ramp up your shop floor with AI solutions

Artificial intelligence (AI) is enjoying a rapid uptake by manufacturing companies of all sizes. Engineers today are keen to find shop floor problems where AI can create an impact. So, what do engineers need to do? How do they go about identifying these problems? From our experience, we see a clear pattern across manufacturing companies in identifying shop floor problems: a tendency to address only big shop floor problems (such as predicting a component failure time and automated quality inspection). We also observed that they select these based on AI hype given for the big problems in the manufacturing community. They hope to obtain higher shop floor productivity gains in one shot.

What if I told you that there could be another angle to it – that, a big problem can exist because of many sub-problems’ underneath? You’d retort that solving these might individually lead to small shop floor productivity gains. But, in sum, these can in fact lead to productivity gains outweighing the benefits obtained by solving one or two big shop floor problems within the same time frame. End result? Higher and faster returns on AI investments!

Let me give you two hypothetical examples from manufacturing companies in Sweden.

Addressing throughput issues with bottleneck elimination

A company experienced a low shop floor throughput. The engineers wanted to improve the throughput significantly. To realize this, they chose to address the problem of bottleneck elimination using AI. They thought solving this problem would directly help them achieve significant improvement in throughput. After deploying AI and eliminating the bottlenecks, they noticed only an incremental improvement in the throughput and couldn’t reach the target. Improving further was challenging as the production lines on the shop floor were well balanced.

The engineers tried to dig deeper to understand the low throughput problem. This exercise exposed several issues outside the production lines on the shop floor. These problems were collectively hindering the throughput. It included production planning problems (e.g., less accurate manual calculations on production planning and scheduling) and logistics problems. Careful evaluation of the throughput impact of solving these problems using AI was collectively higher than addressing the single bottleneck elimination problem.

Achieving zero breakdown with forewarning system

Another company had the vision of achieving zero breakdowns in a CNC machine. To realize it, they chose to have a forewarning system that can predict the failure times of a set of critical components in the CNC machine. They used AI to predict the component failure time. The engineers were excited and had a sense of satisfaction in achieving the vision of predictive maintenance. The AI solution helped alert maintenance engineers about the unplanned events before they occurred in real-time, thus reducing the operational uncertainty.

However, they realized after a month that the maintenance frequency and the time duration to perform the maintenance hadn’t decreased. They decided to explore the problem deeply and tried to understand the complete workflow adopted to the maintenance problem. This exercise helped uncover a set of questions for which they have to find the answers, including questions on work order navigation in the system, handling the work order by maintenance engineers, diagnostic and decision-making process, and performing the maintenance tasks.

Finally, they found three sub-problems regarding work order navigation in the system that affected the time duration. The problem that affected the frequency was that different engineers followed different maintenance procedures. This process introduced anomalies, which in turn caused the failure of the components. After streamlining the work order navigation and building an AI recommender system to prescribe maintenance actions based on the skills levels of engineers, they reduced the frequency and the time duration. The impact was much higher than just establishing a forewarning system.

One size doesn’t fit all

Every shop floor is different. Big problems on one shop floor might not have the same impact on other shop floors. When selecting the use cases for AI solutions, we suggest engineers look under the layers of the big problem. There could be multiple sub-problems that might collectively influence the bigger one. One way to identify those sub-problems is to think about the decision supply chains of the big problem and explore the weak links in the chain. These problems could be an excellent candidate for AI solutions. When solved, the returns on such AI investments could be huge and outweigh the returns on AI solutions targeted exclusively for the big problem.

We are excited to hear how you identify the use cases for AI solutions on your shop floor. What is your approach? If you find it challenging to identify use cases that benefit from AI-based solutions, reach out to us. Our team has extensive experience and can help you identify impactful shop floor problems in a data-driven way.

About Author

Mukund Subramaniyan

Email | Linkedin | Twitter

Mukund works in the manufacturing sector, mainly advising clients on topics related to digital manufacturing, data analytics, artificial intelligence, and operations transformation. He has extensive experience working with major automotive companies in India and Sweden including original equipment manufacturers (OEMs) and suppliers. His educational and professional background integrates technical expertise as computer scientists with the manufacturing expertise of engineers. This ensures that the deep insights generated through AI translate into real measurable impact in an organization. He is passionate about transforming manufacturing operations using IIoT, AI, data, insights, and actions.