Sign up for the Design News Daily newsletter.
How to Build a Better Assembly Line
The key to assembly line efficiency could be using artificial intelligence to quickly identify and prevent sources of production slowdowns.
December 14, 2020
4 Min Read
Dallas-based ONE Tech, Inc. develops artificial intelligence solutions for optimizing the performance of manufacturing, agriculture, and telecommunications. Design News chatted with Chris Catterton, Director of Solution Engineering, to learn how AI can improve assembly line efficiency in a factory. The company has an automaker customer who ONE Tech says was able to increase the overall equipment effectiveness of its assembly lines by 15 percent using ONE Tech’s MicroAI and Helio technology, and we wanted to know how this happened.
Design News: What are some of the areas where current production practices and techniques can be improved?
Chris Catterton: In manufacturing, some inefficiencies can go unnoticed due to a lack of technology that identifies and alerts upon their occurrence. There are many facilities that operate with little-to-no visibility within the actual cycle-times of assets and production schedules are set with the assumption that the line is running within the parameters of these planned cycle-times.
By getting more granular into measuring the performance, cycles, and utilization of individual assets, not just the end of the line, facilities can understand where underutilized assets reside, and identify problem areas that create large amounts of accumulated wait time.
DN: Why are scheduling and downtime issues so disruptive?
Chris Catterton: Downtime events, regardless of duration, impact productivity. When downtime events occur, it means completed units are not rolling off the end of the line. Manufacturing processes, in essence, are well-choreographed engineering masterpieces.
Raw material turns into completed goods by way of interaction with many moving machines and operators interacting with machines. Eventually, those completed units land on trucks or trains, to be distributed for commercial consumption. When there is a “hiccup” in one part of the process, it has a chain reaction through the rest of the process.
DN: How can AI help manufacturers plan in advance to address these problems?
Chris Catterton: AI is being adopted in manufacturing facilities to understand the “what and why” something is occurring and ultimately, what is about to occur. By converging data from both OT and IT, plant operators can now identify signs of failure in advance of the downtime events occurring.
Furthermore, by building the corrective action steps into the AI models, automation can help a machine to either “right itself” or, if there is a need for technicians to get involved, a work order can be generated automatically with details of what is occurring and what steps need to be taken to correct the issue.
DN: How was your automaker customer able to solve welding downtime problems?
Chris Catterton: The welding tips of MIG welders wear after each hit. As this tip becomes worn, more current or voltage needs to be pushed to the welder to ensure the same quality of weld takes place in the same cycle-time. By inspecting welds through various sensor technologies (arc voltage, welding current, wire feed speed, and infrared weld quality), ONE Tech processed data at each welding asset locally with our MicroAI technology.
MicroAI provided AI-Driven output, which was fed directly back to the welder controller for making adjustments for continuous improvement to the welds. This drastically reduced the amount of manual rework required that was caused by poor quality welds and also reduced the amount of unexpected downtime due to a self-correcting environment that allowed the robotic arms and welder to make adjustments based on the AI feedback that it was receiving.
DN: What were the resulting benefits to that customer?
Chris Catterton: Once ONE Tech’s MicroAI was implemented, there was a large reduction in unexpected downtime events. This allowed for more production from the line, thus resulting in more units and revenue for the facility.
DN: Are there other examples of auto industry customers who have benefitted?
Chris Catterton: Continued rollout to other plants is underway, but specific customers cannot be disclosed at this time. ONE Tech’s MicroAI product is designed to live at the edge. What we mean by this is that there is no dependency on a cloud/server environment.
More specifically, machine learning models can train and process data in a local environment. This has been noticed by auto manufacturers for both their production facilities but also for their vehicles.
Automobiles, trucks, trains, aircraft, etc., all are becoming more instrumented. Hundreds of sensors are being placed throughout these assets and some manufacturers are paying extremely high data transmission, storage, and processing cost to make sense of this data in cloud environments. MicroAI is now allowing for data to be processed directly in the TCU/ECM of vehicles, allowing for a drastic cut in data expenditures.
DN: Can you quantify the cost savings that result or the return on investment?
Chris Catterton: Exact numbers cannot be disclosed but continued rollout to other plants is underway.
DN: Other than the monetary return, what are the other benefits to customers of being able to stay on schedule? It seems like their own customer satisfaction should be better when they don’t fall behind.
Chris Catterton: Being able to meet planned production schedules is critical for continuing to receive orders. When units are produced at expected quality, within expected timelines, customer satisfaction is certainly increased.
You May Also Like
Generative AI Has Taken Over Supplier NewsMar 1, 2024|6 Slides
Innovative Tech Abounds at Mobile World Congress 2024Feb 29, 2024|4 Min Read
Will the Future of EV Charging Be Wireless?Mar 21, 2024|14:00 EDT
Generative AI Can Be Like an Expert in Your Pocket, Says Industry ExpertFeb 29, 2024|3 Min Read