Lean manufacturing looks to promote system-wide efficiency through continuous improvement by refining and aligning design, process, and quality requirements with business objectives. However, without consistent execution and continuous improvement, manufacturers see slow and mixed results from this Lean methodology, losing out on the important investment and efficiency this methodology provides.
The journey towards autonomous manufacturing has become an increasing priority for manufacturers facing increasing pressures in global sustainability, process efficiency, margin pressure, and economic uncertainty.
For many manufacturers, investments in robotic process automation (RPA) have seen massive improvements in production capacity. Robots are fast and consistent but do not improve quality. If your current plant produces 1000 defects, RPA will make 10 times as many good parts and 10 times as many defects. Autonomous manufacturing looks to combine robotic process automation and Artificial Intelligence to continuously optimize quality using robots and quality data.
THE BENEFIT THAT AUTONOMOUS MANUFACTURING BRINGS LEAN MANUFACTURING
The lean methodology is based on the idea of repeated planning, execution, evaluation, and acting based on careful, time-consuming experimentation. The ability of AI to automate root-cause analyses without having to design a new experiment is very different from the trial-and-error approach outlined in Lean manufacturing.
Looking at historical data, we observe a ‘natural experiment’ where AI systems can learn from past plant states used by operators to offer prescriptions that continually optimize the plant. This greatly reduces the risk of experimentation seen in lean manufacturing and its impact on machine availability and scrap rates throughout the particular study.
THE INFLUENCE AI HAS ON PROCESS OPTIMIZATION IN MANUFACTURING
In manufacturing, the use of AI can optimize production without the need to invest in new and expensive manufacturing equipment. By studying the relationships between equipment and process parameters, researchers estimate that using AI solutions in manufacturing can significantly reduce yield loss significantly. This is not just AI identifying safe process parameters; this is AI identifying optimal historic set points that reduce defects that previously required the machine, technician, and operator’s time to fix them.
One of DataProphets’ clients, a major luxury and utility vehicle manufacturer, experienced challenges with stud-welding operations in the most complex part of their process-the body shop, which has at least 200 robots in operation.
The main problem was the additional rework required on the vehicle and increased cycle-time due to stud welding faults, resulting in immense costs to the business. Owing to the complexity of the process, traditional statistical solutions fell short. However, by implementing DataProphet PRESCRIBE (an AI-enabled parameter optimization solution) at the facility, we were able to solve this problem. DataProphet PRESCRIBE was able to prescribe optimal process parameter values for hundreds of stud weld locations simultaneously.
DataProphet PRESCRIBE also captured complex interactions with other related processes—consistently predicting weld quality for the client. After implementation, our client was able to reduce stud welding defects by 55% within the first month. This increase in welding quality led to an equally significant reduction in defect-related downtime at the facility.
Another client, an engine block manufacturer, was faced with the problem of an inefficient process of detecting defects. There was a team of operators who would note which engines had defects and which ones did not. As a human process, this was not consistent nor effective, which led to defective blocks occasionally being shipped to clients, resulting in major penalties and knock-on effects for their production facility.
The operators only captured primary defects, and subsequent defects were not observed. As a solution, DataProphet developed and deployed a predictive and prescriptive solution that is powered by state-of-the-art machine learning algorithms to identify engine blocks that would go on to be defective as well as identify the optimum operating regions for maximum yield with variable process parameters.
AI is building capabilities in autonomous manufacturing, building from process automation, to reduce non-quality and build value in the enterprise. This leads to greater agility in meeting the competitive demands of the ever-changing markets and leads to a quality and customer-driven process which, in turn, generates greater levels of customer satisfaction and strengthens customer relations.