In a day in age where constant innovation and continuous improvement are a must to stay at the forefront of competition, automation in just about every sector is more prominent than ever. Automation is the core of Machine Learning (ML), which is ultimately a subset of Artificial Intelligence (AI). We are slowly but surely heading towards a world that leverages AI in just about every aspect of our lives. 

With this progressive dynamic in mind, we thought a focus on automation and some of the domains that aim to monitor, control and improve the production and delivery of products and services we consume on a daily basis would be worthy of exploring. Over the next three months, we will be featuring a series in which we key in on various industries and their respective implementations of ML and AI and how these platforms enhance efficiencies. 

This month, we are taking a deeper dive into manufacturing processes, an industry who has been an early adopter of AI with the adoption of Manufacturing Execution System (MES) solutions.

Specifically, we will be taking a look at manufacturing metrics of production processes and output via an Information Technology (IT) foundation. 

Creating Value in Manufacturing with an IT Solution

By: Matt Holka

A Manufacturing Execution System (MES) is generally understood as an Information Technology (IT) solution used to monitor and control complex manufacturing processes and the data associated with them. While various industries might implement MES solutions for different reasons, the system’s benefits usually boil down to allowing an organization to have greater control of their manufacturing process and improving production output. With an MES platform that is fully integrated into factory floor equipment, an organization is able to gather critical production metrics in real time and make informed decisions efficiently. These metrics include but are not limited to; equipment uptime, machine & operator rate, work station yield, active schedule enforcement and part specific traceability. By leveraging the vast amount of information that is generated, an organization has the ability to reduce time spent manually generating reports and paperwork, increase inventory accuracy, improve schedule and shipment attainment, and increase customer satisfaction.

There is an adage that says, “What gets measured, gets managed” and this is never clearer than in a manufacturing plant. Daily production meetings and report outs are filled with various numbers and metrics that are being tracked, measured and micromanaged throughout the facility, sometimes as often as in 15-minute intervals. We can consolidate a majority of these metrics into a couple Key Performance Indicators (KPIs) that directly impact overall business performance: Overall Equipment Effectiveness (OEE) and Throughput Yield. 

Overall Equipment Effectiveness (OEE) is a convenient way to compare specific equipment or complete production lines against one another since it takes into consideration multiple metrics that are likely already being measured and combines them into a single and easily understood “score”. The simplest definition of OEE is what percent of time is being used productively in a specific area. A value of 100% OEE would imply that a machine is running at maximum speed for the entire shift and only making good parts. All three metrics that are used to calculate OOE are easily tracked by use of an integrated MES, they are; Equipment Availability, Performance, and Quality.

The first metric used while calculating OEE is Equipment Availability, which is the ratio of Actual Run Time verses Planned Run Time. Planned Run Time is the duration in which the equipment was scheduled and staffed. Let’s say that for an 8-hour shift, the operator has two 20-minute breaks, this means that the Planned Run Time for that machine is 440 minutes (480 min Total – 40 min Breaks = 440 min Planned Run Time). Actual Run Time is the time in which the machine was actively producing parts, taking into account Unplanned (machine failure) and Planned (changeover) downtime events. We could say that during a shift we had 20 minutes of setup time after a changeover and the equipment was broken for an additional 60 minutes at some point during the shift. In our example, this would mean that our Actual Run Time calculates to 360 minutes (440 min Planned – 20 min Changeover – 60 min Downtime = 360 min Actual Run Time). With these two numbers we can calculate our equipment availability by dividing the Actual Run Time by the Planned Run Time, in this example our equipment availability would equal 81.8% (360 min / 440 min = 81.8%). 

Calculating equipment availability is fairly straight forward, the trouble in manufacturing comes in when we start to rely on human operators to accurately record dozens of different metrics throughout the day, such as the planned and unplanned downtime events. During a shift, it is very easy for an operator to lose track of the exact time their equipment failed since they will likely either be trying to troubleshoot the issue or contact someone who can repair it. At the end of the shift when their paperwork is turned in, it might say their machine was down from 10:30am to 10:45am for a total of 15 minutes. By interfacing the equipment directly with an MES, we would be able to see that the machine actually went down at 10:22am and wasn’t running again until 10:57am, giving us an actual total downtime of 35 minutes. The difference in Equipment Availability in this situation is 96.6% with operator reporting vs 92.0% with the MES recorded downtime. You can see that with this scenario and only one downtime event, management might look at the report and think their equipment is running better than it actually is. This type of misinformation could lead to the problem-solving team looking in the wrong location for why a specific shipment might have been short. In addition to being more accurate, the MES platform will free up the operator to focus more on making good parts than trying to keep track of the start/stop times of the downtime events that happens throughout their shift. Since the MES is integrated directly with the equipment, we can be certain that the data generated from the system is accurate and reliable.

Equipment Performance and Quality are the two remaining components required to calculate OEE. Equipment Performance is simply, “how many parts did a specific machine make in a given timeframe” or “how fast can a machine make parts”. While Equipment Quality on the other hand answers the question “how many of the parts that were made were actually good parts?” In the future, we will explore how a fully integrated MES takes the guesswork out of these metrics as well as diving into two of my favorite topics; Part Serialization and Traceability and how they relate to throughput yield.

If you have an emerging or innovative AI or ML topic you would like to contribute to our future newsletter content, please contact Adrienne Moulton, Marketing & Communications Manager at [email protected]

Matt Holka is a PMI Certified Project Management Professional and has several years of experience in quality assurance and process improvement in pharmaceutical and automotive manufacturing industries. Currently, he is leading the design, development and deployment of MES solutions as part of a team dedicated to process improvements for a West Michigan division of an international automotive parts supplier.