It’s Not What You’ve Got, It’s What You Use That Matters


Michael Ford, Sr. Director Emerging Industry Strategy, Aegis Software

Data Abstract
Data Abstract

An unusual occurrence, or is it a daily thing? The reports show all the machines in the factory are performing well, but it seems the factory itself seems to be in a coma, unable to fulfill crucial delivery deadlines. Some managers are asking for more automation to fix the issue, while others call for an improved machine data analysis detailing the problems and where they occurred. Is digital technology the answer to factory data management, or is it just adding to the issues?

The use of machine data is not necessarily a good or bad thing. I could go into any factory that is currently closed due to Covid-19 and pull production reports for the last few months which depict that there was no scrap, parts were not lost, lines were balanced, zero-defects, all deliveries were made on time, and there is no indication that any machine had a decrease in output. There was zero-time lost due to operators needing bathroom breaks, no employee lateness' or overtime, and zero incidents. In this situation, these statistics may be actual, but they are worthless. It just proves that you can rationalize anything, depending on how you look at the data. Albeit this example may be a bit excessive. This practice is quite common. After all, everyone wants to get a good annual review so that results can be presented in a positive manner; however, simple, isolated metrics that are pulled out of the overall report can be deceiving. The financial costs of this start off slow but substantially increase over time, only showing signs of limitations and concealing serious operational deficiencies.

The data needs to be examined better. Two systems that are excellent at automating process improvement from raw machine data are machine learning and line-based closed-loop systems. Other than these limited solutions, analyzing just the machine data in isolation is rather useless. By now, we would anticipate that automated machines perform well, when possible and deliver favorable results. The difficult task is figuring out how to analyze what is taking place between the machines, where there is zero information available for analysis.

If we look at things in the most basic way, zero data equates to potential loss. Machines come to a halt, possibly because they are jammed, empty, not needed, malfunctioning, or there might be problems with the quality of material, or maybe there are not enough operators due to vacation or pandemic. The machines do not give any insight into why they are stopped. They only indicate the stoppage. What must be explained are the cause and effects of any crucial exception during the operation that stopped the process by connecting the data from several machines and various controls, such as material logistics, planning, and quality management (QMS).

Then, if we examine things in a more detailed manner, we ought to consider the product's advancement during the manufacturing process, instead of just examining how the machines are functioning. Take the average international traveler's airport experience. Exactly how much time is required to check-in, drop a bag, go through the security line, walk to the gate, and board the plane? Most likely, 10 minutes from beginning to end. However, we are instructed to arrive at least 2 hours before departure time. The actual amount of productive time at the airport is roughly 8%, and the other 92% is time wasted, although shop owners may feel otherwise. An 8% manufacturing efficiency rate would most likely result in job loss; nevertheless, I could pull reports from most factories operating at a normal capacity that indicate much less than 8% efficiencies. Instead of truly evaluating the factory effectiveness of creating finished goods out of raw materials, we are preoccupied with the machine data. Raw materials supplies should be reduced, as well as keeping sub-assemblies, semi-finished goods, and finished goods in the warehouse. There should not be volumes of products pending repair or re-test, being fixed or tested, quality inspected, quarantined, or just stockpiled for processes that are not ready to begin. All of these things have a bigger impact on manufacturing than how a particular machine is performing.

Obtaining machine data has now been transformed, with the ability to collect factory data from machines much easier, faster, more precise, and detailed, without using middleware or custom integrations, especially when utilizing the IPC Connected Factory Exchange (CFX) standard.

Unfortunately, there are no connections for the intervals between the machines, which affects production the most. For instance, let's look at an individual product moving from one process to another. The goods reach the end of the line and come to a halt. It is then stored someplace, waiting for the rest of the job to finish.

The next stage needs to be as efficient as it can. As a result, production planning pushed back the start time until all products were completed in the prior process, and there was an opening available and at an optimal time. This period could take minutes, hours, or even days. Thereby increasing the potential for mishandling, added delays due to faulty equipment, lack of materials, engineering changes, contamination, possibly leading to more inspections, cleaning, processing, additional interruptions, extra storage. The ability to digitally track the products throughout the assembly process generates more ways to increase productivity and reduce costs. To accomplish this, the data from the individual machine and factory processes also need to be used to create a real-time virtual "movie" of all that is going on in the plant, a live manufacturing "digital twin," not just a simulation.

It is unnecessary for software to generate elaborate 3D visuals and graphics to harness a rules-based engine that fully contextualizes data from machines, materials, quality and Planning, line configurations, and product to make a digital twin that encompasses the process in its entirety. The manufacturing digital twin utilizes near-term performance history to discover what works properly and the current state. It also looks ahead to analyze current trends to uncover any potential issues that could occur by making adjustments and decisions beforehand. Ultimately, this rules-based manufacturing digital twin is guiding the production process in its entirety, providing complete visibility and automatically avoiding obstacles, focusing on the essential business requirements and improvement opportunities throughout the plant. This is not the average MES (Manufacturing Execution System) solution. It is the enhanced, cutting-edge IIoT-driven MES solution, designed explicitly for the rules-based digital twin.

Acquiring machine data from the entire factory is just the first step towards improving the manufacturing process using digital solutions. What you do with the data is much more important than merely having it, creating dashboards, performing machine learning, and analyzing it. The real IIoT-based MES digital twin for manufacturing has arrived.

Sign up for our blog

Stay up-to-date on the latest in manufacturing trends, insights and best practices.