The global disruption from COVID has caused many manufacturers to fast-track robotics efforts to improve resiliency and create production processes that can continue to function with less hands-on human dependence. This new wave will, in theory, focus on so-called “collaborative robots” that can perform multiple tasks and can work together with other robots in less structured work environments. This is a wonderful goal, but it belies the profound complexity of most manufacturing facilities. Without systems to visualize and contextualize down to the most granular level the most relevant data for a plant, you risk creating automation in a vacuum that hides potential problems from view and frustrates remote operations and coordination.
Industrial Automation Is Rapidly Accelerating
Investment in advanced manufacturing, like industrial robots and AI, has risen quickly around the globe. Many studies have associated robots with increased productivity. In recent years, however, total factor productivity has actually stagnated or declined. This mirrors some other areas of the economy where technology is making big inroads but not yielding better outcomes, such as the health care system and even office work. Economists believe that there is a lagging period during which organizations learn to better utilize and reorganize around novel technology. In the Industrial Revolution, factories did not reorganize to their current flat-and-long configuration to take advantage of the economies of electricity and controlled power sources for several decades, replacing the old vertical integration required when the power source was flowing water and proximal belts. The key shift happens when organizations start to understand how a system is changed by a new technology in totality, not just the impacts of the local changes. The same is true with robotics and advanced automation.
How Robots Might Negatively Impact OEE
Building a fully automated production line with robots that enjoys less downtime is a double-edged sword. It can mean that you produce more goods and can operate more consistently, even in the face of COVID-like disruptions. However, in reality, automation has a broad potential for unintended negative consequences. Manufacturing optimization is a tradeoff between four items: quality, cost, rate, and flexibility. Improving uptime (rate) through automation can have a negative impact on the others.
- Quality: Automation can make it more difficult to detect quality problems; or, even if it takes exactly the same amount of time to detect a problem, if you’ve made more product during that time (through a reduced cycle time), there’s more defective product that requires rework or scrap.
- Cost: the cost of installing automation in the first place may not be returned in a sufficient time to justify the investment.
- Flexibility: like humans, automated systems also need to be trained, and by people with a more technical skill set. Each new product requires retraining the system at potentially significant expense. This also comes into play if, during a crisis, demand for a new product suddenly skyrockets and a line has to be completely reconfigured.
Without granular sensing, automation can make it far harder to detect and identify quality and other problems that come from the complex interplay of factors in every factory: speed of line, environmental conditions, and material variances from suppliers. Even sensors have limitations compared to the human eye and hand. There are also the subtle but often crucial variances that everyone who works in manufacturing knows are par for the course. When slight variances multiply, they can degrade a production process and create bad outcomes, like glass that breaks or paper and packing that is not sturdy enough.
To Make the Big Leap, Operators Need Observability and Context
To understand how growing numbers of robots are interacting in a plant and impacting processes, you have to be able to see the processes as data. This is part of the promise of IIoT—but only part. The data must also be properly structured and wrangled so that all the actions of robots and automated systems on a line can be viewed both discretely and comparatively: as a whole and as thousands of tiny variables. Each part or cycle or unit can then be tracked as it moves down the line or out the factory door into real world use. Without this system-level view, adding more automation and robotics may actually make it harder, not easier, to match outcomes to inputs and processes. With this system-level view, coordination becomes possible—and even inevitable.
Automated systems with good visibility and observability can also collect and process information that humans can’t. For example, it’s easy to tie information about the ambient environment to a problem (defect/downtime) analysis. With automated systems, process engineers and data scientists can interrogate and test hypotheses on datasets much larger than humans can hold in their heads. If the system can recognize a time that the process was similar and quality was high, it can adjust its settings accordingly. A human may not realize that 14 weeks ago, while running the same product, the weather was the same but quality was higher.
In fact, good visibility can lead to conclusions that human intuition doesn’t. While expertise can be a powerful place to start—for example, knowing that the cooling rate affects the brittleness of the product—there may be other factors that can only be observed by sensors and associated by the background system.
Automate, But With A View Towards A Greater End – Observation and Contextualization
The nature of modern manufacturing is testing hypotheses to find the root cause of problems and identify ways to improve quality and efficiency. Forward thinking manufacturing companies already understand that adding automation in a vacuum may improve performance of a specific asset but the automation may contribute very little, and even may detract, from facility-wide productivity and quality. In the Industrial Revolution, they finally realized that building vertical factories where workers still had to walk up and down stairs and production was fighting gravity was not a great way to leverage the power of combustion, steam and electricity. In the Robot Revolution, we need to keep the big picture in mind—namely, how all the pieces of our manufacturing environment are interacting and how to measure that—if we want to escape the productivity stagnation of the past decade.