Manufacturing organizations pursuing industrial Internet of Things (IIoT) initiatives have an ever-growing list of intriguing solution options to consider. And while many IIoT solutions make a lot of sense in concept, as with virtually everything in manufacturing, things are more complicated in practice. Early warning systems are a case in point. We’ve talked with many clients who are implementing or considering early warning systems, and the reality is that getting clear value from an early warning system isn’t a given. In many cases, early warning systems can actually lead to a lot of “noise” that people just end up ignoring. Let’s explore some of the potential struggles related to early warning system implementation and what it takes to get things right.
What is an early warning system?
For the purposes of this blog post, we’re defining an early warning system as any IIoT-based solution that is designed to provide advance notice about something going wrong in a process or on a machine. These systems typically rely on a combination of IoT sensors, the related data and analytics to predict or identify exceptions.
The dangers of warning overload
How many alerts do you need? And who will respond, and how, to different types of alerts?
These questions are fundamental to success with early warning systems because if you don’t have the right alerts and response processes in place, early warnings are useless. For example, alert tolerance is set too tight triggering alerts too soon; or unnecessary alerts are set up where the sheer quantity of notifications overwhelms and frustrates the people receiving them. The problem is that it’s not easy to answer these questions, so many manufacturing organizations gravitate toward a trial-and-error approach to determining and setting alerts starting from a “more-sensors-and-data-are-better” mindset.
This was the experience of one of our clients who worked with a third-party vendor to implement an early warning system. With their initial setup, their supervisors were getting upwards of 70 alerts per hour. With everything else going on, it was impossible for them to respond to or even review all of the alerts and they ended up tuning them out.
Key considerations for early warning system success
The example above is a simple one and we’re not arguing that you can’t get value from an early warning system, but we do counsel clients to use a disciplined approach for choosing and sharing alerts.
The best practices for selecting what to track in early warning systems are no different than those you should use for any other business metric. They start with systematically exploring areas where you see the potential for alerts to improve operational performance and deliver the most returns to your business. It’s important to take your time because you really need to work through and understand the drivers of a process capability before you can pick useful alerts. And the reality is that determining if there are correlations between things like speed, tolerances, temperature and humidity and the outcome of a process often takes careful analysis over a period of days, weeks or even longer. By taking your time and using analytics to understand the key levers that drive performance of a process from productivity, cost and quality perspectives, you can find a few key areas to focus on. If your analysis shows promise, then it’s time to answer some key questions around potential alert options, including:
- How could an alert for a specific metric help optimize processes or improve performance?
- Can you clearly measure the impact of tracking and responding to that metric?
- Who would get the alert for a metric and how fast would they be expected to respond (and is that realistic)?
At the end of the day, the key takeaway here is that your focus should not be on the warnings themselves. It should be on improving performance. Simply choosing alerts that show promise in improving performance is half the battle. Ensuring that the right people pay attention to them and are prepared to take timely action in response is the other half. Using a methodical evaluation and roll-out process will keep you on the right track. If you are currently evaluating early warning systems and need help creating the right process and metrics that drive improvements, give us a call. Our operations management experts have experience in helping clients implement technology in a practical way that creates value for your manufacturing organization.
VICE PRESIDENT, SUPPLY CHAIN & TECHNOLOGY
Ken is a 25+ year veteran of manufacturing, operational excellence and supply chain optimization. He believes that it is critical for organizations to address the challenges of strategic planning and use of big data in manufacturing. At TBM, Ken is actively leading the effort to our suite of services to include emerging technologies that improve productivity and convert complex data into information for improved decision making.