A resource for safe and effective troubleshooting from the leaders in simulation training.
Manufacturing data and how to cut through the noise
Welcome back once again, Troubleshooters! Thanks for joining us today on Troubleshooting Thursdays.
Last week we began our series of blog posts on big data in manufacturing, what to do with all of the manufacturing data being generated in the modern smart factory. There’s no doubt that harnessing the power of new data streams is absolutely imperative for manufacturers to compete and thrive in this brave new world of the digitally connected factory.
However, as Industry 4.0 unfolds, each new piece of production machinery has ever more sensors that generate an ever-growing torrent of data. Manufacturers who heard the siren song of data are now finding that it can very quickly become noise—noise that could be blocking out the relevant signals that would translate into greater efficiencies and higher profits.
A 2016 MPI study found that 44% of manufacturing executives surveyed said their biggest hurdle in using IoT data was that they didn’t understand how to apply it practically to improving their operations and products. Today, we’re going to look at how to manage manufacturing data so that it can be translated into increased efficiencies and, ultimately, a better bottom line.
The difference between data and information
Taylor Milner and Molly Tracy, analysts at Stroud International (a professional services firm that specializes in “driving breakthrough improvements in operations and capital projects”) believe that there actually is such as thing as too much data. The main risk is that the overload will desensitize manufacturers to the point where they miss the important signals.
Milner and Tracy argue that there is a big difference between data and information. Not all data is usable information. It’s possible to have a stream of data that doesn’t tell you anything that will actually help you improve your production processes. It might for example indicate correlation but not causation. Data has to be transformed into information in order to be meaningful and useful. According to these experts, data tells you there’s a problem, but not what to do with it.
As an example, they cite the case of one of their clients’ factories, where multiple alarms were always sounding. There were more than the maintenance department could handle. Stroud consultants stepped in and started looking to reduce all of the noise (literally). They divided the alarms into categories from most serious to least serious based on the risks they posed to the production line. Stroud first tackled and resolved the underlying problems that were causing the most serious alarms, taking them out of the equation altogether. They then increased the system’s tolerance for the least serious alarms that posed very little risk. That way, the maintenance teams could easily handle the remaining critical alarms whenever they did come up.
Leveraging the relevant manufacturing data
According to Milner and Tracy, there are a few key steps manufacturers need to take to avoid becoming overwhelmed and to be able to leverage meaningful data:
- Decide what information you need from your data before you begin the analysis.
- Realize that data is different from information.
- Prioritize which alarms to fix first.
- Don’t hide behind an electronic dashboard. Go out to the floor and smell the problem.
We would add that it is critical for VPs, directors, and plant managers to strategize—to isolate and prioritize their goals for what they want to accomplish with their data in the next 5, 10, 15, or 20 years. Output, maintenance, training, HR, and logistics systems will generate a lot of data that will have to be sifted for meaning. It’s necessary to have a big picture in place of which areas will yield the greatest gains, and what order to tackle them in.
Next, they will have to have a plan for collecting and analyzing the data (a fine-tuning of their strategy). This may include streams such as data from an electrical troubleshooting training solution like the Simutech Training System , which also generates relevant data on maintenance trainees’ progress, enabling HR managers and training program admins to quickly analyze who’s ready for the factory floor, and who still needs further training. Leveraging this data can lead to huge savings by reducing production line downtime, which can cost thousands of dollars per minute (or more).
Big data needs big analytics
It’s clear that the amount of manufacturing data coming into the state-of-the-art, digital factory is only going to increase. At some point, the sheer volume will surpass the ability of mere mortals to sift through, and big analytics will necessarily be part of the solution. Big analytics simply refers to the computer technology—a cognitive computer platform—that can independently and instantly analyze and make decisions about a flood of data coming in from multiple sources. These are programs that can adapt, learn, detect trends and patterns, and, in some sense, think. Managers and VPs need to be thinking and researching these analytics to prepare for the day when they can’t keep up with this manufacturing data themselves.
Whether it’s AI or human beings searching for the signals in the noise, looking at the patterns and correlations and finding useful associations and root causes, there needs to be a management plan in place that involves defining clear data goals and strategies for accomplishing them. In the coming weeks, we’ll be looking at these steps in depth to help manufacturing decision-makers plan for the management of their data, both now and in the future, to prepare for the tsunami of manufacturing data that is headed their way.
And that’s it for today, Troubleshooters! Join us next week on TST for Part 3 of Cutting Through the Noise: Managing Big Data in Manufacturing.
Follow Troubleshooting Thursdays
Tune in to Troubleshooting Thursdays for reliable tips, general troubleshooting process, and industry insights. Stay up to date with Simutech Multimedia:
- LinkedIn: Simutech Multimedia Inc
- YouTube: Simutech Multimedia
- Instagram: @simutechmultimedia
- Twitter: @SimutechTrain
Have a subject you would like Troubleshooting Thursdays to cover? Send Simutech Multimedia an email at [email protected].
Looking to give simulation learning a try? Get started with our award-winning first solution here: Get Demo.