Michael Taylor, a mechanical engineer and project manager with the Manufacturing Extension program at the US National Institute for Standards and Technology, wrote in a blog posting that digital applications are becoming frequent in manufacturing. While implementation of such techniques may be daunting for small enterprises, due to initial purchase price and training costs, he recommends five applications: digital performance management; predictive maintenance; yield, energy, and throughput analysis; automation and robotics; and digital quality management.
What does it mean?
The article briefly describes each application:
- Define important performance metrics, devise ways to collect the metrics in real time, and display them in a digital dashboard. Start with machine operating data or production output, as examples.
- Predictive maintenance uses equipment condition, often monitored by sensors, to detect the need to perform maintenance before machine failure.
- What is your yield for each step and for the entire process? How much energy are you using? How can each step in your process be improved? Digital applications can help you answer these questions.
- Automation and robotics are becoming more common, with turn-key solutions more available. The NIST article advises starting with applications such as low-speed material handling.
- Take the performance metrics identified in the first application and tie them directly and automatically to decision making.
What does it mean for you?
Three key ideas underlie the NIST suggestions: industrial engineering, vendors, and data.
The author of the NIST article is a mechanical engineer. My field, industrial engineering, has historical roots in mechanical engineering, but adds considerations of efficiency, quality, and safety to the design of mechanical – and electrical – devices and systems, especially in manufacturing enterprises. Many of the manufacturing trends of the last decades are central to industrial engineering, including performance and productivity measures, preventive and now predictive maintenance, process improvement and optimization, automation and robotics to improve the operations of a manufacturing system, and automatic, data-based decision making. Your organization will have an easier time implementing the applications suggested in the NIST article if you have industrial engineers to help you.
You can get industrial engineering help by hiring an industrial engineer, of course, but often vendors can be your friends. Yes, a vendor is trying to sell you some technology, hardware, or software, but a good vendor wants to make sure your organization is successful in implementation. Many companies use engineers, often industrial engineers, in technical sales positions. Even if the sales person is not an engineer, a good vendor will provide training opportunities for your staff, access to web resources, and someone on their staff to help you install and operate the new application. If the vendor isn’t offering such support, try another vendor. Always remember that you aren’t just purchasing a physical device; you are also purchasing the support and service that comes with it.
Quality guru Deming is well known for saying; “In God we trust; all others bring data,” and I am much less well known for saying: “I love data.” Deming also said “Measure, measure, measure.” Data are your messages from the real world. Using data requires that you choose what to measure, design a system to collect and report those data (collect from where? with what devices? measured and reported at what intervals?), and, most importantly, use that data to make better decisions.
I cannot shout this final point loudly enough: data are useful because they help you make better decisions. While this point is explicit in the fifth suggestion from NIST, it is implicit in all of the suggestions. And this point has many implications. You probably shouldn’t bother to collect some data if they won’t be used to make decisions. Using data to make better decisions involves a lot of work to understand the relationships among the data, the real world system they were collected from, and the decisions your organization makes that affect that system. You need a good model of those relationships.
The NIST articles rightly focuses on getting started. For all organizations, collecting data and learning to use them to make better decisions is the path to quality improvement and to better decision making.
Where can you learn more?
Yes, “data” is a plural noun.
I used the word “dashboard” in the title of this post because many use that phrase to describe the types of applications recommended in the NIST article. An Internet search on the phrase “manufacturing dashboard” yields much good advice. See, for example, 6 Manufacturing Dashboards for Visualizing Production.