Industrial Internet of Things: A spectrum of adoption
07 January 2022
The Industrial Internet of Things (IIoT) has progressed a great deal in the past few years, especially regarding adoption ease. Many machines and industrial equipment are already equipped with the capabilities to connect to an existing IIoT infrastructure. But implementing a new stream of data can be paralysing to the uninitiated.
Creating a successful IIoT system for a factory, or enterprise, requires time to learn what data is valuable and then slowly build an infrastructure to efficiently collect the most valuable information from your machines.
Another major hurdle for implementation is understanding how the data is processed. A vision system, for instance, may be included in an industrial machine purchase, but there must be a way to handle the terabytes of data it can produce in a very short period. Or the data being collected might be less voluminous but just as actionable at a large enough scale. Process data is only valuable if it has a purpose. This is why smaller initiatives with a short-term impact are important. Allowing for learning through limited scope projects with an option to fail before finding the right solution creates an atmosphere of trust with employees and the system.
Implementing an IIoT system can differ dramatically from company to company, and there are a variety of ways to start your project. Let’s take a look at a few important insights based on customer challenges starting out with IIoT and our own understanding of these experiences.
The first step in implementing a successful IIoT project in the factory is working from the current maturity level of the site. Those that have already been collecting historical data of processes and quality may be fit to start analysing more complex processes out of the gate. But those just starting to collect data through an IIoT infrastructure may want to start with the basics: tracking inventory, understanding machine efficiency, or logging energy consumption. As the team becomes more comfortable, they can continue to adapt and expand what is being tracked and analysed from the shop processes.
High data volumes
As more sophisticated data collection is implemented for the shop floor, tracking this data accurately and timely becomes a greater hurdle. And many of these data sets are far too complex to analyse manually. Furthermore, vision systems require algorithms and even machine learning or artificial intelligence to sift through the data fast enough for it to be actionable. This challenge extends to increasingly diverse data sets. For example, silicon fabrication facilities collect atmospheric measurements at each machine along with environmental conditions to optimise the manufacture of highly complex products. These data sets could even be used to improve the sustainability of the shop; high-energy processes could be shifted to off-peak energy times to prevent stresses on the power infrastructure or even as a means of optimising an individual machine’s energy efficiency.
The best first step to collecting data is to start small and learn with hands-on experience. This provides value in two ways. First, it reduces the impact on the greater factory by limiting the size of a new program. You could even start with a post-analysis, which has no impact in the moment on current operations. And second, starting small provides time to learn how IIoT systems function. Small projects at the beginning of an IIoT initiative compound with time. As more-complex functions are created, they often rely on existing and straight-forward data collection processes. Machine maintenance schedules may be influenced by the shop environment, the operating temperatures within a unit, or the rate of production of the machine. Each of these data sources could have been collected in small pilots and integrated later for this more complex task.
Time to fail
Possibly the most important insight to remember when implementing a new IIoT initiative is to allow time and space to fail. Failure is not the goal in a new project, but assuming success on the first attempt is a very costly assumption to make. By building in a buffer area for failure, required production will not be impacted to the point of total business failure. And oftentimes the best learning opportunities arise from a system failure. For instance, in the vision system example, users may want to learn from the amassing data and tweak the amount that gets analysed. If a camera is outputting so many frames per second that it is overloading the processing devices, but the number of required images to complete the task is far lower, the failure in a safe environment enables a better overall system and a greater understanding.
Wherever your company is on the path to digitalising with the Industrial Internet of Things, there is a solution for you. Starting from nothing is all about creating an infrastructure that allows you to collect the data and pull the low-hanging fruit from the solution. But if there are already workflows collecting, analysing, and acting on the data, there is still room to expand the functions of IIoT without impacting current operations. No matter how you implement IIoT, it is critical that you incorporate it as soon as possible. The historical data that many higher-level functions rely upon needs to be collected and analysed before they can come to fruition. And as the data types expand into vision or other high data-volume avenues, the system needs to be able to handle it effectively. These seemingly small changes compound with time and volume into an expansive digital solution.
Contact Details and Archive...