Where to start with industrial analytics
The value of industrial analytics is in the results they deliver. For a brewery, analytics use delivered a 60% throughput boost. For a pet food producer, it resulted in more than $800,000 in annual energy-cost savings. For a cosmetics plant, it meant a 90% reduction in line stoppages.
Without a doubt, the ability to collect raw data and turn it into useful information for workers is essential to staying competitive. But what if you’re unsure of how to bring analytics to your operations? Where are you if the goal is to deploy analytics, but no defined strategy is in place?
Fortunately, implementing analytics doesn’t have to be a mysterious or overwhelming task. Those that have already made the journey can provide key lessons for how and where to begin.
What you need
Industrial analytics is built around four core elements:
People: Data scientists are vital to deploying analytics. Everyday use of analytics, however, should not hinge on these specialized workers. Today, analytics technologies work almost as a data-scientist-in-a-box to structure data at the start. Then, non-data specialists freely access, manipulate, and analyze the data. This "self-service" analytics approach allows almost any worker to use data to solve various issues.
Data: Identify the data wanted and where it will come from. Some of it may come from sources not directly accessible to you. But most of it will come from the set of the connected technologies known as the Industrial Internet of Things (IIoT). These technologies can include device-level components like sensors, gateways, actuators, and drives; machine- or line-level components like controllers; and enterprise-level components such as software systems.
Connectivity: Analytics requires seamless connectivity across a plant floor. This means unifying any disparate systems that create "islands" of information. It also means having bandwidth for both current and future traffic needs. Standardizing a plant-floor network on a technology like EtherNet/IP helps achievement of real-time control and information. Pre-engineered network products and services, like industrial data centers or infrastructure-as-a-service (IaaS) offerings, can also reduce network design and configuration time.
Software: Look for analytics software that delivers best value to the organization. The software’s most basic job is to add context to data. Context allows combining and comparing different data to get deeper operational understanding. Instead of getting an oven temperature reading, for instance, data contextualization delivers that reading at a certain time, for a specific recipe, and during a given labor shift. That allows tracking key performance indicators (KPIs) and the factors that contribute to them.
Analytics software should do more than contextualize data. For example, it should allow users to drill down into specific analytics to investigate anomalies or troubleshoot issues. By means of machine learning it can even monitor operations and trigger automatic control adjustments, if a process falls outside allowable parameters.
The four elements outlined above form the foundation upon which analytics will be built and executed. But there also are also some key decisions to make early-on to maximize the long-term success of an analytics approach.
An analytics investment can pay for itself. It’s important, however, that the investment be tied to specific business outcomes, and is not merely be about the technology. Build a strategy for investment around a specific need, such as improving overall equipment effectiveness (OEE). Then determine the OEE boost required to meet your return on investment (ROI) goal. Those gains can be used to fund subsequent analytics initiatives.
Security isn’t something added to an analytics strategy or network upgrade. It should be holistic, extending from edge devices to the enterprise. Start with a security assessment to identify risks and potential threats. Then deploy a defense-in-depth security approach to guard against threats on multiple fronts. Work only with trusted suppliers that support security goals.
Existing industry partners play an important role in an analytics strategy. On-machine analytics from an original equipment manufacturer (OEM), for example, help maximize the performance and durability of production assets. With just a gateway device installed on a machine, the OEM can deliver analytics via cloud-based applications.
Finally, make use of available services and resources. Their use contributes to getting the most value from an analytics deployment and alleviates some challenges being faced.
Reference architectures, for example, help in the design and implementation of a network upgrade. Training and certification programs equip workers with skills needed to design, deploy, and oversee a secure information infrastructure. Connected services support analytics deployment, and even take on key roles, such as remote monitoring of operations.
Now get started It has never been easier to deploy analytics. One thing is clear though. The foundation needed is a well-established fact. These technologies put the power of data into the hands of those who need it. Therefore, deploying industrial analytics shouldn’t be a question of: Where do I begin? It should be: When can I get started?
David Stonehouse is global consulting leader, connected enterprise services, Rockwell Automation.
This article appears in the IIoT for Engineers supplement for Control Engineering and Plant Engineering.
See other articles from the supplement below.