If you engage in manufacturing, asset management, or logistics management, chances are high that you’ve already completed, are currently engaged in, or are getting ready to build an Industrial Internet of Things (IIoT) platform. Whether the core purpose of your IIoT solution is to gain better insight into your customers’ usage or to measure and track your business’s assets, IIoT provides powerful ways to get the most out of your data.
At its most basic level, an IIoT solution covers three fundamental aspects: ingestion, analysis, and visualization. The goal is to derive insights, predictions, and knowledge from the increased data acquired from assets in and around the working environment. The technologies behind this are not new, but the ways in which the information flows and can be combined and used change constantly. Let’s consider how each part of an IIoT solution works, individually and in concert, to deliver the highest possible value for your company and customers.
The intake of data is called ingestion. Think of the system as an organism that feeds on data to produce knowledge and insights. Ingestion has a variety of forms and is not limited to just device sensors: data can be acquired from financial markets, weather sources, email servers, and anything that has a data feed attached to it. An industrial ingestion framework must consider data integrity, device health, and data storage flexibility.
If the source of the data cannot be trusted, the data should not be used for decision-making. Modern cryptographic technology is pervasive on the internet-scale and the device-scale, and hardware to accelerate symmetric algorithm-based security is cheaply available. Both Microsoft and Amazon provide APIs and guidance for working with these industry-standard technologies and offer them freely. The device itself should be secured with a “Secure Boot” mechanism to ensure the integrity of the software running in the field. Compromised software may produce compromised results or, worse, it may affect other devices adversely as well.
Each device or data feed has a “health” aspect to it. As the health of a device deteriorates, its data may become less trustworthy. If its sensors go out of calibration, or its energy source becomes unstable, the values it produces could be questionable. For example, a simple temperature reading might drift if the battery cells on the device have deteriorated and are producing fluctuating voltage outside of the sensor’s operating limits. The device should report its health so that the downstream systems can alert the operator of a problem, but also to cast doubt on the measurements from that device if it’s faltering.
As data flows into the system, it must be stored somewhere. Cloud-based storage is available in several different flavors based on the access patterns of the data. If you don’t plan on reading the data at all except in rare cases, you can get a deep discount on storing even terabytes of data, but pay a higher fee only when retrieving it. Conversely, if you have data you need to access frequently, you’ll end up paying a higher price per gigabyte of storage, but paying less on the access fees for the data. This balance means that you can design your storage solution around your usage patterns, and there is no excuse for not storing raw device data. Keeping the raw data in “cold” storage gives you the flexibility to reprocess the data or transform it when you find a better way to present the data or want to copy it into an analytic workspace.
An IIoT architecture that considers these three areas effectively will allow you to transform the design of the system to handle future requirements without impacting existing data flows and while maintaining the security and integrity of the system.
Data gains tangible value once it has been transformed and analyzed by your application. This is an opportunity to provide insights back to your customers based on their personal use, and also gain insights on the state of your industry if you’re measuring components within a larger machine or if you’re gathering like measurements from various manufacturers. These insights could include average lifespan of equipment/asset based on components used, instances of incidents requiring maintenance, or performance comparisons of different manufacturer equipment/assets in like conditions.
Meta-measurement for maximum value
You may notice that all these examples require an analysis that expands beyond the actual measurements you are ingesting. The true value lies in the metadata associated with those measurements. Looking at things like work orders that stem from poor measurement(s), equipment/asset specifications, environmental factors, equipment/asset changes over time, and sales figures enhances the insights you are able to provide. Therefore, it’s not only necessary to gather edge device data: the greatest value will come from combining this data with additional details concerning the environment and conditions. This metadata-gathering may depend on the end-user supplying the metric (either one-time or at every instance of measurement), or integrating with existing applications. For example, knowing the rate a customer orders replacement parts for manufacturing equipment delivers much greater value when you can compare this to the rate at which other customers order parts based on the conditions, performance, and specifications of their equipment. Consider an example of the type of analysis you could conduct with proper metadata and insights derived from integrations with your cost and maintenance-tracking systems:
A $150,000 investment in newer equipment within a machining shop of medium size (metadata: “medium size” is defined as 10,000 sq ft, producing at least 1,000 components/ per day) has led to a 75% reduction in replacement part purchases and a 50% reduction in maintenance reports over a two year period. This reduction in parts orders and maintenance time produces a savings of $200,000 over the same period (metadata: average maintenance investment is one full-time employee for five hours, per request). Thus, an investment of $150,000 in equipment can produce a net return of $50,000.
This type of integrated analysis will result in actionable business insights and uncover the real value of an IIoT solution.
Notifications and predictive analysis
Integration doesn’t need to stop at receiving data into a single big data engine. You could program notifications that are triggered at preset thresholds to not only be received by a user, but to be sent to other systems to trigger an automated process. For example, a component experiencing dangerous level of heat can be set to not only notify the appropriate user, but also to trigger a high-priority ticket in a management system where the maintenance engineer can actively communicate with the individual reading the measurements. Measurements should also be captured so that trend analysis can be performed—data trends allow you to move past the realm of the reactive and predict upcoming events, and perform proactive maintenance or action on a given component or asset. Machine learning models can be trained to recognize potential failures and notify the operators ahead of time. Manufacturers can use this type of automated data to improve their services: Once the operational and maintenance data across customers has been aggregated, you can compare maintenance actions across the entire portfolio. This will allow you to create programs that drive higher performance from equipment/assets and support recurring sales.
Recent advances in If-This-Then-That (IFTTT) command integrations allow you to program actions across your equipment/assets that automatically shut down the equipment, reduce energy consumption, or reduce the risk of downstream issues. It’s become possible to tailor the performance of components so that only the absolutely necessary wear-and-tear can be applied at a point in time for a specific application, or based on its usage environment. This can increase the lifespan of the components and equipment/assets. These types of automated actions do require a machine learning module with a certain level of artificial intelligence.
Powerful insights generated by your analysis engine lose value if these insights can’t be easily digested by your end-users. Visualization should be developed around specific customer user stories, not in a generalized “show me all my data” interface. When confronted by a flood of data, end users will not always know how to interpret the results. Experts may be able to look at a chart of pressures and temperatures over time and immediately identify certain operations and features, but not all end users are experts. When designing a visualization for an IIoT solution, the user experience (UX) must be kept squarely in mind.
Consider the outcomes desired
A customer wants to improve the efficiency of a process or predict faults in equipment. Consider the hypotheses put forth by the process experts about where certain issues might originate. Proper visualization brings clarity to these ideas and enables the end user to more easily draw conclusions from the output of the system. During the UX design process, the experts and end users are interviewed to understand their goals and the product is designed to support those goals. A feature that cannot be expressed in terms of outcomes for a stakeholder will not be implemented, focusing the development team on creating deeper value for the customer instead of implementing shallow features.
Consider the user’s expectations
Users interacting with industrial systems require information to be organized in recognizable hierarchies. Familiar identifiers make it easier for the user to navigate the system and to anticipate where certain functionality may lie. If we use words like “site”, “plant”, “asset”, and “equipment,” the industrial customer will make assumptions about the various levels of the hierarchy. Industrial customers also have expectations on color schemes and symbols to reduce the cost of training new users for a system. Users expect modern software systems to be tailored to their workflows and to present the relevant information prominently.
Consider who needs to know what
Visualization also has to address the entitlements of users on the system because industrial data is a competitive advantage for customers. Keeping the data locked down provides two major benefits: it limits the exposure of the company by implementing the principle of least privilege and it simplifies the view of data and analysis for end users. When a user is based in a plant in Texas, they typically don’t need to know, or care to know, how a plant in New York is operating. However, a regional manager would need to see roll-up data for all the plants in his or her region. While industrial companies have similarities in their entitlement design, there are usually hidden requirements that arise from interviewing different levels of users of the system and understanding their respective workflows.
IIoT powers combined
As you walk (or run) down your own path of designing an IIoT solution, keep in mind how ingestion, analysis, and visualization all support the needs of your organization. While each component of IIoT has its own unique requirements, understanding how these feed into each other is how true value will be realized.
Ingestion + Analysis = Control
When you use your analysis capabilities to look at the ingestion process, you can optimize your bandwidth, tune your equipment, and predict failures in equipment. The ingestion process itself yields additional metrics on every data point acquired that can be analyzed to find inconsistencies and problems.
Ingestion + Visualization = Connectivity
Visualizing the ingestion process makes it easier for your operations team to support the IIoT solution by giving them the tools to see what is going on with the devices in the field and interact with them more effectively.
Analysis + Visualization = Compute
Visualizing the data analysis will enable you to better present a solution to the end users, but also to better understand how the analysis process works, allowing the teams to continually improve the process—and ultimately the insights you derive.
There is a natural progression from ingestion to analysis to visualization. However, the different combinations are just as important, and when considered together they become truly powerful and insightful tools. Stay tuned for a deeper look into the ways in which these powerful components can be combined to make both your solution and your business stronger.