The Internet of Things (IoT) connects the physical world of sensors to the Internet. Sensors deployed in heavy industries are categorized as the Industrial IoT (IIoT). The insights generated from IIoT can fundamentally transform industries and create far greater economic value than consumer IoT [5] [6]. There are many companies that aim to advance IIoT, so we’ve developed a framework that allows one to visualize the components.
The Industrial IoT market is large and growing.
Research analysts peg the value of the IIoT market at hundreds of billions of dollars ($150 billion — $250 billion) and forecast a growth rate of nearly 26% CAGR [1] for the next decade. Some [2] forecast that the manufacturing, transportation and information industries will each invest $100 billion in IIoT over the next five years. Among the growth drivers in IIoT are decreases in the cost of sensors, bandwidth, storage and processing. There’s also an increase in the adoption of handhelds, technology miniaturization and advances at the big data application layer.
GE has estimated [3] that a 1% improvement in efficiencies in just 5 industries — railways, aviation, healthcare, power and oil & gas — could save over $276 billion in the next 15 years.
A framework to figure it out
While IIoT is promising, less than 1% of the data collected by IIoT sensors is used today[4]. We meet hundreds of IIoT startups and use this framework (among others) to evaluate their products.
Abstract view of Industrial Analytics
Industrial Analytics is conceptually composed of Data / Knowledge Generation (the top row in the framework) and Analytics Applications (the bottom row in the framework).
Data / Knowledge Generation: gathering industrial data such as temperature, pressure, humidity, light and sound through sensors. The data collection processes labelled Streaming Analytics push to data management/historian systems in the form of event, time series or machine data. Such systems, labelled Analytical Databases, are designed to ingest millions of events per second and allow data aggregation, cleaning, labelling, storage, contextualizing, and real-time comparison of multiple streams. Analytical Databases, in conjunction with visual data exploration and monitoring systems, labelled Visual Analytics, enable discovery and learning from this data. This learning can be useful when reiterating or updating historical models, or identifying anomalies. When a specific anomaly or condition is identified in the ingested data stream, it may trigger an automated action (such as a digital adjustment) or alerts (as designed in the reporting system) as part of Analytics Applications (the bottom row in the framework).
Example: Improving airline operations
Planning flight route, capacity utilization, pricing and yield management is an approximate science because of the many variables in the equation (altitude, speed, weight, maintenance and the weather). Real-time data from thousands of embedded sensors, together with historical data and sophisticated analytics, are needed to precisely utilize resources and streamline processes. Any resulting operational efficiencies, such as an optimal flight plan, that save even 1% in jet fuel use would save airlines hundreds of millions of dollars per year, and the global commercial aviation industry $30 billion over 15 years.
We can better understand — and thus further improve airline operations — using this framework. Sensors on the airplane body measure airplane speed and altitude (Streaming Analytics) then streams of this data are compared against weather data recorded on the airline’s data historian (Analytical Databases). The optimal flight path that avoids rough weather is determined using visual tools on streaming and historical data (Visual Analytics). The system will recommend a correction to the flight path if any deviation is detected (via dynamic Digital Adjustment) to continuously monitor and save fuel consumption for this one flight. Data logs recording these changes (Reporting) can be harnessed over-time to learn data-driven efficiencies to save the airline millions of dollars.
The framework can similarly capture the key components of almost any industrial analytics solution. This allows us to visually map a solution and better comprehend the implementation challenges.
Framework allows us to grasp the challenges in industrial analytics
IIoT has the potential to transform businesses and industries. Analytics-driven processes with learning and feedback can significantly improve business objectives. This is a rapidly growing opportunity that is attracting several hundred companies to build solutions that reduce human intervention, and to eventually automate all the components.
The core challenge in industrial analytics is in automating every component of this framework. This is the first of a 2-part article. The second part will use this framework to describe the subset of challenges that we consider to be most immediate and impactful.
References: