Data is the foundation of IoT value in manufacturing. It provides visibility into operations, enabling you to build rich models for condition monitoring, boost overall equipment effectiveness (OEE), and generate new revenue streams through connected products.
Manufacturers have made good use of data to monitor, optimize, and control production for decades. However, the amount and diversity of data—and the extent of its impact—are new factors. By 2020, to increase speed, agility, efficiency, and innovation, 80 percent of manufacturers will need to extensively restructure, placing data at the center of their processes.
This transformation will bring change not just to the production lines, but to the ways employees work. A survey by Forbes and Microsoft showed that 41 percent of companies will outfit firstline workers with wearable devices by 2021, and 40 percent will deploy RFID devices to these employees. Solutions such as these will enable workers to get in-context recommendations, understand how to optimize their performance, and know where to find equipment. Increasing automation will require upskill training so workers can master “the machines that run the machines.” At the management level, understanding and working with data will become a foundational skill across all types of manufacturing.
To save time, money, and headaches later on, it pays to think through the main challenges you’re likely to encounter on this rapid ride. They can be summed up as the “three Vs” of big data: volume, velocity, and variety.
Volume: taming the flood
As you connect more elements of your production and supply chain processes, data volumes will grow. Plus, you may find that you need to store some of it for longer periods in anticipation of future analytics. Cloud-based systems can greatly reduce the cost and complexity of big data storage. Using a tiered storage model that uses low-cost “cold storage” archives for infrequently accessed data, and higher-performance systems for “warm” or “hot” data, can deliver significant savings.
Komatsu, one of the world’s leading manufacturers of construction equipment, has found success in using the cloud to handle high volumes of data generated by its smart construction solutions. The company knew that its Smart Construction Cloud Service would store an enormous volume of data. At the same time, it would be used by construction companies, 90 percent of which are small businesses. By eliminating the need for customers to deploy infrastructure, Komatsu was able to keep costs low and bring smart construction to more locations.
Velocity: data at the speed of now
Velocity is the rate at which data flows into your information systems. Time-series data—data points captured and time-stamped at regular intervals—can result in the capture of billions of events in a relatively short span of time. The faster your system can ingest and analyze this data, the sooner you can act on the insights. Choosing a solution that lets you start small and scale easily, such as Azure Time Series Insights, will serve you well in the long run.
Anheuser-Busch InBev, one of the world’s largest manufacturers operating in the food and beverage space, is using a streaming analytics platform to track inventory with unprecedented precision. “This gives us very microlevel data, so that we can identify every pallet and every case and know where it was brewed, when it got to the wholesaler, all the way to when somebody bought it and it left the store,” says Chetan Kundavaram, global director at the company. “We capture and process real-time data using Microsoft Azure IoT services, and the data feeds into our global analytics platform so we can precisely track the movement of inventory around the world.” Not only is this a smart application of real-time data ingestion, it’s also an example of how companies are looking beyond the production line to create efficiencies throughout the manufacturing value chain—in this case, with an intelligent supply chain.
Variety: data diversity as an advantage
The third data challenge is variety. In early stages, you may be working with IoT data from one source or a small number of sources. However, as you follow the value chain and expand your IoT capabilities, you’ll find that the variety of data can quickly grow. Standardizing data can allow you to use more if it, leading to more reliable, actionable outcomes.
For example, one major IoT innovator, Rolls-Royce, ingests many types of data in its engine-as-a-service solution: snapshots of engine performance that planes send wirelessly during a flight; large-volume downloads of “black box” data; technical logs; flight plans; and weather data provided by third parties. Making all this information available to analytics workflows enables Rolls-Royce to monitor engine condition, predict maintenance needs, and optimize fuel usage for its customers.
Looking broadly at types of data you might use can also inspire new solutions. Consider how data from industrial controls, business applications, sensors, wearables, GPS devices, audio/video capture devices, CRM, ERP, and even social media could add value to your solution—and choose a flexible platform that can handle a wide range of data types throughout the IoT lifecycle.
The three Vs illustrate some of the challenges common to IoT manufacturing initiatives. What are some of the ways to reduce these complexities?
Keep your options open
Open standards can prevent data silos that hamstring IoT value. Such standards simplify connectivity, interoperability, security, and reliability. That’s why Microsoft strongly supports open data standards such as OPC UA, which are essential to creating factories of the future. In fact, Microsoft is the top contributor to the OPC UA codebase. In partnership with SAP and Adobe, we’ve created the Open Data Initiative (ODI) to enable a single, comprehensive view of your data from all your lines of business, across all your systems. We’re also working with partners such as BMW to create an open technology framework and cross-industry community known as the Open Manufacturing Platform. Getting locked into a proprietary data format is no longer necessary for IoT innovation in process or discrete manufacturing.
Protect data from threats
With increasing dependence on data comes a need for better digital security. Increasingly sophisticated attackers will look for weak points at all levels of your solution—devices, networking, and cloud services. Protecting each layer using point solutions can quickly become cumbersome. With the potential for data to exist in many locations at the edge and in the cloud, manufacturers will need technology that provides end-to-end protection. This can greatly simplify security operations by giving you “one pane of glass” to identify and mitigate risks.
Data on the edge
The cloud is essential for big data storage and analytics that drive IoT value in many manufacturing scenarios. However, it’s increasingly apparent that a hybrid model allowing data to be processed and acted upon at the point of need is the right choice in many contexts.
Take the example of Bühler, a food-processing equipment supplier that uses IoT and machine learning to reduce waste while making foods safer to eat. Its LumoVision system identifies and eliminates individual kernels of contaminated grain at the rate of 15 tons per hour. This has the potential to dramatically reduce the amount of aflatoxin, a powerful carcinogen, that enters the food supply, while cutting food waste and energy consumption. The company developed the powerful models required for this solution using cloud computing—and deploys those models at the edge to enable a real-time precision process.
Data processing capabilities at the edge can help you minimize three-V data challenges, maximize data quality, and cut costs. In many situations, only a small minority of IoT data is worth storing for later analysis. Locally installed, pre-trained models are one solution for separating out the meaningful data for delivery to the cloud. Another solution is Azure Stream Analytics, which can combine with other products in the Azure IoT platform to provide a complete solution for ingesting, processing, visualizing, and acting on data. Finally, Azure Time Series Insights provides an option for end-to-end IoT analytics.
Data has the power to transform manufacturing—and not just on the factory floor. Next time, we’ll discuss how it can apply across the enterprise to help you create new levels of value.
Read the whitepaper and learn how to turn data into insights and discover Azure IoT solutions for discrete manufacturing.