Manufacturers face a host of pressures to optimize plant performance and stay competitive—from knowing when to perform equipment maintenance to tightening inventory control. Data analysis has served as a primary tool to gain more visibility into operations, but many correlations are missed as data languishes in unconnected equipment or unused databases.

The emerging Internet of Things (IoT) will unleash even more data across an enterprise as additional sensors, actuators and other devices with built-in intelligence connect to the Internet. In fact, 80 billion Internet-connected devices are forecast to be in use by 2024—more than quadruple the number of current connections.

“It presents a big opportunity for manufacturers to understand how things like schedule information, material sourcing and equipment diagnostics play into the lifecycle of the products they produce, and how they can better service customers over that lifecycle," says Chet Namboodri, managing director, Global Manufacturing Industry at Cisco.

The key to unlocking this opportunity rests with advanced analytics—a process that evaluates the resulting “big data" and other information sources to find previously hidden connections and derive valuable insights. The challenge for the industry, however, is learning how to convert these insights into action that is valuable to both front and back offices.

Finding Value in Analytics

The initial challenge with “big data" is understanding the terminology. The definition is subjective, ambiguous and often in the eye of the beholder.

“To some people, big data is looking at a large set of their end customers, reviewing their buying patterns and determining how to stock the retail chain," says John Nesi, vice president of market development for Rockwell Automation. Others, he says, look at big data in the context of running diagnostic analytics on fixed assets in plants to predict when the next breakdown might occur.

Because the big data nomenclature quickly popped into existence, many do not know what it entails, says John Larson, vice president and global leader with IHS Advanced Analytics. “Big data has been co-opted to really mean much more about the technology behind data than the insight from data."

John Larson, IHSNo matter how organizations define big data, they are accessing unprecedented volumes of data from multiple sources—everything from electrical loads on rotating equipment to third-party information to social media commentary. But almost all of it is meaningless without applying advanced analytics. In fact, the proliferation of data and the ability to analyze it is the result of two critical developments: increased storage capacity and rising processing power.

“In the past, you had a sense of what was out there, but you could not maintain all this information," Larson says. Now, with a fall in data storage costs and the advent of cloud computing, “you have an ability to maintain every single data point, and every data point has value."

(Watch a video as John Larson discusses big data and highlights major trends, challenges and opportunities for industry.)

Meanwhile, more powerful, less expensive processors help users extract meaningful intelligence from the flood of data. This capability has revolutionized analytical tools such as machine learning, which is a subset of artificial intelligence that detects and extrapolates patterns in the data and predicts trends.

With this stream of information in real time, companies can simulate what is going to happen by building models on a population of data in real time rather than what may be inferred from a static sample.

“You move beyond predicting what is going to happen into the prescriptive world," Larson says. Not only can an organization find out what is transpiring and why, but it can also take real-time action to effectuate outcomes.

Real-time analysis often happens at the production level, says Rockwell's Nesi. “It's a matter of determining, 'Am I running as efficiently as I can today, at this minute, with the production order that I have in my hands right now?'"

Operations managers can make adjustments to production based on what they see at that moment “because they have the ability to do scalable computing at the edge of the network to instantly understand where bottlenecks, quality issues or other things occur that affect time-to-want for your customer," Nesi says.

Analytics Move Up the Chain

The insights extracted from advanced analytics may offer the most value in the context of the enterprise. That means fully integrating manufacturing with supply chain management and other business functions.

On the plant floor, manufacturing systems have produced significant amounts of data for years. However, much of it is discarded in the production process or remains shared only between machines. Supply chain decisions that do not harness this type of operational data may spell missed opportunities.

Manufacturers can take several approaches to linking often-disparate functional areas. Image source: DARPA IT software and service companies are developing platforms that specifically target industry, removing data from silos and analyzing it to find previously unknown connections.

Another solution involves narrowing the gap between operational IT and business IT, which often have different viewpoints on how the other operates and prioritizes tasks.

“When we get them to the table to work together for a purpose, things tend to align pretty quickly," Rockwell's Nesi says.

Crossing this chasm is both an organizational and a technical challenge as most manufacturers use several forms of middleware between machine control systems and enterprise resource planning (ERP) systems to handle workflows, quality metrics and product tracking.

“Many older systems are disjointed, unsupportable and not optimized for today's newer technologies and open standards," Nesi says. The middleware can be upgraded with a dashboard software tool or replaced with a production workflow system, deriving insight that can be used; for example, eliminating an external cause of internal production defects by going back to the supplier.

Big data analytics application does not necessarily stop once the end user takes control of the product. A jet engine or medical device in the field, for instance, can remain connected to the IoT. This enables product manufacturers “to analyze the data produced to help the customer optimize the use of that equipment," Cisco's Namboodri says. The data also can yield insights influencing future product design.

Interconnectivity Challenges

Several challenges persist in making connections between the factory floor and larger decision-making processes. Many factories still use legacy systems that may or may not be reconcilable from a data standard—or even a network standard—point of view.

To leave as many existing assets in place as practical, manufacturers may be able to perform some intermedialevel of ry network integration. In other instances, however, “some of those assets need to be upgraded to a more contemporary control and network because they are almost obsolete anyway," says Nesi.

Another obstacle is that not all systems speak the same language. “Automation companies have grown up over the last 50 years with their own proprietary purpose-built networks and control systems, along with their own terminology," Cisco's Namboodri says. Similarly, Germany is developing what it calls Industry 4.0 even as firms in the the United States are taking a similar approach but calling it advanced manufacturing.

“There are a lot of different sets of dialogue," Namboodri notes, “and that is a challenge in being able to compare apples to apples."

More frequently, manufacturers seek data from external sources. “If you are only building models around your own data, you end up optimizing around your view of the world," IHS' Larson says. “Your view of the world might not be the true view of what is going on out there."

As such, organizations are forming partnerships to exchange data. Potential problems arise, however, when companies have proprietary standards or policies that do not allow data to reside in a cloud environment and thus be accessed externally.

Analytics in Action

For all the talk about advanced analytics offering insights across the enterprise for better decision-making, the strategy must deliver results. When Cisco outsourced its manufacturing processes a decade ago, the company developed and supplied its own data-gathering tool on a device that was attached to the quality systems within the contract factories. This allowed Cisco to have direct real-time visibility into a system and process beyond its control so the company could react and change if needed, says Namboodri.

“This is one of the first silos of information we were able to integrate into other systems for better visibility and integrity."

Now, cloud computing has replaced many physical appliances on systems, allowing Cisco to gather data not just from the contract manufacturers but the component suppliers as well. The resulting big data and analytics “have become a mechanism for us to reduce our cycles and risks because we have a good understanding of schedules, quality and all the parameters associated with uptime," Namboodri says. “We've been able to improve our productivity and profitability by having a better grip on our supply chain all the way from component suppliers to contract manufacturers."

Rockwell Automation recognized the need to connect its own enterprise before setting out to help its customers do the same. The company comprises worldwide manufacturing facilities and supply chains responsible for 387,000 stock keeping units and custom orders. An average product order includes 200 different part numbers.

Rockwell took a five-step approach in developing its strategy: assessment, upgrading network and controls, defining and organizing working data capital, analytics and collaboration. Thanks to data analytics, Rockwell reduced lead times by half and lowered inventory from 120 days to 82 days, among other milestones.

Advanced analytics promises to unite the entire enterprise by unearthing previously concealed relationships among data, leading not only to valuable insights but action. To make such a strategy successful, manufacturers will have to address the challenges of establishing the appropriate infrastructure and finding the right expertise and partnerships.