Big Data Analytics for the Manufacturing Industry: How to Find the Hidden Gems
A wealth of information contains secrets about process behavior over time
When you begin using big data analytics for the manufacturing industry to gain insights about production, you might find that you have a few “Rembrandts” collecting dust in your industrial attic.
There is unrealized value just sitting inside the data stored in your historian. You may be sitting on priceless works of manufacturing “art” and not realize it. Big data in these historians contain details about process behavior over time and clues about ways to optimize operations. Often, however, these insights remain locked away like precious gems in a lead-lined vault.
The problem is not unique to manufacturing companies. An article in the Harvard Business Review, Unlock the Hidden Value of Your Data, references a book called Rembrandts in the Attic. Written more than 20 years ago by Kevin Rivette and David Kline, the book compares unused patents to priceless artwork—say, by Dutch “Golden Ages” painter Rembrandt Harmenszoon van Rijn—just sitting neglected in the attic. The authors said these patents are intangible assets that are extremely valuable but often overlooked. Instead of being “lost in the attic,” these patents should be treated like currency and valued.
Stefaan G. Verhulst, Co-Founder and Chief of Research and Development at The GovLab, based at the NYU Tandon School of Engineering and author of the article, agrees. Rather than patents, he sees big data as modern-day Rembrandts in the attic.
“It is widely accepted now that the vast amounts of data that companies generate represents a tremendous repository of potential value. This value is monetary, and also social; it contains terrific potential to impact the public good. But do organizations—and do we as a society—know how to unlock this value? Do we know how to find the insights hidden in our digital attics and use them to improve society and peoples’ lives?”
– Stefaan G. Verhulst, Author, Unlock the Hidden Value of Your Data
Like the priceless art hidden away from view, production data locked inside a historian is process behavior gold waiting to be found. Without a solution to leverage this resource and make data available in a central location, however, it remains locked away in independent silos.
Advanced industrial analytics solutions help process experts find hidden treasures among their big data in storage. They empower process engineers to tap into a wealth of resources to help improve sustainability, optimize performance, and increase the company’s bottom line.
The manufacturing industry has been capturing time-series and contextual data for years. But the data’s full potential often is not being reached because data analytics traditionally has required the help of a data scientist. Today’s advanced analytics solutions are designed to work at any analytics maturity level so process experts evaluating all four types of analysis find value. From trend viewing to Python notebooks integration, process manufacturing experts can solve even the most challenging production problems without data science training. They also can apply machine learning and artificial intelligence capabilities to iterate patterns of good behavior or eliminate bad ones.
Realizing the Value of Data
Not all industrial data appears to be as valuable at first. If you’re looking for a Rembrandt in your process manufacturing attic, your scope may be too narrow. An Arizona man proved that with a discovery he made in his garage.
In 2015, the retired man found a lost painting by impressionist artist Jackson Pollock. It was estimated to be worth $15 million. While Rembrandt is a common and well-known artist, Pollock is not as popular. Realizing the value of his work requires special consideration and an understanding of the art market. Industrial manufacturing data is much like an underappreciated Pollock in the garage. Appreciating its value requires analysis.
Industrial manufacturing plants capture two main types of production data: time-series and contextual. Time-series data records process behavior (and influence factors on process behavior) from sensors throughout a factory over a given timeframe. Process experts can use time-series data in a historian to learn more about their processes. For example, they might use it to determine what caused an anomaly, so it does not continue to occur. Similarly, engineers might use the manufacturing data to record periods of good behavior and create a golden fingerprint—the ideal operating conditions.
However, with historians, process experts are limited to 30 days of data, and they generally only provide descriptive analysis. To get the most from time-series data, process experts must use an advanced analytics solution designed to analyze big data.
Contextual data includes production information about quality, maintenance logs, operational events, environmental conditions, and process performance. It typically resides in external business applications. Because contextual data is not created from the same source as time-series data, it often is stored different, too. The data can remain in separate applications that are maintained and controlled by different departments within an organization. When used in conjunction with time-series data, however, contextual data can help engineers study process behavior as it relates to other business decisions. For example, while a process is offline for maintenance, its sensors also are not collecting operational data. Process experts can eliminate the time when the processes were offline during their evaluations.
Because of the additional value it can provide, contextual data should be studied when analyzing time-series data.
Uncovering Hidden Process Treasures
Searching for value in process data is much like exploring the beach with a metal detector. You know that value exists in the form of small metal objects under the sand, but you can’t see it without help.
Beachgoers stroll the sand with metal detectors in search of coins and other precious metals. Much like the metal detector, advanced solutions to analyze big data in the manufacturing industry detect production treasures that process engineers do not know exist and cannot see otherwise. Self-service analytics solutions empower process experts to search through years of data for secrets to improving operational performance.
Advanced analytics solutions empower process experts to:
- Increase plant efficiency, flexibility, and agility,
- Provide greater visibility into factory operations to control processes and production more effectively,
- Support plant management more productively because they can identify process deviations and take corrective measures,
- Reveal the root causes to process anomalies and any hidden process restrictions,
- Determine the right time for maintenance and predict process failures,
- Alert key stakeholders about the potential for process issues, which gives them enough time to intervene before there is a problem,
- Streamline plant processes and team corroboration, and
- Offer the organization insights for continuous improvement.
Analytics Becomes Your Guide
The promise of using data analytics for the manufacturing industry is enormous. So is finding a Rembrandt in the attic. If your attic is like most, you’ll need a flashlight to see what you’re looking at.
Think of data analytics as the flashlight to illustrate production data value. Analytics is the interpretation of data patterns. Data scientists break these patterns into meaningful and understandable portions of information. This information allows process experts to make data-driven and thus, informed, decisions. Process experts using advanced analytics software can make the decisions themselves without the help of a data scientist.
From pattern recognition to machine learning capabilities, an advanced analytics platform makes sense of production data. It taps into a wealth of captured time-series and contextual data captured over many years. Engineers working at any analytics maturity level benefit from its intuitive user-centric interface and seamless overall experience. Process experts get a more thorough understanding of plant operations, which allows them to make better decisions about production.
Advanced analytics software also helps take data scientists out of the equation. Although data scientist are experts in analysis, they generally do not understand the manufacturing process they need to analyze. Engineers often spend time explaining a process to a data scientist so he or she can crunch numbers. As a result, process experts can make better, faster, and more effective decisions to maximize plant operations.
Time-series and contextual data can be used to evaluate processes continually. Engineers can use this big data to measure, observe, check, control, and improve the process. They also can use it to solve previously unsolvable problems and prevent them from recurring. The analytics platform provides engineers with constant and increased awareness about what’s happening throughout the factory.
Leveraging Data Value: A Business Use Case
By giving operational experts access to the data so they can connect the dots, advanced analytics solutions provide opportunities to save time and energy, reduce waste and downtime, gain a greater competitive advantage, and increase profitability. A German specialty chemical company has reached this level of operational performance thanks to a self-service advanced analytics platform.
LANXESS has embraced new technology in its mission to become data driven. Chief Digital Officer Jörg Hellwig said economic advantages were a driving decision to digitalize its manufacturing plants.
“The use of data analytics tools in production is a clear business case,” said Hellwig after rolling out the self-service analytics software. “LANXESS is now leveraging the optimization potential throughout the (company).”
Hellwig said he understands that using the right solution is the most effective way to uncover hidden value in industrial process data. The company has been collecting time-series and contextual data for years but did not have a way to unlock its value before investing in the self-service solution.
“Data by itself is worthless,” he said. “It doesn’t generate any ideas on its own. We now have to leverage this treasure trove of data. We can’t exhaust its full potential until we are able to work with all this data and analyze and understand it.”
Advanced analytics for the manufacturing industry will help you find the Rembrandt or Pollock hidden in your industrial attic.