Heiko Claussen is SVP of AI at AspenTech, responsible for the company’s industry 4.0 strategy, industrial AI research and data science.
Getty
The volume of new data worldwide is projected to more than double by 2026. There are few industries in which the impact of big data is more evident than in the industrial and manufacturing sector. In fact, manufacturers expect data volumes to increase more than any other industry.
The sheer quantity of data presents a major challenge: Industrial data only continues to grow in size and significance—and if organizations can’t keep up, operations and decision making will spiral out of control. Simply put, the era of big data is over, and smart data is taking its place.
What’s the problem with big data?
Whether it’s sensors that can detect recurrent failure patterns, analytic insights that improve sustainability or models that solve bottlenecks and optimize process design across distributed factories, today’s manufacturers have more data analytics tools and digital technologies at their disposal than ever.
For example, manufacturing processes in high-temperature, chemical or other sensitive environments—such as a gas turbine, food production or metal smelting—can employ virtual sensors that infer data from other sources. While traditional sensors would malfunction or compromise performance in those environments, virtual sensors can use contextual information from other parts of the process to provide data that would otherwise be unavailable or unreliable.
With such abundant use cases, why aren’t manufacturers achieving more benefits from industrial data? The answer is that many businesses are stuck in a big data mindset, struggling to manage massive amounts of data rather than extracting actionable, valuable insights from it.
In particular, the growing array of data sources, digital platforms and analytics tools each has unique access controls, user permissions and data formats. When manufacturers are stuck with data silos and disconnected information, they fail to capture a complete picture of their business and miss out on opportunities to inform, improve and optimize manufacturing processes.
How can you make the most of your industrial data?
As the world enters the Fourth Industrial Revolution (a.k.a. Industry 4.0), manufacturers need clear, concise and contextualized data that provide meaningful information, actionable insights and newfound value.
Consider the following steps to build a smart data strategy capable of driving improvements, fueling better decision making and generating value throughout the entire manufacturing lifecycle.
• Getting started is half the battle. Nearly half of manufacturing companies are prioritizing connectivity and data visualization to enable operational transparency, while more than a third are focused on predictive analytics for better forecasting and planning. That should serve as inspiration if your organization hasn’t yet started on data analytics projects. Start with small data analytics projects that you can refine, improve upon and scale as your organization matures, whether it’s working to validate models, monitor assets and refine data quality or optimize process design. No matter where you’re at in your data journey, investments now pay dividends down the road.
• A strong data backbone is essential. Smart data requires having the right infrastructure and technology tools in place across your organization. Well-designed data management strategies foster an industrial environment and culture that values and fully leverages data organization-wide. One vital component is a dedicated repository that collects, stores and disseminates data from varied origins—what we call a “data historian.” Data historian solutions have become increasingly important given the high volume and complexity of data created throughout industrial facilities. A strong data backbone ensures your data projects not only get off the ground but continue to receive buy-in over time.
• Data quality makes or breaks insights. Your data insights are only as good as the data that goes into them. Consider a sensor with a floating ground: If the sensor is feeding inaccurate and inconsistent data points, teams may make decisions based on faulty data—and by the time they uncover errors, it’s too late to make a difference. As AI, machine learning and industrial IoT tools become more widespread, it’s crucial for manufacturers to continuously track and refine data quality practices. In particular, you can (and should) track data over time and compare historical scenarios, allowing teams to automatically monitor sensor quality and make adjustments along the way.
• Context matters more than ever. You can have all the data in the world, but it has little value without the right context to connect information and extract meaningful conclusions. Consider all the information necessary to tie together each data application in advance. Even a single sensor requires a high level of context: You need data about the particular sensor, the asset it's monitoring and its historical track record; the facility’s location and its conditions; and information about how the data will apply to a particular use case. Contextualizing data across several sources allows you to increase efficiency and ease of use for both current projects and future use cases.
• Use your imagination—and the right resources. The most potent hindrance to smart data is a lack of awareness about what’s possible. In manufacturing, IT domain experts possess a unique set of skills and expertise, but if knowledge isn’t shared throughout the business, it can hold back even the most technically advanced organizations. That’s where an industrial data scientist helps. This relatively new role ties together knowledge of toolchains, algorithms and other data-intensive processes with domain-specific expertise. Industrial data scientists are best positioned to recognize use cases, carry out end-to-end deployments and envision new functions and features. If it’s not feasible to build your own data team, collaborate with outside partners, universities or companies.
On its own, big data doesn’t always equate to big value. Manufacturers that amass more and more data without a clear plan or functional use cases risk losing out to competitors and falling behind the curve. A smarter approach to data empowers your organization to unlock the full potential and power of its industrial data.
It’s clear that we’re entering the era of smart data—and effective management and incorporation of industrial data will be crucial in determining who comes out on top. Are you ready to make the most of it?
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
