Breaking Down Dark Data Barriers in Asset-Intensive Industries 

The phrase “Dark Data” – referring to data that is unknown or untapped – has been bantered around technology conferences for years. Many industries are well on their way to becoming data-driven, but asset-intensive industries, like utilities, oil & gas, mining, rail, and others that earn revenue off self-constructed, long lived, complex assets, are still challenged to figure out how to take advantage of what can be a hundred years of data. It’s not that these companies don’t understand the value of data. They absolutely do. But the task of finding, organizing, analyzing and gaining insight from data seems like a herculean and expensive task, while strategic initiatives like improving reliability and reducing carbon are top priority.

The reality is that harnessing data doesn’t have to be overwhelming or take attention away from other priorities. In fact, being a more data-driven organization can better support top priorities. By committing to a data-driven future defined by specific, articulated business goals, artificial intelligence (AI) and machine learning can help organizations break through the data chaos to more quickly achieve strategic goals.

The Value of Dark Data

Many asset-intensive businesses know there is value in tapping into their dark data to help them optimize the use of their critical infrastructure. Fixing a critical asset when it fails requires new parts and immediate crew deployment. It is costly. With 80% of asset work being reactive, there is money in being able to predict when an asset will fail and perform maintenance before it does. The goal is to flip the ratio, maximizing proactive maintenance, reducing costs, and improving reliability. AI, machine learning and the ability to take advantage of dark data makes this all possible now.

What’s the Holdup?

The way asset-intensive businesses have been built, and their industries evolved, over many decades, creates some hurdles. Data spans decades, locations, sub-companies and varying levels of sophistication. For context, the oldest utility was founded in 1816 and first steam-powered railway systems were started in the 1820s.

Information about how expensive and vital equipment was specced and when some components have been replaced or maintained existed before the days of formal data collection and warehousing. There is no single place where all this information lives. Often, the knowledge of specific legacy data retired with someone years ago. That information can be a pile of papers in a bottom drawer, an excel sheet from 2001 or in the cloud.

This can be overwhelming if the assumption is that the value of data lies in having all the data. But do you really need all the data? Today’s technology can help. What you need are tools to find enough of the right data, to help fill the holes and run simulations to predict the outcomes that 100% of the data would produce.

Becoming Data-Driven

A technician uses ABB Digital Enterprise solution

So, what can asset-intensive organizations do right now to tap into their dark data? Contrary to popular belief, addressing the problem doesn't involve going out and buying the fanciest new tool out there. What you need is a clear objective. The technology is useful only when the business goal is established. Then AI and machine learning can assist, not just because it’s powerful technology, but because it's being pointed at accomplishing a clear business objective. The shiniest new tool that everyone is scrambling to get won’t help you much if you don't know how to measure success. Data on its own won’t get any company where it wants (or needs) to go.

Leadership also needs to buy into a data-driven future and develop a programmatic approach for the organization. A couple of engineers with limited decision making power and little visibility into company-wide objectives can’t bear this lift on their own.

Top leadership must clearly define the goals and anticipated business benefit of leveraging dark data, establish priorities for the effort, and communicate that prioritization across the business. For example, what problem does the business want to address first? Is the goal to defer capital investments, reduce operational spending, make more fact-based decisions, model additional scenarios? Based on the priority, a specific, actionable plan for leveraging dark data to address this problem can be established. Boiling the ocean by trying to wade through everything without a specific objective won’t deliver value in the end.

Dark data gives asset-intensive industries more operational agility than they have ever had and that translates into savings and improved resiliency, among other benefits.

Harnessing 100 Year Old Data to Predict Equipment Malfunctions

Take the example of a hundred-year-old power generation and distribution utility. It produced around 40% of its power through hydroelectric plants. However, an increasing number of its plants were reaching the end of their designed operational life, with some facilities being over 100 years old and others refurbished several decades ago. In order to dig out their dark data and tap into its benefits, the company turned to an AI-powered Asset Performance Management (APM) system. The team reviewed asset condition data histories and correlated those conditions with critical asset malfunction modes to determine future malfunction risk profiles. Analysis then revealed that a generator’s increasing level of vibration was driving the risk of future malfunction and was highly sensitive to load.

The team was able to take the useful data and simulate what would happen to vital equipment at different levels and intensity of use. In this case, the team was able to calculate when the malfunction would occur if the generator continued operation at the previous load and recommend the optimal load for achieving stability of the vibration gradient. With that information, the utility now more efficiently and cost-effectively manages its assets. It has become more proactive, intervening to prevent future malfunctions before they occur, saving on costs and achieving their original business objective.

Many asset-intensive industries have been slower to leverage dark data because of the scale and complicated nature of how these businesses have been developed over past decades. AI and machine learning provide opportunities to accelerate their development and reap the benefits. And it’s not about having all the data or even tackling everything at once. Asset-intensive industries can start with small applications of AI and machine learning to predict, simulate and then expand. To see this success, dark data initiatives need to be guided by tangible business objectives from the top down – and that’s non-negotiable.

About the Author