This content is part of the Essential Guide: IoT analytics guide: Understanding Internet of Things data
Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Maritime company cuts through the noise of sensor data analytics

Sensor data presents a major big data problem due to its volume and velocity. One user's experience shows that managing its flow is key to not drowning in all the data.

Sensor data may be among the most valuable types of big data, providing businesses with insights into the operational health of pricey machinery and potentially enabling them to head off breakdowns that result in costly repairs and unplanned downtime. But the primary challenge on any sensor data analytics project inevitably comes down to how to make sense out of the mountains of data that come from sensors.

Big data problems get really big when sensor data is involved. The data typically streams in from industrial equipment and other machinery that often generates thousands of data points every second. Finding the signal in all that noise can be tricky. Despite the potential benefits of analyzing sensor data, that hurdle has kept some businesses out of the game.

"The biggest challenge has always been, you need to know what you're looking for," said Ken Krooner, president of Engineering Software Reliability Group (ESRG), a company in Virginia Beach, Va. that offers a set of software and services to help the U.S. Navy and other maritime organizations monitor the condition of machinery on ships and offshore drilling platforms.

Many companies have trouble getting started with sensor data analytics because they lack one or more of the necessary components, Krooner said. To be successful, he added, a business needs to have a clear problem that can be solved by analyzing sensor data, workers with both analytics skills and domain knowledge (in ESRG's case, knowledge of maritime machinery), and the right tools to handle all the data.

"I believe what has slowed down adoption is those three legs of the stool have been difficult to put under one roof," Krooner said.

For ESRG, the first piece of the puzzle is clear: Its clients' business models rely on shipping vessels and oil and gas drilling equipment working at full capacity. ESRG solved the second issue by hiring people who have worked in various maritime industries and know all the machinery involved. That makes it easier for them to spot patterns in the data that are likely to be meaningful. To address the tools piece, ESRG chose a data integration and analysis platform from software vendor Pentaho, and offers its own business intelligence (BI) tool. The software ingests streaming data from machine sensors installed on customer equipment and organizes it in a relational database for analysis.

But ESRG doesn't save every piece of data that comes off its sensors. Krooner said trying to do so would overwhelm ESRG's system and make performing meaningful analytics impossible. The solution was to set conditions under which the sensors report data back to the company, such as when a piece of machinery starts up, runs for a long period of time or goes through other potentially straining operations. That approach catches meaningful data while still keeping data volumes under control, according to Krooner. "What it helps us address is data overload," he said.

Examples of what ESRG has done for customers with the tools and the team it has assembled include helping to spot a failing generator on a Navy destroyer just before it went on a deployment, identifying a problem in a shipping vessel's engine that was leading to fluctuations in fuel usage, and finding a faulty fuel injector aboard a supply ship. Krooner said fixing the faulty fuel injector reduced the ship's fuel consumption by 5%.

While adoption of sensor data analytics may be slower at this point than some commentators had predicted, Krooner said he thinks more people are starting to see the potential benefits, and may be ready to invest in systems. "There's less and less resistance toward putting resources like that out there," he said.

Ed Burns is site editor of SearchBusinessAnalytics. Email him at and follow him on Twitter: @EdBurnsTT.

Next Steps

How to turn sensor data into your own personal mechanic

Sensors help trucking company keep traffic flowing

Big data project uses sensor data, other info to predict weather for farmers

Dig Deeper on Internet of Things (IoT) Analytics

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Kudos for a fantastic article.

With all the enthusiasm for sensors, M2M and the Internet of Things, there is a very high likelihood of "drowning in data". Your article is an excellent preventative to this problem.

(FYI, I've written about this issue elsewhere. You could say there's an economic principal involved: "data is cheap but business analysis is expensive". If it's OK to reference, here's an item on "alarm fatigue" and what manufacturers can learn from healthcare, on the same topic:
Insightful post, John - it's important to remember (but easy to forget) that all these data sources are just indicators, and you need to understand what is behind these indicators in order to focus your attention, not just react to the immediate stimulus of the beep, the ping or the alert. Makes me think of the hospital sketch from Monty Python's "The Meaning of Life."