Model Behavior: Eliminating Waste in Heavy Industry

Welcome to the seventh installment of Model Behavior. This ongoing series provides an inside look at Uptake’s industrial data scientists and the challenging work they do on behalf of our customers. Why is this topic so important? Because tough problems don’t solve themselves. See how model Uptakers build and deploy Uptake’s models — and how both of those elements must work together in harmony to create a world that always works.

As heavy industries become more connected, data affords them the opportunity to run more efficiently than ever before.

Yet, around the world, there’s still a lot of waste. Can data be used to change that? Not only are we certain of it at Uptake, we play a role in making it happen every day.

For proof of doing more with less, look no further than the manufacturing sector.

Manufacturers are no stranger to innovating for efficiency’s sake, and to focusing on reducing non-value-add costs or indirect costs. But today’s leaders are separating from the pack by harnessing machine data and process data to improve their productivity and eliminate waste.

Let’s take a closer look at how.

Now Here’s the Catch

Uptake industrial data scientist Tony Bouril is an expert at identifying and analyzing patterns. As a member of our Data Assessments Team, Tony spends his time turning data into meaningful intelligence. He puts our customers in a better position to ensure their critical assets are performing as best as they can.

Today, Uptake works with a major manufacturer to prevent oil leaks in its critical production assets. The customer originally turned to Uptake because it had been suffering from an unidentified leak in one of its machines, which was prone to spilling oil onto the surrounding environment of the plant.

The leak posed obvious safety and cost concerns that needed to be addressed. What’s more, the uncontrolled oil consumption was an indication that there could be something wrong with the machine’s health and performance. The manufacturer sought Uptake’s help in solving the problem and making the oil waste disappear.

Using oil-level data from the manufacturer’s hydraulic presses, our model processed and filtered the data to separate signal from noise. It then calculated the slope of the signal — in other words, the direction and steepness of the fluid level over time. Whenever the slope would start to decline, our model would alert our customer to the behavior.

Like any real-world problem, our model had to overcome challenging obstacles. It had to contend with machine vibrations, erratic behavior and performance, and missing data. Why was it essential to solve for each of those factors? Because they all can influence the resulting insights produced by the model.

At the end of the day, it’s pointless to have insights without action. The real keys to solving our customer’s problem were the subject matter experts who understand how the machinery is intended to function on the front lines. They used their firsthand knowledge to validate the insights produced by our model, identifying expected versus unexpected behavior.

Additionally, in this case, industrial data science enabled plant personnel to spend more time on value-add activities — like building product. Now that our model was monitoring the oil levels of the machine, they’d alert plant personnel with a specific, recommended action when needed.

To illustrate how vital it is to couple model performance with operating context, consider the following scenario:

  • A critical production asset — which shapes pieces of metal using hundreds of tons of force — shakes, vibrates, and bounces up and down in its normal course of action.
  • When the machine is shut down, oil drains down from its pipes back into an internal reservoir. This causes a large yet expected increase in the oil-level signal.
  • However, after the shutdown, the oil-level signal didn’t return to its previous level. In fact, it was much lower than normal, indicating that oil had left the machine entirely.

How do you approach solving the problem described above? In short, it takes a blend of insights from models, usage by subject matter experts and an understanding of the manufacturing process to determine expected versus unexpected results.

Uptake revealed that this issue had been causing roughly 120 gallons of oil to be spilled onto the plant floor each week — unlocking approximately $70,000 in potential annual cost savings on cleanup and materials at just the one site location.

For any manufacturer, plant personnel are largely focused on keeping overall production going. They’re usually too busy to manually check all visual indicators and available gauges. That’s precisely why they need actionable, data-driven insights that alert them to problems and recommend fixes. It enables them to quickly take care of any issues and get back to building product.

Ask Me About My Failures

Check out this video to hear Tony explain the catch in his own words:

Want to meet more of Uptake’s industrial data scientists? Check out our past episodes to see how they’re using their skills to:

Drive greater fuel economy from industrial engines.

Power a brighter future for society.

Keep manufacturing lines up and running.

Make sure trucking fleets stay on the road and deliver on time.

Help locomotives run better for more reliable rail transport.

Keep wind turbines turning to produce more renewable energy.

New to Our Model Behavior Series?

Here are some helpful definitions to know:

A machine learning model is a codified structure and digital representation of an industrial asset that physically exists in the real world. It organizes the data generated by the asset and standardizes relationships within that data — and, importantly, how all of that data relates to the physical properties of the actual asset. A model targets a specific component of a specific asset, growing increasingly accurate over time with each new data point.

At Uptake, our machine learning models are the output of our AI engines, which are tools that automate the development and deployment of our models. Our engines use valuable insights gained from different industrial use cases to come up with approaches that solve common sets of problems.

A failure happens when an asset or one of its components shuts down unexpectedly in its operating environment, causing problems like unplanned downtime, lost production, service disruptions, unsafe working conditions and more. Failures are what models try to predict and prevent from happening.

A catch is when one of our models — running in a live production environment for our customers — generates an alert that correctly warns of a failure or performance issue that’s bound to happen. Or, in other words: A catch means our model did what it set out to do.

Think you have what it takes to become an industrial data scientist at Uptake? Visit our Careers Page to view our open positions.

Stay tuned for our next installment of Model Behavior.