At Uptake, we believe that success requires a depth of understanding. This holds especially true in heavy industrial sectors that not only rely on machines and equipment, but that also run on interdependent processes. Where a sudden, unexpected problem upstream can result in even bigger consequences downstream.

Out here, the stakes are high. The answers are hard — not obvious. And problems don’t fix themselves.

That’s what we’re all about. We see technology as playing a pivotal role in defining the next generation of products and services our world and economy will depend on — new ones and all the others we’ve come to expect.

Care to meet some of the people behind that technology? Be sure to check out our past episodes to learn how Uptake’s industrial data scientists are doing big things like:

Keeping wind turbines turning to produce more renewable energy.

Helping locomotives run better for more reliable rail transport.

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

Now Here’s the Catch

Uptake industrial data scientist Agnes Bao puts her skills to work for a major metal fabrication manufacturer. The goal? To help improve the overall equipment effectiveness (OEE) of its critical production assets.

Around the world, today’s manufacturers are struggling to win the fight against unplanned downtime. For most of them, it’s the single largest cause of lost production opportunity (LPO) events. In fact, industry analysts estimate that the cost of unplanned downtime for manufacturers can skyrocket to as much as $540,000 for every hour that critical production assets or constraint machines are out of service.*

For them, the objective is to conduct planned maintenance on their own terms. Why? Because by catching smaller issues today — before they turn into bigger issues down the line — manufacturers can take immediate, proactive steps to increase the availability and effectiveness of their production equipment and improve their overall production control.

Uptake’s industrial data science helped increase the availability of our customer’s equipment while minimizing the unplanned downtime incurred. By turning machine data into predictive insights, we improved the efficacy of the key asset in its manufacturing process — the stamping press.

We took a holistic view of the stamping line process, which includes placing a flat sheet of metal onto the stamping press, where a tool-and-die surface then punches out the metal into the correct shape of the product.

A key variable that indicates the efficacy of the stamping press is its tonnage — the force that the press is designed to exert against the sheet of metal that’s in the die.

Using sensor data from each individual tool-and-die punching stroke, our model measured the difference between the actual versus the rated peak tonnage. The greater the difference between the two values, the duller the stamping press and the higher the probability of product defects and resulting scrap.

That is quite difficult to determine because:

• There are numerous confounding factors that can contribute to tonnage signal anomalies.

• Technology must be capable of filtering through the noise of those potential contributing factors to pinpoint the real problem.

• It’s difficult to decipher what’s really causing the jump in signals versus what’s acceptable variance in material or die data.

• At the end of the day, machine data on its own is not enough. It must be combined with operator data and maintenance records to understand the full story of what’s actually happening with the stamping press on the front line.

Uptake’s Failure Prediction Engine helped solve the above challenges for the manufacturer. As one of our canonical AI engines, it predicts the probability of future failures by using historical asset data plus real-time sensor data on the current conditions of the machine. Using the engine, Agnes then built, trained and deployed a model that fit our customer’s stamping press — from data input to predictive insight output.

With machine data streaming from our customer into the Uptake Platform every two seconds, our AI technology turned that mountain of data into actionable intelligence. Our predictive insights provided advanced warning of failures, informed smarter preventive maintenance tasks and demonstrated the ability to increase uptime by up to 15 percent.

As a result, Uptake unlocked hundreds of thousands of dollars in cost savings for one of our customer’s stamping presses. Scaled across the company’s stamping presses and plants, that translates into millions of dollars in potential cost savings.

Ask Me About My Failures

Watch Agnes talk about the catch in this video:

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

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 how that information relates to each other — and, importantly, how all of it 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 a tool that automates 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 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.

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.

Check out the fifth installment of Model Behavior here.

*Gartner Research