GE Plans Software Platform For Creating 'Digital Twins'

GE already uses an internal platform for modeling jet engines, turbines, and other physical assets. Now, the company will be offering it as a service by the end of 2017.

Charles Babcock, Editor at Large, Cloud

July 20, 2016

6 Min Read
<p align="left">(Image: pluem05/iStockphoto)</p>

11 Programming Languages For DevOps Success

11 Programming Languages For DevOps Success

11 Programming Languages For DevOps Success (Click image for larger view and slideshow.)

GE is working on a modeling platform that will allow it to create a "digital twin" of all its manufacturing products by the end of 2017. Whether it's a jet engine, a windmill turbine, or a locomotive, the physical asset will be duplicated digitally through a series of models, which can be operated in a virtual environment.

For a manufacturer like GE, creating a digital twin to a physical asset has many advantages.

It would allow succeeding generations of a product to be created on the drawing board, simulated using data drawn from the sensors of an existing model in real life, and revised into a more fine-tuned product -- before going to manufacturing -- explained Rusty Irving, services technology leader at GE Global Research.

With existing jet engines loaded with 20 information-gathering sensors that take readings every second, a large amount of operational data is available for collection and use in simulations with the digital twin. By studying the simulations, GE engineers can track how well the engine is performing and what it's maintenance schedule should be.

Leaving on a Jet Plane

Jet engines have always had maintenance schedules, but they've been compiled without a lot of feedback information on their current operational state.

By plugging operational data into the twin's virtual reality operation, GE can come up with a maintenance schedule "that is condition-based instead of time-based. It starts to advise us how we can get more operational value" out of the physical asset, Irving said in a recent InformationWeek interview.

As one example, the expensive and crucial high-temperature alloy turbine blades in a jet engine, which are performing much of the work of creating thrust, can be replaced as they produce data indicating it would be wise to do so. Employees do not, then, have to swap them on a routine schedule. The approach ensures maintenance gets done when it's needed, but puts that schedule on a more realistic, feedback-oriented footing, he said.

GE has been working on a platform equipped with tools that can take various digital representations of a turbine or other physical asset and convert them into a set of models that make up the digital twin. The process starts out with the CAD drawings and blueprints through which the product was created, but builds toward a more complete picture that includes the conditions under which it operates.

Among other things, aircraft manufacturers want to know what happens to jet engines as they fly into and out of dry and frequently dusty regions. There's now a dust belt that extends from North Africa, across the Middle East, and into China.

It's well known that the airborne debris from a volcano's eruption can play havoc with a jet engine. What about the debris from desert or near-desert environments, where the dust might seem to have a gentler, more innocuous impact? What's the long-term effect of frequent exposure to particulates that are 10 microns or less?

The digital twin can be fed temperature, wear, and vibration data to form a preliminary assessment, which can be used to allow testing under mock conditions that might step up the exposures. It isn't one model that would be used to create a digital twin, but perhaps 20 that would allow the testing of different parts under different conditions.

Likewise, GE would like to create digital twins of the steam turbines that are used in power-generating plants.

Focus on Power Plants

A previous generation of plants were operated continuously "for decades," said Irving, with data collected on how to perform maintenance with such a schedule. Today, however, some older plants are being taken off line and replaced with some form of green power, then brought back on-line when the source of that power, such as solar or wind, has been temporarily exhausted.

The managers of the power grid then call for more plants to be brought on-line, and that's when GE's power generating customers want to accommodate that demand. But what is the effect of this cycling of steam turbines from hot to cold, and back to hot?

The best practices of the past dictated that a gas steam turbine be brought up to its high operating temperature before being called upon to generate electricity.

"We've found that we can run it a little colder and still produce electricity," said Irving. That finding significantly alters how quickly the plant can be brought on-line, and how much fuel is consumed before it become productive.

By using the turbine's digital twin, GE can explore the trade-offs of that earlier cut-in to the grid. It knows the turbine runs less efficiently when it's cold. What is the best point to make the trade-off between temperature, time, and fuel? Turbine blades are made of an expensive nickel alloy.

At what point is their longevity affected by this change in operational procedure?

With 200,000 assets like the power generating turbine to monitor daily, the digital twin becomes a key component in how GE maintains the asset and extends its life span through knowledgeable operation, Irving said.

A New Service

GE has developed a platform for building digital twins for internal use, and he said the company "will make a business out of this" in which it sells a platform equipped with the same tools for other companies to produce their own digital twins. He expects such a platform to become available as a service by the end of 2017, he said.

Use of digital twins will become increasingly important if the world's climate continues to change, as in the dust belt condition previously described, he added.

"Climate change has a dramatic effect on many of the products we produce," he noted. With the digital twin, GE engineers can plot the change and forecast conditions soon to come, plugging that information into how long it can project that the service life of its products will last.

[Want to see where some machine-learning analysis might come from with a GE digital twin platform? Read Microsoft, GE Partnership Targets Industrial Cloud.]

GE is building machine learning or "self-learning" analysis into its platform to constantly make use of the data that's collected from equipment sensors, and also what is known about it operating environment. "We need the physical domain knowledge. Customers trust and rely on us to have that," he said, and to know how to make use of it.

GE is not the only party working on the problems.

The concept of a digital twin was actually originated by Michael Graves, a researcher at the University of Michigan in 2001. The federal government has established and funds a Digital Manufacturing and Design Innovation Institute in Chicago to allow companies to conduct research into the concept. Dow, Lockheed Martin, Rolls-Royce, Siemens Product Lifecycle Management Software, and GE are participants in the institute.

In the future, the success or failure of a product may hinge on how skillfully its digital twin has been constructed, how successfully operational data can be fed into it, and how well the resulting data can be used to forecast its future, Irving said.

About the Author(s)

Charles Babcock

Editor at Large, Cloud

Charles Babcock is an editor-at-large for InformationWeek and author of Management Strategies for the Cloud Revolution, a McGraw-Hill book. He is the former editor-in-chief of Digital News, former software editor of Computerworld and former technology editor of Interactive Week. He is a graduate of Syracuse University where he obtained a bachelor's degree in journalism. He joined the publication in 2003.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like

More Insights