A Closer Look at the Digital Twin With Russell Gentry

The director of the Digital Building Lab at Georgia Tech, Russell Gentry, gets a unique view of the digital twin. Positioned as the connective, digital tissue between the fields of architecture, computing, building construction, and civil engineering, the DBL is helping push digital twin technology and application past its initial stages.

"It's not just the geometry of a building. Architects and engineers have been doing geometry for years. We've already seen what CAD looks like, and images of BIM," he said. The digital twin takes all that modeling one gigantic step further, he said. 

"Digital twinning is about the building and all the components that are in the building. Where they are, what condition they're in, all kinds of qualities," Gentry said.

How does DBL use digital twin in its research?

Gentry: The Digital Building Lab, even from day one, under Chuck Eastman's leadership, was really about imbuing the physical models that artists, architects, and engineers create with data. So that was baby steps at first, tagging attributes on drawings so that we could count them, schedule them, specify them. And so those those were early efforts to make the drawing or the model something smarter, smarter than just the geometry of the building, and the quantity of things.

Then at some point, everything we know about an object in a building or the building itself could be affiliated with the building model. That's the Holy Grail, if you will. Any questions that you had about the condition of something, the age of something, even active asset tracking — where something is in the building, that becomes a model that is much more powerful. That’s the digital twin: the model of the building contains enough information that it can answer your questions.

The DBL’s work on wind turbine blades for the NSF is a good example. We know the nominal geometry of the blades from the factory, but over time, that geometry changes — things wear out. A digital twin of a wind turbine blade could capture the original geometry as it was designed and how it changes over time. With a digital twin, you could run a condition assessment and get a notification once the geometry of a blade is so far out that it needs to be serviced. In buildings, the more common example might be with air filters. You could have a building where air filters are on an exchange cycle of 30 days, 90 days, or once a year. Or you can have a digital twin of the building where the the H-Vac system is set up to self diagnose. The individual filters can report to the digital twin and say, ‘there's a larger pressure drop across this filter, because it's dirty.’ And thanks to that you replace one filter instead of going in and replacing lots of clean filters that aren't dirty.

How does digital twin show up in Georgia Tech’s architecture

Gentry: The graduate course I teach, in building systems and data, is really where we talk about the birth of building information modeling (BIM) and the future of BIM. The students learn how to create databases, to affiliate data with objects and query those objects, in addition to drawing or model buildings. We recently had a student working on a project to do fire code evaluation for Starbucks. They wanted to automatically determine how many people could legally be in a Starbucks at any one time and still meet the fire code. So, there has to be sufficient data: how many tables, how many chairs, where the exits are, how far are people from the exits? If you imbue that data into the building model, that creates a digital twin of the Starbucks, in this case, and you can determine whether a certain configuration is likely to pass inspection.

My colleague Daniel Willkins, who's leading a project in the Digital Building Lab for the Smithsonian Institution, is looking at all of the data that's imbued onto models of historic buildings. Buildings of historic significance very likely have components that are irreplaceable. And the building itself might become an exhibit. If you're going to The Mall in Washington, D.C., to see the Arts and Industries Building, you're looking at the exhibits in that building, but the building itself is an exhibit. That building’s digital twin becomes a way to know what specific cornices were made of. What is that material on the floor? Where was that limestone? Those are stories, if you will, or narratives that are embedded onto objects that can talk about its past and its preservation.

Very much of the digital twin is about performance. There are active and passive elements in buildings that affect the building’s performance. An active element is like the example I gave previously about the H-Vac system. Passive element could be an element of the facade, a window, an opaque wall, a roof. Those elements change over time. You could have water ingress into a facade, which would totally change the performance of the thermal envelope, the R value. My colleague Tarek Rakha’s work with drones and the Department of Energy is really determining ways to embody updated information directly onto the facades of buildings. Tarek and his students use drones to map the heat of building envelopes in real time. The infrared technology that he's developed could make digital twins more complete.

Do other College of Design professors research digital twin?

Gentry: Pardis Pishdad in the School of Building Construction is doing such interesting on integrating the Internet of Things (IoT) and digital twin. And as far as I know, there are not a lot of people who are working to fuse real time data from buildings to building models. Traditionally, building models are static. A building model is handed over to a sophisticated client — like Georgia Tech — and that building model is accurate the day it's delivered. But it immediately goes out of date. What if we refit someone's office, or we cut a door into a wall? The promise of the digital twin is a model that's updated and kept alive and imbued with all this data, so that it's always accurate and it's always useful.

One of the questions we struggle with in the AEC industry is which profession will maintain the digital twin, build it, and keep it up to date? And can we demonstrate the value proposition to building owners? In other words, if I say I need a $5,000 contract a year to keep the digital twin of the class building up today, right all the data is cleaned and relevant any changes to the building or model. Any changes to the data infrastructure are added to the model and the model is always up to date. So that becomes a data store for the people maintaining the building. That doesn't come for free. A lot of our research is focused on automating those processes to make digital twinning possible.

And I'm keeping my eye on the new Master of Science in Urban Analytics in the School of City and Regional Planning. The research Subhro Guhathakurta, director of the Center for Spatial Planning Analytics and Visualization and this new program, does could lend a lot to scaling digital twin to the city and even mega-region level.

Is the AEC industry keeping pace with digital twin?

Our students are capable of using digital and that's what I hear, continuously, from DBL clients. In fact, I was talking to our friends from Aptitude just yesterday. They told me one of the most important reasons to be a member of the DBL is to have access to our students and hear what they're thinking. In the so-called “war for talent”, it's good to be on the cutting edge of things.

But I think that the future of digital twinning and buildings is not defined yet. It is still in the developmental stage. It's going to be most effective when we're talking about high-value, active systems and buildings that require a lot of maintenance. Lighting, H-Vac systems, building access control, you know, what are the things that wear out quickly and and cost the most to refresh? And can we track their condition or their need for maintenance? Our students are capable of building models with a lot more intelligence than is the standard in the AEC industry today. But what we have to continue to do is to demonstrate the use cases for these models in the industry.

Questions?

 
If you can't find the information you were looking for, we'll get you to the right place.
Contact Us