A Concrete Example of a Digital Twin
Phillip Keane posted on June 30, 2019 |

We have been talking about digital twins on engineering.com for a number of years. All told, digital twins do make for a lot of hype.

 If that hype is to be believed, the value of the global digital twin market is set to grow to somewhere between $15 billion USD and $40 billion USD over the next decade. So it’s a little weird that we don’t see any working examples of them more often. It’s almost like they don’t exist… We will examine exactly why this is later in the article.

But first, we are pleased to say that we have seen one working in real life, thanks to a demo from the folks at COMSOL. And interestingly, we saw it running on Android.

A Concrete Example

Finally after all this time we have finally seen a concrete example of a digital twin running in real life. And that’s not just a turn-of-phrase… we got our hands on a digital twin for monitoring the drying of cement.

If cement industry was a country, it would be the third largest producer of carbon dioxide (after China and America). It is a market that is ripe for innovation.

Cement takes a while to dry/cure. And knowing how long it takes to cure can determine how the next phase of a project begins, because you can’t build on concrete that hasn’t been cured properly. Getting that estimate wrong can cause delays of days or even weeks to the next phase or completion of a project.

This is where COMSOL comes in with their cement drying digital twin application.

Cement drying times are normally estimated based on several factors, including the dimensions of the cement section and environmental factors such as wind, temperature and relative humidity. In the past, the information used to determine these factors was based on historical average weather values.

However, weather can fluctuate wildly over time, and these deviations from historical values can have a significant impact on the prediction of the drying time for the cement. Weather is dynamic, and so a dynamic solution is needed to get the most up-to-date information.

With the COMSOL Concrete Drying application, we saw that the app had been connected to some sensors remotely, which were accessible over the web. These sensors can be positioned at a construction site and can measure environmental data, sending the data back to the COMSOL server once an hour, enabling a much more accurate picture of the situation on the ground.

In one example, COMSOL showed us how the predictions relying on historical data and those relying on real-time data were different by 20 days.

We spoke to Daniel Ericsson, Applications Product Manager at COMSOL about the relevance of mobile digital twin access.

“The important aspect here is that we can access instances of the digital twin from a mobile platform, Android, iOS, or whatever OS it may be,” said Ericsson.

“The digital twin and the digital twin aggregate (DTA) are specific for a specific product through its life, from design, manufacturing, operation, to disposal. This also implies that you may have to access the digital twin from wherever you may be.

“If you are a constructor and you need to know the status and properties of the concrete structure during curing, then you may actually be at the construction site, trying to decide if you should add water or not based on the predictions made by the digital twin. The COMSOL Client for Android simplifies the access to the digital twin using a smartphone or tablet for people, for example, at a construction site.”

How to Build a Digital Twin

Building a digital twin in COMSOL Multiphysics follows a fairly simple workflow.

The mathematical model is first built within COMSOL Multiphysics’ Model Builder. This is where you can add the CAD geometry and physics stuff such as data tables, formulae, boundary conditions and so on.

Next it is sent to the Application Builder. This is where you can specify the behavior of the application, such as information required to display, and how the completed simulation will appear. This is also where you add the instructions on how your digital twin should communicate with its sensors.

Advancing from the Application Builder, there are two options available. You can use COMSOL Compiler to develop a standalone executable file (we have written a tutorial how to use COMSOL Compiler in this article). Or alternatively, you can upload the application to COMSOL Server to be viewed in a web browser, or from the Android Client.

There are many paths to creating a digital twin.

Show Us the Twin Already!

We downloaded the COMSOL Client for Android from the Google Play Store (free download), and used the login details provided by COMSOL to access the COMSOL Server, where we browsed a few applications before finding the Concrete Drying App.

Starting the concrete app up, we were offered the option to run a plain old, non-connected simulation, or to run a simulation with the sensor data. Opting for the latter option, we were presented the option of selecting a construction site in Sweden, or one in the U.S.

Figure 2: Concrete Twin Welcome Screen on Android.

Next up is an input page, where you can specify the dimensions of the concrete under analysis, and other simulation parameters such as step time, simulation length (in days), and the RH (relative humidity) limit, which is an important factor in monitoring the moisture in concrete (because moisture is needed to help concrete cure).

The next page is where the magic happens: the simulation page. A number of 2D and 3D plots can be accessed and, when ready, the computation can be performed.

The 2D graphs will show how long it takes to reach a target humidity based on historical data as well as sensor data, and the 3D FEA plots will show a variety of factors changing with respect to time.

Figure 3: Simulation time. On the left is an FEA plot showing temperature of a concrete slab. On the right, sensor data vs historical data.

And finally, you can generate a Word document report on the simulation results, just to summarize your findings.

You can see our video of the whole process running on Android below.

If watching a simulation of concrete drying feels somewhat anti-climactic to your expectations of what a digital twin should be, then you can blame the marketing people who have been ramping up the digital twin hype over the last few years.

When people think of a digital twin, they tend to think of large complicated 3D assemblies such as jet engines or wind turbines, displaying colorful representations of some huge datasets. In fact, if you go on Google Images and search for the term “digital twin” right now, chances are you’ll end up with a results page full of pretty wind turbines with glowing blue colors and X-Ray effects…

But this is pure hype. Hype and Photoshop.

Image result for digital twin

Figure 4: Is this a digital twin? Or just an artist's depiction. (Image Credit: GE.)

A digital twin is about fusing CAD, simulation and sensor data to create actionable data—data that can be used to predict outcomes based on certain scenarios—and to help businesses make productive decisions.

Where Are the Twins?

So, having finally seen a digital twin working for ourselves, we asked COMSOL why there are not more of them?

“Digital twins have been exclusive for very expensive products, such as jet fighters and other defense or space applications,” said Ericsson.

“The defense and space industry seldom discloses how they manage their fleets. The information that has been disclosed clearly shows the benefits of the concept when used for devices and products that are strategically important, that are made in relatively few units, that often have to operate at their maximum performance, and that even these few units have to operate in very different environments (tropics, hot deserts, Arctic climates).

“Early adopters of digital twins are expected to operate in similar industries, where a few expensive and very complex products are designed, manufactured, and operated. One example is wind turbines. These industries are usually high-tech modeling and simulation intense. Taking the next step to adopt the digital twin concept is not a large step when the benefits are clear. But these industries have something else in common with defense and aerospace: there is fierce competition, there are few players, and these few players have relatively large resources.

“They have enough critical mass to make their own digital twin implementation. They do not need to cooperate with others in the community and therefore they do not need to share knowledge.”

But, as we see here, digital twins are well-suited to complex systems, they aren’t limited to them. Pretty much anything that can be mathematically modeled and can benefit from predictive analysis can potentially make good use of a digital twin.

Recommended For You