Building a Digital Twin? Consider a Vendor-Agnostic HPC Cloud
Shawn Wasserman posted on October 18, 2017 | 4170 views
Digital twins help engineers predict real world behaviour, optimize designs and gain new insights and understandings of use cases. (Image courtesy of Rescale.)

Digital twins help engineers predict real world behaviour, optimize designs and gain new insights and understandings of use cases. (Image courtesy of Rescale.)

The concept of the digital twin brings versatility to the engineering world. By creating a virtual representation of a product, engineers can investigate designs to further product development, in-service optimizations and quality assurance.

Now with the implementation of the Internet of Things (IoT), the potential of the digital twin grows.

Engineers can link real-time data into their digital mock-ups, allowing for better understandings of the physical world.

However, digital twins don’t just come off the shelf. Since every modeled behavior is built in different system, engineering and IT teams experience a considerable challenge linking their disparate tools into one model. Joris Poort, CEO of Rescale, explained that this is where vendor-agnostic cloud HPC, like Rescale, can really shine.

Why Engineers Need a Digital Twin to Lead Product Development

Let the digital twin lead product development by helping you test and redefine your systems and designs. (Image courtesy of Rescale.)

Let the digital twin lead product development by helping you test and redefine your systems and designs. (Image courtesy of Rescale.)

So, why would an engineer want to link their products, the internet and their simulations into a digital twin? Well, with the IoT and digital twin, engineering teams can better predict the behaviour of their products, assess trade-offs and fuel innovation. This linkage also allows for improved definitions and validations of design requirements,as well as, explorations into previously unknown use cases.

“Companies that invest into linking the digital twin to the IoT can better predict how their designs, equipment and tools operate,” Poort said.

“You will be able to get feedback from the real-time data you input to the digital twin,” Poort added.“This is powerful for a company, as it opens opportunities for in-service predictive maintenance and other insights. This tight feedback loop means you can more accurately tune your digital twin model, reduce margins of safety and produce higher quality products.”

One of the biggest limitations of the current simulation-based product development workflow is that you can’t always predict every use case.

“Over-engineering is typically done to deal with these unknown use cases,” Poort said. “However, the more you get back from the IoT and digital twin the more you can change in the product development as you learn about these unknowns.”

The idea is to analyze the IoT data that comes in, feeding it back into the digital twin with predictive software and/or simulation software. You can also compare the IoT data with the models you create to validate their results.

This feedback cycle is particularly useful in industries that produce products with long service lifetimes. Take an airplane, for example. An airplane might be in service for 25 to 30 years. This is an intolerably long feedback cycle to improve next year’s designs. However, if the airplane manufacture assesses its products during operation using the IoT, that feedback cycle becomes near instantaneous.

From this information,the company can better design its products or even schedule maintenance cycles so it can improve its products that are already out in the field. In other words, redesigns get to the customer faster.

How to Build a Digital Twin on Rescale’s Cloud HPC Platform

Rescale helps IT and engineers integrate business tools and processes into one platform. (image courtesy of Rescale.)

Rescale helps IT and engineers integrate business tools and processes into one platform. (image courtesy of Rescale.)

So, now you know why you need a digital twin, but how do you create it? Well,one of the biggest challenges is that data crunching, data storage and linking tools don’t always want to be linked together.

If you think simulations create large sums of data, then imagine the data collected from a system connecting the IoT to simulations in a digital twin.

This isn’t something you can easily download and fit into a database or crunch on-premise.

HPC cloud platforms, like Rescale, offer a great way to crunch these numbers. Cloud storage platforms, Amazon Web Services (AWS) or Azure, for example, are then recommended to store this data.

For Rescale’s HPC, the information coming in from the IoT, digital twin and simulation is all just data streams feeding into the company’s open platform.

“Rescale lets companies layer their tools from machine learning, to simulation, onto each other within our cloud network,” Fanny Treheux, director of solutions at Rescale, explained. “We are vendor agnostic. You use your own sensors and streams of data which feed into Rescale’s ScaleX platform.”

“The user typically stores their simulations and digital twin on ScaleX,” Poort added. “You just connect the IoT platform into ScaleX which then brings the data sets into the digital twin and simulations.”

Poort explained that one of the biggest challenges to implementing a digital twin is getting the initial project running, considering the significant investments organizations have already implemented into legacy CAE tools.

The question becomes: how do you get these disparate tools to interact without needing to purchase new tools from a single vendor?

The obvious deterrent to the single vendor digital twin is that it will force employees to change how they currently operate. Your team will be shoehorned into using tools they may not like and which may not truly fit the team’s needs. This will slow your digital twin development to a crawl.

“There are many ways of building a digital twin but the most effective is from the bottom up,” Treheux suggested. “You will have teams doing simulation work in independent silos, but the goal of the digital twin is to tie these sources together into one model that captures all the physics into one virtual design. The common misconception is to switch everything into one simulation or IoT platform. With Rescale, you can tie all these disparate software together in the cloud and access our computational centers together.”

Treheux further explained that this helps avoid a nightmare scenario for IT departments that are desperately trying to link software platforms together that don’t always play nice with each other. This can become a considerable logistical and data security issue.

“At the application level, an effective digital twin will not change the tools you are already using. It is more of a concept than something you are providing. It’s about bringing the data into a single location. Rescale’s approach is to harvest and manage this data from sometimes hundreds of applications. We tie these tools together from the bottom up using the hardware and software you already have within our pay-as-you-go service.”

Another added benefit to this bottom-up approach to the digital twin is that organizations will be able to gain insights as they continuously grow the system. Treheux noted that partially competed digital twins will still shed light onto your operations. This brings the ROI of the digital twin project to the initial stages of its setup. With every tool you link into the digital twin, you will continuously start to learn more and more about your system.

Rescale’s digital twin offering does not focus on providing a data storage solution. However, as Rescale's ScaleX is an open platform, other cloud services providers can provide the data storage solutions. Poort said that Resale allows any external data repositories, simulation tools or IoT platforms to be connected.

Some of the connections to these disparate tools have already been made available through the Rescale system. However, if a tool you need to connect into the digital twin isn’t in this database, then application programming interfaces (APIs) can be used. This gives end users the ability to link tools from various vendors, be it PTC, ANSYS, AWS or Dassault Systèmes into one ecosystem.

The API system in Rescale is particularly useful, as much of the IoT data collected by Rescale’s customers are from sensors. Much of these sensors can be 10 to 15 years old and are from various vendors and don’t have a platform.

“We take these legacy sensors and connect them so you can extract data from that,” Poort said. “Many of these sensors don’t even have a platform. Platforms drive insights. But if insights are coming from a digital twin then you need a vendor independent platform.”

To learn more about Rescale click this link.

Rescale has sponsored ENGINEERING.com to write this article. It has provided no editorial input. All opinions are mine, except where quoted or stated otherwise. —Shawn Wasserman


Recommended For You