Skip to content

Virtues of the Virtual

Ilene Wolff
By Ilene Wolff Contributing Editor, SME Media
Jamco Corporation, which makes seating, lavatories and galleys like this one for Boeing, Airbus, and others, created a digital thread based on Aras Innovator, a low-code product innovation platform, to manage 3.2 million CAD data sets.

When passengers enjoy a beverage and a treat on a Boeing 787 Dreamliner, their snack comes from a galley made by aircraft interiors manufacturer Jamco Corporation.

Along with Jamco’s success in making galleys, lavatories and seating for Boeing, Airbus, and others, came integrated management needs for 3.2 million CAD data sets. Despite having all of that data, the company found itself unable to answer some basic questions. Why a product was designed the way it was? Why it often put design flaws into a part despite its past experience of changing the CAD in other parts to eliminate that same design flaw? Why it deleted an essential element to reduce costs only to have to it added back later? Its solution was to create a digital thread based on Aras Innovator, a low-code product innovation platform. The platform allows Jamco’s designers and engineers in Japan, France and the United States to search for design history and use the related information to understand why a product was designed the way it was and stop repeating past errors.

“By connecting search results with our information, including deliverables and various regulations, our designers can search every process through to completion at any time and will reach proper information on Aras Innovator without depending on technical succession between engineers,” stated Hiroshi Sakurai, in a case study. Sakurai is the deputy general manager, engineering and technology division, aircraft interiors and components group at Jamco.

Aras’ platform and applications are among many tools manufacturers can use to create digital threads, which are ideally composed of data associated with a product during its entire lifecycle. A digital thread is a log or a record starting with the CAD data, and can include information from the machine’s PLC, maintenance done on it, sensors, energy use, and more.

Connected AR factory operations provide real-time performance insights. (Provided by PTC)

The company, and others, also has tools to create digital twins, which are virtual models connected with physical assets that are used to understand how to make changes and gauge the impacts of those changes without having to affect the real world. These assets can include people, processes, workflows and a single machine—or a system such as an entire factory. Among other uses, digital twins can help with safety planning and in error-proofing a part at the design phase.

Generally, the tools for building a digital twin and thread include ones already familiar in manufacturing—CAD, PLM and IoT applications.

Innovator is open source, while its applications and those of other vendors are offered usually with a paid subscription in a software-as-a-service paradigm. Other variations among the tools include off-the-shelf vs. custom, ease of use, flexibility, support for customized applications, inclusion of low-code apps, models to represent the elements in a digital twin, bundled or a la carte applications, and more.

What various providers largely agree on, however, are the rewards to be gained from creating a digital twin, digital thread, or both.

Digital twin enriches suggestion box

Jamco’s experience demonstrated the utility of using a digital thread to provide history and context in the design stage while eliminating repeated errors, all in the quest for greater efficiency and improved quality.

Another use would be for maintenance, said Craig Melrose, executive vice president of digital transformation solutions at PTC.

“I can provide them the maintenance understanding and experience through augmented reality (AR), telling them disassemble this first, check this item second, inspect this item third,” he said. “If I need to replace an item, here’s the details on how to replace it or reinstall it. How to confirm it’s done correctly. How to re-assemble the piece of equipment and verify it’s all been done correctly and ready to turn on again. And even safety items like turn the power off and lock out the piece of equipment, these types of things, can all be delivered through AR.”

A digital twin offers advantages of its own.

“The benefits are so much larger than the investment,” said Zohair Mehkri, a director of engineering responsible for digital twin at Flex Ltd., a $24-billion global contract manufacturer. “For example, you can have full optimization of your processes, of your buildings, before anything even happens. You can move things around, you can change products, put in new machines, add or remove head count or material and you can try all of that in the software before you even step foot on your factory floor.

“The amount of time and effort we spend going back and doing trial and error on physical manufacturing processes that affect us financially is huge. So the ability to do that in a digital twin is enormous.”

Once a digital twin is created, manufacturers can start monitoring, optimization and analysis. Each of those steps has its own benefits.

“An analysis of a process using a digital twin is very powerful because it will not only describe your process it will do things like tell you areas that need to be improved,” said Mehkri. “It doesn’t have to be a bottleneck. It could even be, ‘You have some empty space here, you could probably utilize it.’ Or you can take a non-value-add task out of your process.”

The key thing to keep in mind is a digital twin is connected to the physical model with hardware and software, he said. This can lead to higher benefits due to the closed loop.

“If something happens in the physical system, the digital model should know that a change in the physical system has occurred, the digital model should adjust itself to reflect the physical system and then optimize itself based on the change and then pass that optimization to the physical system,” Mehkri said.

If a bottleneck exists, a digital twin can offer solutions to correct it, said Ali Ahmad Malik, assistant professor of industrial and systems engineering at Oakland University.

“For example, in an assembly cell your target is 1,000 units in a day, but at a certain hour the digital twin evaluates it and tells you if you continue to work at the same pace you will not be able to achieve the goal,” he said. “But it is not only telling me the potential problem that may occur by the end of the day but also gives me some solutions, because it has the opportunity to simulate it and it’s intelligent. It has information from previous incidents. Suggestions may include adding a robot or increasing the speed of a robot. I can only achieve this if I have data connectivity and my digital twin keeps becoming intelligent from past events and past learning. So a dimension of machine learning should be enabled in it.”

Malik, whose experience with creating digital twins is with robots or cobots, said the twin can also help with assembly process balancing. In assembly cells, tasks are assigned to humans and robots working together to build an assembly in a given amount of time so no operator sits idle. Before digital twin, the process was always done manually.

Zohair Mehkri, a director of engineering responsible for digital twin at Flex Ltd., a $24-billion global contract manufacturer, says the benefits of creating a digital version of an asset far outweigh the time, money and effort spent using trial and error on manufacturing processes. (Provided by Flex)

“With digital twins that continuously evaluate the robot’s speed it keeps balancing the process and assigning tasks to the right source,” he said. “So all this manual activity of process balancing goes to a digital twin. If you have high mix, low-volume production this digital twin can greatly help you.”

It will also help a factory react more quickly to market fluctuations because manufacturing lines or cells can be quickly and virtually added, deleted or modified to react to changes in markets, tested via simulation and accepted or rejected based on results. In regulated industries, there’s even talk of pre-qualifying a line digitally before it’s qualified physically.

Melrose said the quick and virtual changes that are tested with simulation can keep production humming in a plant.

For example, Melrose said, take a fictitious plant with 10 machines. Workers have a hard time changing parts in and out of a machine because they’re so heavy, so the factory owner wants to add a lift assist.

“Rather than interrupt production, I can test all of that offline in a computer to make sure that it’s going to work well,” he said. “I’ll install and test it over the weekend and turn it on Monday so that we can start running production again with that new added item.”

Above all, what the digital twin provides is a context and framework for sensor data flowing from a smart machine, said Rob McAveney, CTO at Aras.

“If you think about it as I have two different airplanes with two different engines giving feedback on how one engine is operating vs. how the other engine is operating,” he said. “That feedback is valuable but it’s much more valuable if I know which aircraft they are installed on, where that aircraft is flying, what the maintenance history of that aircraft is. All those things are important in terms of analyzing the data that’s coming back from those sensors. So the digital twin, that’s for me where it provides the most value.”

Thread, twin apps could be more CAD-like

Where McAveney sees the biggest challenge is in disconnected data from disparate systems.

“You have data in different formats that are used by different individuals in different departments that never talk to one another,” he said. “If they are required to provide their data, they have to manually massage it to make it applicable to someone else’s needs.”

Having disparate systems in and of itself is not bad, it’s the fact that they’re disparate systems with no connection.

“And I don’t mean [don’t talk] from a technology perspecitive, I mean [don’t talk] from a semantic perspective,” said McAveney. “Meaning I have a bill of materials in PLM, I have a bill of materials in ERP, and there is no mapping, there is nobody sitting in the middle saying ‘Oh, that field in that system is the same as that field in that system.’ We need to make sure they’re synchronized. Now there’s not even a conceptual connection between them never mind a physical connection.”

Malik compared the situation McAveney described, of disconnected data, to that of data from various CAD software vendors.

“For example, AutoCAD data was [once] not exportable or importable into SolidWorks, but now it is a very normal thing to export and import data from one CAD to another CAD and there are standard formats,” Malik said. “The same thing needs to be done in this dimension and not only CAD data but also dynamic data and the information we get from physical systems. Right now we’re limited to one company’s platform.”

Vendors and users agree the first and most important step to take in using a digital twin is to determine your objective and which parts of your physical system you want to model.

A factory owner may simply want to visualize their plant, Mehkri said, and answer questions such as: Where are my materials? Where do people spend their time? What are high-traffic areas? Where are issues in terms of the line?

“When you start to visualize those types of challenges, then you can go into that second factor, which is how do I solve those problems, how do I optimize those problems, or how do I get rid of those problems?” he said.


Hands-free augmented reality guidance provides visual work instructions for complex tasks. (Provided by PTC)

Melrose agreed choosing the right objective is key.

“Companies are rightfully risk averse but in being risk averse maybe choose the wrong use case which is low value and, potentially because they’re not sure of the value, choose the wrong technology based on price rather than based on impact,” said PTC’s Melrose. “They can get into pilot purgatory.”

While some vendors discourage customization, at least one company not only allows customization, it encourages it.

“You are free as a customer to augment those applications with additional capabilities and what Aras does is provide the ability for Aras subscribers to use applications as-is, augment applications as they see fit, build new applications as they see fit,” said McAveney. “Then it’s our job to make sure that all those changes that the customer has made are able to move forward with technology.

“As Aras continues to release new capability and new versions of the platform that work with newer compute stacks we guarantee as part of our subscription that anything you build today will work next year.”

In Mehkri’s experience creating digital twins, he relies on lean manufacturing principles when there’s a mismatch between his twin and the real world. Listing all of the possibilities for what caused a difference is impossible, he said.

“There are some possibilities that can happen however; I could have modeled incorrectly, there could be a user error, there could be different data sources or the data source itself can have a problem,” said Mehkri. “The way that we usually determine that is by using the Power of Gemba, a Lean Manufacturing technique, which is going to the place where the work is done. By going to the physical floor we are able to view and immerse ourselves in the process and perform an RCA (root cause analysis) on what has caused the discrepancy and then go after that and figure out the problem.”

One area Mehkri and the rest of the industry struggles with is not having enough off-the-shelf content to model materials, machines and the other elements in a digital twin. When there is no model in the software, the user has to create it himself, which is time-consuming, he said.

Malik said a user may need multiple products to build his digital twin. “Maybe one tool will enable me to design what elements are there in a physical system,” he said. Another tool may allow me to define the dynamics or kinematics of the physical system and then another software may enable me to integrate the data from the physical system into the digital system. So a combination of four or five different softwares may help me to complete a digital twin.”

Some of the prominent names among digital tools vendors have a comprehensive set of tools for building digital twins, but small and medium companies often can’t afford or don’t have people with the skills to use it, Malik said. Even among the bigger vendors, various applications may not transfer data smoothly and easily.

“But if you can find some software that has more features within one environment, for example Visual Components offers continuous simulation, data connectivity, discrete event simulation or stochastic analysis in one single environment,” he said. “That is a very good feature in my understanding because then you don’t need three or four softwares.”

  • View All Articles
  • Connect With Us

Always Stay Informed

Receive the latest manufacturing news and technical information by subscribing to our monthly and quarterly magazines, weekly and monthly eNewsletters, and podcast channel.