Real-virtual loops

Visionary companies leverage operational data and virtual models in “digital twins”

William J. Holstein
6 December 2017

7 min read

Manufacturers worldwide are building sensors and communications into their devices to collect real-time data. The most advanced are feeding this data into what analysts are calling “digital twins,” creating real-time feedback loops between in-use devices and the 3D simulations used to create them. Compass looks at how four companies are using the insights to improve their customers’ experiences.

The four-stroke diesel engines that Finland-based Wärtsilä designs and manufactures provide power for oceangoing freighters and cruise liners in 70 countries. With a typical lifespan of 25-30 years, they are some of the world’s largest engines. The 2015 edition of Guinness World Records also recognized Wärtsilä’s engines as the world’s most efficient.

How has Wärtsilä done it? Beginning in the 1970s, Wärtsilä recognized that building physical prototypes of each giant engine to identify and eliminate errors was an impossibly expensive proposition. So the company became an early pioneer in using sophisticated 3D modeling and simulation to achieve engine designs that were “right first time” when manufactured.

“Simulation comes very naturally in this business,” said Juho Könnö, manager of the company’s digital design platform. “It’s a lot about getting closer and closer to reality and, when you do that, you better understand your product and what the critical things to look at really are.”

Today, Wärtsilä is feeding its models and simulations with real-world performance data from hundreds of sensors installed on each new engine, creating a “digital twin” simulation that replicates actual, in-service operating conditions.

When used to drive the scientifically accurate 3D simulation, the data allows Wärtsilä experts to visualize how a specific engine is being used and to run “what if?” scenarios in search of opportunities to improve performance. Based on the analysis, Wärtsilä can recommend changes to settings and operating parameters to improve how ship owners operate their engines.

Or, Wärtsilä can incorporate identified design improvements into future engine designs.

GE Power Division is seeking to create fully accurate 3D digital models of its turbines, such as this electric power-generating 9Emax. The models are the first step toward creating a full digital twin to use in optimizing the design for different applications and operating conditions. (Image © GE Power Division)

“Getting the data from the real engines is really important for developing the simulation methodology,” Könnö said. “We try to use that as much as possible to calibrate our engines. We are also using it the other way around, using simulation models to see what and where we should measure on the engine. It’s a symbiosis.”


Worldwide, other innovative companies are striving to arrive at similar sweet spots in their own industries. Combining rapidly expanding computational power, sensor-equipped machinery and real-time data collection and analysis via the Internet of Things (IoT), these companies are driving intelligent 3D simulations to new levels, dramatically improving design and construction processes, manufacturing environments and customer-engagement outcomes.

As a result, Gartner, the information technology consultancy, ranked “digital twin” as one of its top 10 strategic technology trends for 2017.

While most companies have not reached Wärtsilä’s level of sophistication, leaders in other industries also are chasing the benefits of creating a virtual-real loop of continuous feedback and experimentation.

In automotive racing, for example, Onroak Automotive, a design and production unit of the Everspeed Group of France, is in the second year of a three-year project to overhaul the way it builds cars, trains mechanics and drivers, and manages races at Le Mans, the world’s oldest sports car endurance competition. The annual race requires that a vehicle be operated for 24 hours straight and complete 12 loops of a track that is a combination of closed public roads and a race circuit.

Wärtsilä gathers data from physical tests of its finished engines, then uses the data to fine-tune its simulation methodology. That methodology, in turn, is used to calibrate finished engines and plan sensor placements in future engine designs. (Images © Wärtsilä)

Even small advantages can make the difference between winning and losing, and Onroak Automotive believes its digital twin capabilities will give its teams a competitive edge.

“We can use this new system to design and to produce cars,” said Sébastien Metz, the Le Mans site director for Onroak Automotive. “We can also use it as training for the pit stop mechanics. You can know where the parts are placed and how they all fit together and what are the spaces for your hands and whether you can get access to the parts. You can train the mechanics and the drivers. It’s amazing what you can achieve.”

Onroak Automotive is particularly excited about using its twin to manage an actual race. Currently, Onroak Automotive has 10-15 data points on each car, which it can monitor in real time as the data flows into its 3D simulation via the IoT. But Metz anticipates the day when he could have 500 points of data feeding into the digital model, giving the team insights it can use to better manage the race, reduce cost and enhance its victory prospects.

“At some point, we will be able to simulate a pit stop (before it happens),” Metz said. “We can help the crew choose the right type of tire, depending on track conditions. We can simulate different weather conditions, whether it is raining or dry. We can simulate the real life of the car on the track.”

Like Wärtsilä, Onroak Automotive is seeking to achieve a two-way flow of information between the real and virtual car, insights that he expects will save 5%-8% on fuel. Eliminating even one pit stop, Metz said, could be the key to winning a race. Onroak Automotive also hopes to apply what it learns on the racetrack to high-end features for consumer vehicles.


In terms of sheer size, one of the world’s largest rollouts of the digital twin concept is occurring in Palu Special Economic Zone (SEZ) on the Indonesian island of Sulawesi, where an international consortium of energy companies is planning a €9.8 billion (US$11.5 billion) integrated crude oil refinery, strategic oil reserve and downstream petrochemical processing complex.

The Palu GMA Refinery Consortium (PGRC) in late September 2017 announced a four-year project to build a new greenfield refinery, inspired in part by the live-data simulation that nearby Singapore is creating for its entire city-state. Known as Virtual Singapore, the simulation allows city planners to visualize, understand and optimize everything from traffic, waste disposal and air quality to the placement of new buildings. PGRC’s digital model will help optimize the complex’s physical design and then streamline its construction, managing construction sequencing, coordinating subcontractors, eliminating errors and reducing waste and rework. Mohammad Rusydi, PGRC’s CEO and investment and finance director, said it is the first integrated refinery built using a complete virtual replica. Going beyond a 3D model, Rusydi said, PGRC’s digital twin is linked to two other dimensions – scheduling and project management, plus budget control.

“We call it 5D,” he said. “The project will be completed on time and we will not be facing any cost overrun.” Once operations begin, Rusydi said, “We can run a parallel digital refinery. Everything can be tested and simulated in the digital twin. We can attack a problem before it happens.”

For example, the digital model will alert managers to equipment that is wearing out, allowing them to replace it before it fails, and will be used to train operators on how to respond to a fire or other emergency. “It’s like an airline pilot doing training in a simulator,” he said. “If there is a fire in the cockpit, they know how to handle it.”

Rusydi said the consortium expects that insights provided by its digital twin will allow PGRC to build the complex for 25% less than blueprint-driven construction, with operational improvements of 15%-20%, compared to industry standards.

Virtual Singapore offers many services for citizens and visitors. Currently in development: a service for identifying personalized commuting options. Consider a sample scenario: Marisole usually takes the 7:45 am MRT rush-hour train from Jurong East station to her office near Raffles Place. In the future, she will be able to check Virtual Singapore for a less crowded option.The commuting function will show her choices, including a pre-peak train at 7 am and a bus at the same time. Both options qualify for a free pre-peak fare; because the train station is a shorter walk and the system shows an approaching rain storm, Marisole will choose the train. (Left image © monkeybusinessimages / iStock; right image © Virtual Singapore)


While new projects like the PGRC refinery complex begin life as a digital twin, project development and manufacturing processes created in a pre-digital era have the difficult challenge of retrofitting the concept into long-established processes. General Electric’s turbine division is dealing with just such a challenge. Its products – used to generate power for industry and utilities – include as many as 60,000 parts, according to Jeff Erno, head of Virtual Product Development for GE Power Generation in Greenville, South Carolina.

Because they are installed worldwide in a host of different climates and elevations, fine-tuning each turbine to achieve optimum efficiency in its particular setting is an ongoing challenge.

“A machine running on a top of a rainy mountain has completely different components than one operating in a dry valley,” Erno said. “That’s the nature of the beast.”

Each of the division’s different functions – design, parts engineering, systems engineering and manufacturing, among others – have traditionally focused on that team’s relatively narrow challenges, without full visibility into the overarching vision. Instead, each function works in specialized, disconnected computer silos, forcing them into sequential processes that interfere with real-time collaboration on interrelated issues.

“Nobody sees in virtual 3D what our product looks like,” Erno said. “The (legacy) tools historically do not handle the data well. The CAD systems and simulation systems don’t give you a good way to see what it looks like.” The first task for GE Power Generation, Erno said, is to create a “digital thread,” a single set of consistent, real-time data that each function and department can access and that automatically updates as changes occur anywhere along the value chain. Converting that digital thread into a robust digital twin of a core turbine, he said, would allow each function to optimize the design for different applications and operating conditions.

“That’s what we’re trying to use this technology for – to handle these different configurations,” Erno said. “You would not be burdened by managing them completely differently.” A single master model, he said, would accelerate the development of site-specific turbines and greatly reduce costs.

While companies in different industries and countries are at different points on the journey toward realizing the benefits of digital twins in their operations, Wärtsilä, Onroak Automotive, PGRC and GE Power Generation consider the technology a genuine game-changer – one that could enable enormous advantages over less visionary and committed competitors. 

For more information on Virtual Singapore

Related resources