Skip to main content
Exploring what’s next in tech – Insights, information, and ideas for today’s IT and business leaders

Answer these 5 questions to fine tune your digital twin strategy

Let the business goal define how closely your twin must mirror the real world.

Imagine a city water utility with a software tool based on sensor and telemetry data combined with customer service, billing, and other kinds of data. The tool allows the utility to not only monitor the water supply systems in real time but create forecasts on water consumption and simulate the impact of equipment failure

Aguas do Porto (AdP), a Portuguese utility that supplies the city of Porto, has such a tool. It's an example of a digital twin, a software model that represents physical objects or processes using data.

A digital twin can be of great value in a variety of applications, but people make a lot of mistakes getting started. Use these five questions to get grounded and set your project up for success.

1. What do you mean by a digital twin?

Digital twin has many definitions. Something as simple as an online transaction balance can be a digital twin, since the digital record represents actual money, says Lin Nease, chief technologist for IoT at Hewlett Packard Enterprise. Some claim a static 3D representation of an object is often good enough. For others, only a dynamic 3D representation with real-time updates about the state of every component—and even its service history and maintenance requirements—is a full digital twin.

In a February 2022 Forrester Research report, principal analyst Paul Miller describes design twins consisting of 3D CAD designs and simulations, process twins including simulations of factory floors and devices, and service twins that use operational data to simulate a product's performance in the field.

The definition and use of digital twins varies by market. In real estate rental and sales, a static 3D model is good enough for the vast majority of customers looking to sell or rent, says Harry Ravenhill, director and founder of Motion Real Estate Media, which provides 3D models and virtual views of property.

Only a small portion of the real estate market needs digital twins that show "every aspect of the physical building environment" and the ability to drill down and follow, say, a steam pipe on its way through the building, agrees Jason Pohl, customer success manager at software and services provider Buildings IOT. But if an engineering group needs to assure the cleanliness of the air in a semiconductor plant, the digital twin will need very precise, real-time updates from sensors at every critical location.

2. What do you want from your digital twin, who will use it, and for what purposes?

In the manufacturing, healthcare, and innovation-driven industries he serves, most clients have a clear business goal for their digital twins, says CV Ramachandran, partner at innovation and transformation firm PA Consulting. Such goals include increased productivity, higher quality, and the creation of new product designs. An increasingly important benefit, he says, is the use of digital twins in training new workers quickly, helping to ease staff shortages.

Please read: Modeling and simulation: Achieve next-level results with AI

Among the business uses for service twins, says Miller, is empowering product manufacturers "to pivot to as-a-service business models," using performance data from digital twins to charge customers based on ongoing machine usage and output or to sell data and recommendations based on that data.

Other companies are using digital twin data to predict and prevent machine malfunctions and influence the design of future products, says Miller.

However, some real estate clients "have heard of digital twins … but start to lose interest when you start asking hard questions about what data" they need to create them, says Eoin Kiely, director and digital asset lead for global consulting firm Turner & Townsend.

"Most customers don't understand what they need," says John Zimmerman, COO of digital twin software provider Beast Code. "They need a lot of education about what you can build for them and what they can use it for."

Digital twins are also useful in city planning, says Nitin Agarwal, worldwide lead for smart cities at HPE. The city-state of Singapore, for example, has used 3D visualizations to help model objects such as the pipes and cables workers will encounter when building a bridge and determine the most attractive design fo

3. How smart or knowledgeable do you need your digital twin to be?

"The tricky part is how much detail you put into the digital twin" to achieve the business goal, says Ramachandran. "You could spend all your time collecting data and almost get lost in that." If the goal is to reduce the percentage of scrap in a production process from 15 percent to 7 percent, he says, the digital twin can't contain so much detail that the costs of that data outweigh the benefits. Such costs can include wireless communication, analysis and data visualization, security, and the training required to make the digital twins usable.

Please read: How edge-to-cloud computing powers smart grids and smart cities

The "dream behind the digital twin" is that a user with augmented reality glasses could see all the maintenance procedures for a part that needs repair, down to the proper torque to apply to a bolt, says Nease. However, the data integration required "to make such a reality is usually quite extreme," he says, making it vital to create proper processes to ensure the right data is available and usable.

There are "not a ton" of use cases where a building owner or manager needs data updated in real time or even every few seconds to a digital twin, says Pohl. In most cases, updates every five or even 15 minutes are adequate.

Industry expertise can help determine which data will be most useful in a digital twin by allowing its designer to quickly develop a list of what-if hypotheses that are most worthy of testing. The data needed to test those hypotheses are thus most worth spending time and effort on. In pharmaceutical production, says Ramachandran, a modest investment in more frequent tests of the raw material in a digital twin of a production line can deliver "huge" returns by improving production yields.

4. How will you manage the required data?

Data is both the enabler of a digital twin and a huge cost. Practitioners recommend not only determining how much and what types of data to gather but also creating an integration and management layer to ensure the right data can be easily shared among current and future digital twins.

Pohl recommends having "a normalized dataset available to other systems to make intelligent decisions on. You have to have the relationships in the data model complete. You can't have an independent data layer without data modeling." Such an independent data layer makes it easier to add more and different types of data to the digital twin over time. "This is incredibly important and must be done at the very early stage of product architecture and project scope," he says.

Usage data from sensors and contextual data from enterprise systems is essential "to understand a product's current state in the field," says Miller. "But creating this 'digital thread' is challenging, involving an enterprise integration and data management strategy that crosses the boundary between information technology and operational technology teams." To meet this challenge, he recommends using the data aggregation capabilities found in some IoT software platforms.

Organizations also need to determine where they will host the data associated with the digital twin and how they will access it, and assure its scalability and security, says Zimmerman. Frequent challenges include data owners failing to follow data standards that make information sharing easier and integrating digital twins with product lifecycle management (PLM) tools, he says.

5. How do you build for the future?

To get the maximum return on investment from a digital twin, a business may want to expand it over time to simulate more physical devices or add detail so it can be used for additional purposes, such as training. That's why practitioners recommend creating a data and deployment strategy that can support those expanded uses.

Please read: How data helps Oracle Red Bull Racing's Hannah Schmitz make crucial race day decisions

For example, in creating a digital twin of a naval vessel, says Zimmerman, the original digital twin might only simulate the external shape to determine how much drag it will encounter and thus how powerful its propulsion system must be. Later in the design phase, engineers might want to add much more information about internal systems such as the propulsion system itself, weapons systems, and crew quarters.

For a new destroyer for the U.S. Navy, Beast Code is creating "an initial twin they can use for the engineering phase to make design decisions" but that can later draw information from a PLM system to create an application that will train the ship's crew while it is still being built. "So it's a digital twin for the full lifecycle of the product," he says, reducing the time required for training once the ship is completed.

Pohl recommends against using one-off data integration to create a digital twin for only one purpose. "What about the use case that will surface for a customer a year or more after the initial project is completed? Would a customer be able to tap into its own data to build a new feature or digital twin based on all the previously completed data integration and subsequent modeling? Would all the integration and data modeling required for digital twins be redone?"

In creating a long-term strategy, Zimmerman also recommends assessing where to most cost-effectively perform the graphics processing required to render highly detailed, interactive digital twins. Depending on the graphical requirements and the devices available to users, this might be best done using graphic accelerators offered by public cloud providers or by high-end client devices such as workstations.

Two areas into which digital twins will expand is simulating the state of objects not only within one company but across the value chain, from suppliers to customers, and to incorporate data from new sources such as text and video, Ramachandran predicts.

"We're in the first inning of a baseball game," he says. "We're miles away from the potential that digital twins are going to take us."

This article/content was written by the individual writer identified and does not necessarily reflect the view of Hewlett Packard Enterprise Company.