Digital Twins and Simulation: Unlocking insights into the future
By now, digital twins are not so futuristic—these digital replicas of physical and virtual systems are already making impacts across a range of operational environments. Racecar teams can test and interrogate the performance of virtual replicas of car parts, like brake systems and tires, before the car ever hits the track. Manufacturers can create virtual replicas of their smart factories to simulate performance and uncover new efficiencies. And operations specialists can map out their entire supply chain networks to pinpoint potential vulnerabilities.
But what about the more strategic—and sometimes messier—business challenges many companies face? In our recent article, New uses for digital twins in the race to navigate an uncertain future, we highlight cases of how business strategists are working with digital twin architects to combine digital twins with advanced simulation methods (think agent-based modeling or Monte Carlo[i]) to help tackle more strategic questions that incorporate human preferences and decision making—something less predictable than machine or factory performance.
In the new frontier of digital twins, unlocked by enhanced access to data, computing power, and AI advancements du jour, many businesses are beginning to deploy more strategic simulations for better decision-making, including:
· The simulated M&A transaction. Companies considering a major acquisition should consider myriad uncertainties: How will it impact the combined workforces? Will the market respond positively enough to the newly combined company to merit the substantial capital investment? Should the additional physical assets, like stores or warehouses, remain separated or should they consolidate? By building a digital twin of the separate ventures, a company can simulate various choices to better understand if a merger or acquisition makes financial sense and its optimal path forward.
· The marketer’s new best friend for product launches. Companies could create a digital twin of their customer environment to help marketers train and prepare for a new product launch. Before the product ever launches, teams can create a virtual environment where they could test various marketing strategies (e.g., channel investments, messaging) to illuminate the best course of action—and the potential paths to adjust their strategies—once the product hits the market.
· The impact of climate events on city planning. Getting a clear view into traffic patterns is often no longer enough for many city planners dealing with climate events. Cities can have digital twin architects use these technologies to simulate the impact of rainfall and flood dynamics across the city. This can help city planners to implement better flood management strategies and emergency response plans.
The next frontier of digital twin simulations:
While these new examples of strategic use cases appear promising, there’s a reason it’s taken longer to gain traction here versus the more operational endeavors mentioned at the onset of this article. Consider: 1) the data can be harder (sometimes impossible) to capture, 2) the computing power required to model complex environments is often massive, and 3) the ability to scale such methods across a variety of business problems isn’t always straightforward.
Still, just as recent advances in data and AI have led to new opportunities for strategic simulations, new developments appear poised to further expand the capabilities of digital twin technology, to help tackle another frontier of highly complex and uncertain challenges.
Not enough computing power? Quantum simulations are emerging on the scene. Data is the backbone of simulations, but sometimes the questions that are trying to be answered can require so much data that traditional processing methods fall short of efficient analysis. In response, some companies are in the early stages of building “quantum digital twins” that rely on quantum computing power to help fuel their digital twin simulations in a more efficient and scalable manner.
For instance, some major pharmaceutical companies are experimenting with quantum digital twins to more holistically analyze the impact of new treatments for cancer and Alzheimer’s, simulating how potential treatments might affect the molecules involved in these diseases.[ii]
Have data gaps? Generative AI’s synthetic data can help. Rather than having too much data to process, some problems lack the data necessary to build a digital twin in the first place. This could be due to a lack of representation of certain demographics or simply the inability to capture the data in the first place. For this reason, some companies are experimenting with using Generative AI to produce synthetic data. In this process, the user can feed the available data, along with basic parameters, to a machine learning program, which can then produce a larger sample of synthetic data to inform a digital twin.
Similar to the characters in a video game, these synthetic agents can interact with a simulation to hopefully provide a clearer picture of how an environment could be expected to respond to changing dynamics within the system. While these types of models require extra vetting to confirm an accurate representation of a system, they can provide a new potential pathway to simulating complex environments.
The future looks clearer. Challenges will still arise as companies expand the scope of use cases for digital twin simulations, but as technology continues to evolve, the collective ability to gain a clearer understanding of complex environments will likely continue to grow in-step. The next frontier of corralling uncertainty could be on the horizon. One day, the phrase “predicting the future” may just give way to “simulating the future.”
- Tim Murphy, Senior Research Leader, Deloitte Center for Integrated Research
[i] This method uses computer simulations and probability distributions to estimate a range of potential outcomes for a given model. It helps leaders understand the uncertainty in their business decisions by considering different “what-if” scenarios for key factors.
[ii] Dr. Christina Yan Zhang, “Quantum, AI and Brain Computer Interface Powered Digital Twin Applications for the Future,” TechUK, October 2023.