Last year, consumer products leader Unilever pilot tested a digital twin of a proprietary manufacturing facility in Brazil. The digital twin has since saved the site $2.8 million in operating costs and provided a 3% increase in productivity. Through a partnership with Microsoft, Unilever plans to deploy virtual twins of 70 more plants by year-end.

The demand for efficiency is at an all-time high, and as (thanks to the “Amazon Effect”) lead times are being squeezed into nigh-impossible windows, supply chain visibility becomes key to success. While no crystal ball can predict supply chain setbacks, it’s now possible to achieve real-time, detailed visibility of entire processes. With a digital twin of supply chain processes, companies can simulate, predict, and gain insights into every level of logistics execution.

What is a Digital Twin?

A digital twin is a virtual model of the physical supply chain that includes a digital counterpart of every piece of the process. But unlike other graphic renderings, this model is dynamic, as data streams from IoT-connected devices and is then paired with AI to continuously monitor and update, effectively reflecting the current state of every moving piece.

How Does a Digital Twin Work?

While the logistics applications for digital twins have only recently been made possible through strides in technology, the concept of a “digital twin” has been applied for over 30 years. NASA used numeric models to simulate the first rocket launches, and countless manufacturers have since been using CAD models and process simulations to design, qualify, and implement production.

Digital twins are made possible through IoT-connected devices, in which the ecosystem continuously relays information and synthesizes data. This synthesis allows process engineers to identify historical patterns, define root causes, and optimize processes. When paired with machine learning, an IoT-enabled digital twin can reduce machine downtime, decrease waste from production runs, and minimize quality errors. They further enable engineers to see data points in their environmental context, and to analyze root causes to pinpoint the exact conditions or material that caused failure.

Supply Chain Digital Twin Use Cases

Through sensors embedded in machines, data on temperature, speed, and other production variables is sent to the digital twin where algorithms process this data to identify the optimal operating conditions. With a facility twin that mirrors a network of machines, managers can monitor equipment health in real-time and practice robust predictive maintenance.

The visualization provided by a warehouse’s digital twin can give timely and accurate visibility to inventory levels, while AI software can make predictions – or even autonomous decisions – about deliveries or stock. In a fully integrated shipping system, IoT sensors can be placed in individual containers to track location, monitor for damage, handling conditions, or contamination, and predict the most efficient routes.

Perhaps most invaluably, an AI-powered digital twin can model scenarios and test solutions to production problems through machine learning. By running simulations and what-if scenarios within the twin, process engineers can save time and avoid risks before actual implementation.

Implementation

Companies can create multiple types of twins at successive levels of visibility, representing a single process, a manufacturing site, and eventually an entire ecosystem. These twins can be deployed as on-premise, cloud-based, or hybrid systems, depending on the needs of the company. Wherever the twin is housed, companies should consider the end-users and who will be accessing the data, starting small and testing prior to full-scale implementation.

Challenges

Because IoT-connected devices are the building blocks of a digital twin, sensor integration is key to enabling even small-scale visibility. Designing, procuring, and integrating these sensors can be resource-intensive and time-consuming, and it can be difficult to stream data from older machines that aren’t technologically compatible.

Digital twins require an incredibly complex systems landscape, and in supply chain, complexity can induce risk and inefficiency. While computational complexity taxes hardware, operational complexity can overwhelm operators and detract from tangible benefits. Getting employees on board is one of the biggest challenges in technological change, and if the end-users cannot engage with the software provided, the entire implementation can fall short of potential returns.

Deep strategic analysis is necessary when considering the benefits of a digital twin, as expertise is required for both the implementation and maintenance of the twin upon deployment. Once deployed, a digital twin will require monitoring of components and software, and periodic fixes and updates.  OEMs must be capable of supporting hardware and software, and agree to terms regarding data protection. CIOs should collaborate with functional leaders to develop business models that weigh benefits against material and maintenance costs.

Future

Gartner predicts that by 2021, half of all large industrial companies will use digital twins, deriving up to 10% improvements in production capacities. As competition in analytics and IT sectors increases, the market for digital twins is expected to continue growing at a CAGR of more than 30%, eventually reaching $26 billion by 2025. Although their use is novel and requires high-level planning and integration, digital twins have the capacity to transform supply chain operations at all levels. Leaders would do well to explore digital twin capabilities in their own facilities at a small scale to test what can eventually become a crisp, birds-eye view of the entire supply chain ecosystem.