The rise of digital twins in modern manufacturing

For manufacturers that are not familiar with digital twins, the concept may seem amorphous. Simply put, a digital twin is a virtual representation of a real-world process or environment such as a manufacturing plant or supply chain. Manufacturers can use digital twins to optimize productivity by modeling improvements to their factories or supply chains digitally before making expensive changes in the real world.

There are two main types of digital twins. The first is a simulation built from historical data—on a production line, for example—to create a high-fidelity model that emulates real processes. This is an inexpensive way to run multiple scenarios to predict equipment failures and spot bottlenecks and inefficiencies. It lets businesses test the value and impact of improvements or innovations before they are made on the physical production line; think of it as a virtual laboratory to test a hypothesis.

The second type is a live digital twin that uses real-time data to monitor the performance of core processes (like picking, packing and shipping in a warehouse), manage logistics or capture real-time data on a production line. When enhanced by predictive artificial intelligence and machine learning tools, live digital twins can generate forward-looking performance estimates that respond in real time to production changes or disruptions.

Manufacturing use cases

Manufacturers can use a point-in-time simulation to create a digital production model before investing in a physical model. Modeling the design of a new manufacturing line to reduce suboptimal capital expenditure decisions is one example. Alternatively, organizations can create a digital replica of an existing factory or warehouse to simulate operational improvements and validate process adjustments before making costly changes to existing operations.

With a continuously running digital twin model, use cases tend to get broader, more iterative and increasingly scalable. Model capabilities include predicting sales demand by ingesting external econometric data and correlating it with historical sales. To scale these insights, companies can use demand signal data to inform purchasing requisitions. Managers can then review, modify and approve requisitions based on current raw materials inventory and operational needs. This creates a predictive demand profile, which can be expanded to include product, geography or customer data, enabling “what-if” scenarios to support data-driven materials planning and production. For this model to work, companies need to integrate enterprise data from multiple systems in the cloud to break down data silos and gain visibility.

Using a continuous digital twin model on the manufacturing line can also improve machine uptime or utilization through sensors that capture factory floor data such as vibration, amperage and thermal measurements. Using predictive tools, a digital twin can flag anomalies that indicate, for instance, an overheating bearing or amperage variance, suggesting a motor may need to be replaced and maintenance should be scheduled.

The value of various digital twin use cases for manufacturers typically comes in the form of optimized inventory, lower capital expenditures, increased EBITDA, enhanced throughput, reduced risk and improved customer service. All of these benefits can drive more efficient and reliable operations.

Digital twin uses in other industries

An analysis of five years of companies’ public filings—such as quarterly reports, annual reports and earnings calls—shows the increasing prevalence of digital twins in the market.

The technology and industrials sectors lead the pack by a wide margin, with the highest mentions of “digital twin” in public filings during this time frame. For both sectors, such mentions have generally been on an upward trend over the last five years.

Creating and implementing digital twins

Typical challenges implementing digital twins include data integration and quality, resource availability and skills, and dealing with organizational change.

Working with a third-party advisor that has experience implementing digital twins can be a worthwhile investment. Here are some steps an external advisor can help with:

• Testing and validating a digital model for compatibility with real-world scenarios and adjusting it as needed for accuracy
• Training and supporting employees to effectively adopt and maintain the digital model
• Integrating live internal and external data tables and developing a unified, cloud-based data model
• Mapping processes, materials and information flows between internal and external silos
• Equipping the digital twin with machine learning and AI capabilities to predict outcomes

The potential of digital twin technology for manufacturers is vast. Implemented strategically, this capability can save money upfront by helping companies ferret out which investments may not be worthwhile. It can also equip companies with valuable information to fix bottlenecks, thereby increasing throughput. More broadly, the power of these simulations can help manufacturers improve supply-demand forecasts, enhance the accuracy of sales and operations planning, and reduce overall materials costs.