Manufacturing Intelligence: Unpacking Digital Twin Use in Pharma Manufacturing

In this episode of Manufacturing Intelligence, Rick Jaenisch, Senior Director of Education, Outreach, and Digital Experience at Open BioPharma and Chris Cole, Associate Editorial Director at PharmTech, offer a wide-ranging explainer on digital twins in pharmaceutical manufacturing, covering definitions, current applications, practical requirements, and future possibilities.

Jaenisch grounds the discussion in the FDA's own definition, describing a digital twin as information constructs that mirror a physical asset and are dynamically updated with real-world data, emphasizing the crucial distinction between a static virtual model and a true digital twin. "If it isn't updated with either real time or somewhat periodic updates... it's not really a digital twin,” he says. “It's a virtual model."

The conversation highlights two primary use cases: risk reduction and scale-up modeling. By running hundreds or thousands of simulations virtually, manufacturers can identify potential errors and optimize processes before committing to a physical run, effectively letting "the computer fail" cheaply and quickly. Facilities management, AR-assisted operator training, and inventory tracking via RFID are identified as active applications today.

Scaling digital twins to full manufacturing lines requires robust data unification across building management, laboratory information management systems, and inventory systems, a significant integration challenge. Jaenisch identifies AI as essential, noting that "Digital twin without AI is pretty much just a virtual model." Generative AI's ability to find the needle in a haystack makes it indispensable for anomaly detection at scale.

Looking ahead, Jaenisch sees virtual inspection—in which regulators conduct non-disruptive audits through a real-time digital twin—as a compelling frontier. He also flags a persistent human challenge: first-generation employees often resist the increased sensor and camera surveillance that digital twins require, even when the monitoring is automated rather than human-led.