Digitization & AI Krones Optimizes Simulation in Beverage Production

Related Vendors

Softserve GmbH

NVIDIA GmbH

CADFEM Germany GmbH

ANSYS Germany GmbH

Microsoft Deutschland GmbH

more less

Wuxi InfiMotion Technology Co., Ltd.

less

At Krones, digital twins with integrated AI agents now ensure greater efficiency: The agentic digital twins combine real-time simulations with artificial intelligence capable of reasoning and making decisions. This increases efficiency in beverage production.

Krones reduces simulation times using artificial intelligence (AI) and Open-USD from 3 to 4 hours to less than 5 minutes. This is made possible by physically accurate digital twins developed based on the Nvidia Omniverse libraries with AI agents. For this, the company collaborated with Ansys (part of Synopsys), Cadfem, Microsoft, Nvidia, and Soft-Serve.

With the Agentic Digital Twins, we demonstrate that digitization and AI are not just visions but already deliver tangible efficiency gains and sustainable benefits today.

Markus Tischer, Member of the Executive Board of Krones AG

AI Agent Reduces Effort for Simulations from Four Hours to Five Minutes

Within just two months, Krones managed to fundamentally improve its simulations of machines and systems and elevate the efficiency of beverage production to a new level:

• Previously, individual simulation runs took three to four hours, but this was too slow for real-time optimizations in a highly dynamic production environment.
• The limited ability to quickly test various scenarios delayed decisions and led to higher resource consumption, inefficient production, and a lack of agility.
• Krones therefore needed a solution that enabled faster, scalable, and smarter simulations to remain competitive in Industry 4.0.

In an optimization loop, the AI analyzes its simulation results, adjusts the parameters, and then initiates new runs.

Together with Ansys (part of Synopsys), Cadfem, Microsoft, Nvidia, and Soft-Serve, Krones developed a physically accurate digital twin capable of performing such simulations in less than five minutes. It allows its AI agents to test various scenarios and settings in a virtual model, thereby automatically optimizing workflows. In an optimization loop, the AI analyzes its simulation results, adjusts the parameters, and then starts new iterations. This enables real-time testing and comparison of factors such as speeds, temperatures, or process controls and their effects.

Digital Twin Independently Finds Optimal Settings

In this way, the digital twin learns independently which settings deliver the optimal results. Its best solutions are then transferred to the real system. This allows production scenarios to be compared in real time, optimizations to be implemented immediately, and resources to be used more intelligently. The execution of these computationally intensive simulations takes place in the Microsoft Azure Cloud on Nvidia's platform for accelerated computing. This scalable infrastructure accelerates calculations by multiple times.

Digitally Map Beverage Production And Optimize Decisions With AI

According to a statement, this is a breakthrough for the entire industry: digital twins with agent AI combine simulation, automation, and artificial intelligence to create virtual models of machines, plants, or processes that continuously improve themselves. Until now, digital twins still required human decisions and interventions. The new generation, on the other hand, utilizes its judgment, decision-making, and action capabilities to make its own decisions based on real-time data. Its AI agents simulate scenarios, respond to changes, and can automatically trigger workflows. This enables the creation of automation systems that independently develop improvements rather than simply reflecting the current state of a plant.

Digitally Map Down to the Last Drop

These advantages particularly benefit Krones' customers in the beverage industry:

• The processing of liquids requires a precise interaction of physical processes that are constantly changing. From filling speed to bottle design to flow behavior, the dynamics are highly complex.
• For this reason, pressure, flow, and turbulence must be continuously monitored, predicted, and optimized in real time.
• Only in this way can resource waste or disruptions in later production steps be avoided. Every single unsealed bottle can pose a new challenge for the control, operation, and simulation of the machines.

The answer to this lies in digital twins with AI agents. They digitally replicate the physical world down to the last drop and can make their own decisions to continuously optimize processes. Together with its technology ecosystem partners, Krones is redefining how machines can behave and be simulated. The use of AI and cloud computing continuously optimizes liquid processing, enabling its customers to produce their products more efficiently, sustainably, and productively.

Strong Ecosystem And Interdisciplinary Collaboration

This success was made possible through close interdisciplinary collaboration: regular engineering workshops, joint development sprints, and a clear focus on business outcomes achieved progress in the shortest time. The results include faster commissioning of machines or plants, higher efficiency, fewer downtimes, and data-driven performance delivery. Water consumption decreases, and the systems are optimally configured around the clock.

The following partners were involved:

• Ansys (Synopsys): High-precision flow simulations
• Cadfem: Application expertise for complex simulations
• Microsoft Azure: Scalable, secure cloud infrastructure
• Nvidia: Open libraries, frameworks, and GPU-accelerated simulation
• Soft-Serve: Integration of data, logic, and AI as the "brain" of the digital twin

Machine Authentication

Machine Operator Authentication - Why knowing "who" matters