GERMANY – Krones has reduced simulation times for beverage production lines from three to four hours to under five minutes by combining artificial intelligence, physically accurate digital twins, and cloud-based computing.
The technology is being applied across machine design, production optimization, and line planning, including the company’s newly unveiled Ingeniq line concept, presented at drinktec 2025.
The development is the result of close collaboration between Krones and a technology ecosystem that includes Ansys, part of Synopsys, CADFEM, Microsoft, NVIDIA, and SoftServe.
Together, the partners have integrated AI agents into digital twins built on NVIDIA Omniverse libraries and OpenUSD, enabling simulations that run fast enough to support near real-time decision-making.
Digital twins that simulate, predict, and act
Unlike traditional dashboards that present historical performance metrics, Krones’ digital twins function as living models of filling and packaging systems.
They continuously simulate physical processes, predict outcomes, and trigger actions based on incoming data. This allows operators to test scenarios, forecast operational impacts, and adjust parameters before changes are applied to physical machines.
In practical terms, the system can simulate what happens if a filling line slows down, a production parameter shifts, or material supply changes.
The AI agents embedded in the twin analyze outcomes, adjust variables such as speed, temperature, pressure, or control logic, and rerun simulations automatically. The process repeats until the system identifies configurations that perform best under defined constraints.
Physics-based simulation meets artificial intelligence
At the core of the system are high-precision fluid dynamics simulations developed using Ansys Fluent, with application expertise provided by CADFEM.
These models accurately represent liquid behavior during filling processes, capturing effects such as flow, turbulence, and pressure changes. Running such simulations traditionally required hours of computation, limiting their usefulness in fast-moving production environments.
By integrating these physics-based models with NVIDIA Omniverse libraries and deploying them on Microsoft Azure using NVIDIA-accelerated computing, Krones has compressed simulation cycles to minutes.
The cloud infrastructure allows the system to scale computing power as needed, supporting multiple scenarios without slowing down production planning.
Agentic digital twins move beyond visualization
Krones refers to this new generation of systems as Agentic Digital Twins. Rather than serving as passive replicas, these twins reason over simulation results and make decisions autonomously within defined boundaries.
The optimized settings identified in the virtual environment are transferred directly to real machines, allowing operators to compare outcomes and implement changes quickly.
This shift alters how manufacturing decisions are made. Instead of asking what happened after a disruption, production teams can evaluate what is likely to happen next and what action should be taken immediately.
Human operators remain involved for oversight and exceptions, while routine adjustments are handled automatically by the system.
Application in the Ingeniq line concept
The digital twin framework is also being used in the design and optimization of Krones’ Ingeniq line concept, introduced at drinktec 2025.
Ingeniq is based on modular machine architecture, autonomous material handling, and AI-supported process control.
Digital twins support these features by simulating entire line configurations before physical installation, allowing engineers to test layouts, material flows, and operational limits in advance.
Through continuous data connections enabled by Krones’ Connect and Secure package, production data from Ingeniq lines feeds back into the digital twin, supporting predictive service planning and inventory coordination.
Rethinking performance metrics with automated decision engines
As digital twins take on a more active role, traditional key performance indicators are also being re-evaluated. Conventional metrics such as line efficiency, cost per case, or on-time, in-full delivery often operate in isolation, requiring managers to reconcile trade-offs manually.
Krones has introduced the concept of a unified performance signal, referred to internally as a “Mother KPI,” which combines cost, service, and sustainability parameters into a single dynamic indicator.
This signal acts as a control reference for the digital twin, allowing the system to balance competing objectives automatically, such as avoiding overproduction when storage capacity is limited or prioritizing delivery reliability over short-term cost savings.
Predictive inventory and automated logistics decisions
One area where this approach is being applied is inventory management. By forecasting consumption rates across multiple plants, the digital twin can anticipate material needs before stock levels reach critical thresholds.
This allows vendor-managed inventory systems to shift from reactive replenishment to predictive coordination, reducing emergency orders and excess safety stock.
The same decision logic extends to logistics. When a shipment is ready, the system evaluates carrier availability, equipment compatibility, and historical performance before tendering the load automatically.
If conditions change, such as fuel price fluctuations or capacity constraints, the twin simulates alternatives before committing.
Implications for sustainability and resource use
Reducing simulation times from hours to minutes also affects resource efficiency. Faster iteration enables tighter alignment between production, storage, and transport, reducing idle time, material waste, and unnecessary energy use.
In fluid processing, precise control of filling parameters lowers product losses and stabilizes downstream operations.
These efficiencies align with other sustainability targets across the beverage sector, where manufacturers face increasing scrutiny over energy use, waste generation, and supply chain emissions.
Industry-wide collaboration behind the system
The digital twin platform reflects contributions from multiple technology partners. Ansys provides physics-based simulation models, CADFEM supports complex application development, Microsoft Azure supplies scalable cloud infrastructure, NVIDIA delivers accelerated computing and open simulation libraries, and SoftServe integrates data pipelines, AI logic, and system orchestration.
Krones employs more than 20,000 people globally and reported sales of approximately US$5.7 billion in 2024. The company operates more than 100 subsidiaries and production sites worldwide, supplying processing, filling, and packaging systems to beverage and liquid food manufacturers.