loading
loading
Vivix is a Brazilian flat glass manufacturer whose products include glass and mirrors for the construction and furniture markets. In float-glass production, the furnace is one of the most critical assets — and one of the hardest to observe. At operating temperatures of approximately 1,550°C to 1,600°C in the furnace, teams have limited direct visibility into the interactions among combustion gases, molten glass and the furnace’s refractory walls. Understanding those conditions while production continues is essential to making informed decisions about energy use, asset life and glass quality. To address that visibility challenge, Vivix is developing an executable digital twin (xDT) of its float-glass furnace in collaboration with Siemens, Snowflake and NVIDIA. The initiative brings together engineering simulations, observed furnace conditions and AI models to help teams estimate conditions they cannot measure directly and evaluate potential operating changes. Vivix is focusing initially on three connected areas: refractory health, combustion efficiency and glass-melting quality. By connecting physical shop-floor dynamics with cloud-scale AI analytics, Vivix brings physics and AI together to model, predict and optimize system behavior in near real time. This helps lower energy consumption, extend furnace lifespan and improve product quality. Traditional physics simulations provide high-fidelity predictions but are computationally intensive, whereas standard AI models often lack an understanding of physics. The xDT architecture overcomes this trade-off by taking high-fidelity Siemens Simcenter multiphysics models, converting them via Simcenter reduced order modeling into lightweight, portable executables, and deploying them natively within Snowflake's AI Data Cloud. To achieve a high level of scalability, the Simcenter Executable Digital Twin Manager allows the management of the xDT lifecycle in a single solution. Snowflake’s role is to make the resulting engineering insights useful alongside Vivix’s operational data. Vivix brings IT and operational technology data, including physical sensor readings, together in a unified namespace in Snowflake. This gives engineering and operations teams a governed view of observed furnace conditions that they can use alongside predictions from the digital twin, rather than reviewing simulation results in isolation. This enables operations teams to access engineering-grade data in near real time. NVIDIA PhysicsNeMo enables Vivix to build AI surrogate models that combine physics-driven causality with simulation and observed data, enabling real-time predictions. Engineering and operations teams can collaboratively review high-fidelity results with technologies built using NVIDIA Omniverse libraries. Deployed on Amazon Web Services (AWS) as the underlying cloud service provider, this hybrid solution functions as a continuous "virtual sensor." It allows teams to reconstruct complex and unmeasurable internal states in near real time, including volumetric heat distribution, gas velocity fields and thermal fluxes across refractory walls without needing constant connection to heavy simulation environments. Across all furnace operations, this integrated approach unlocks targeted optimizations in three core domains: - Refractory preservation and longevity: By fusing real-time thermocouple streams with refractory thermal profiling, the twin calculates localized wear rates and detects emerging hotspots hours in advance. This protects critical assets, minimizes thermal stress and extends furnace campaign life before requiring cold repairs. - Combustion and energy efficiency: The airflow and thermal models dynamically evaluate burner power levels and heat distribution patterns. This enables precise setpoint optimization to stabilize combustion, lower natural gas consumption and reduce overall emissions. - Glass melting and quality optimization: Modeling thermal gradients and fluid flow fields within the molten glass bath provides visibility into residence time distribution and melting completeness. This helps improve glass homogeneity and reduce defect rates in finished flat glass. Based on a pilot project, Vivix expects to cut furnace energy consumption by 3%-5%, extend furnace lifespan by one year and shorten its data collection and analytics cycle by 80%. Figure 1. AI physics predictions for glass melting and refractory wall conjugate heat transfer (model trained with NVIDIA PhysicsNeMo and visualized using NVIDIA Omniverse libraries). Solution architecture Figure 2. The solution architecture for the float furnace digital twin, developed by Snowflake and Siemens DISW. By combining physics-based modeling (Siemens SimCenter) with AI-driven analytics (Snowflake Cortex AI), this hybrid model can predict furnace behavior under new conditions and run virtual "what-if" simulations. To make these insights immediately actionable on the shop floor, a Snowflake AI multiagent interface allows operators to interact with the digital twin using natural language. Engineers can ask direct questions regarding optimal air velocities, evaluate "what if" operational scenarios and receive root-cause diagnostics without risking the physical asset. Grounding AI in first-principles physics on top of Siemens, Snowflake, NVIDIA and AWS infrastructure gives Vivix a safer, highly predictive and sustainable foundation for glass manufacturing excellence by performing what-if simulations with the virtual twin instead of putting physical assets in unknown and potentially unsafe operating conditions. Explore what’s possible: Learn how Snowflake helps manufacturers bring operational and engineering data together to support analytics and AI.
document fetched 2026-10-28 19:02 · canonical https://snowflake.com/en/blog/vivix-executable-digital-twin-float-glass