Atlas Copco Cuts Industrial Pump Design Cycles from Weeks to Hours with SimScale Physics AI
Varisco division replaces slow, license-constrained CFD loops with an autonomous Physics AI workflow, and is now extending the technology across its full pump product portfolio
Atlas Copco’s Varisco division product company, a specialist in heavy-duty industrial liquid pumps, has deployed SimScale Physics AI to reduce the time required to design and optimize a pump impeller from three to five weeks to a matter of days. The engineering AI workflow, which automates design generation, CFD setup, Physics AI inference, and performance reporting, has been validated across ten pump configurations, achieving accuracy within 5% of benchmarks. Atlas Copco is now looking to extend the technology to additional pump product lines within the division.
The Varisco team designs centrifugal pumps for demanding industrial applications; drainage, dewatering, and heavy liquid transfer. Until now, each design iteration required an engineer to build an impeller geometry in CFTurbo, export it to Ansys CFX, wait for simulation results, carry the output back to the design team, and repeat. That loop typically consumed three to five weeks. License availability for Ansys CFX created an additional bottleneck, limiting how many variants the team could explore in parallel.
SimScale worked with the Varisco engineering team to build a solution that connects their existing CAD tools — including CFTurbo and Onshape — with SimScale’s Physics AI and automation infrastructure. The result is a four-stage automated engineering workflow: a configuration module that accepts baseline design files and parameters; an automated CAD preparation and simulation setup stage that generates and meshes variants without manual intervention; a one-click optimization module that runs Physics AI inference across 16 to 128 configurations per suction diameter, falling back to full CFD only when the AI model has insufficient training data; and a reporting stage that outputs pump performance curves, cost predictions, and best-efficiency design files. The Physics AI model — trained on approximately 600 simulations spanning multiple impeller configurations — delivers predictions in approximately five seconds per design point. A 75-iteration optimization loop now runs to completion in under three hours.
Giulio Ferrarese
R&D Manager IFD, Varisco S.r.l., Atlas Copco
“What SimScale’s team delivered for us was not a standard CAE tool — it was a solution built around how our engineers actually work. In the time it used to take us to complete one design iteration, we can now evaluate an entire design space and surface the optimal configuration. We’re already looking at how we extend this across our product lines.”
The deployment at Varisco adds to a growing body of evidence for Physics AI in high-fidelity industrial simulation. Physics AI models accelerate the evaluation of large design spaces by replacing computationally expensive CFD solves with near-instant predictions trained on simulation data, without sacrificing the engineering rigor that industrial applications demand.
Alfred Saad
Chief Customer Officer, SimScale
“What Varisco has built is a template for how Physics AI reshapes industrial design practice. The bottleneck was never the engineer’s ability to generate ideas — it was the time it took simulation to keep pace with those ideas. Physics AI removes that constraint. Varisco can now explore design spaces their competitors cannot afford to explore.”
Atlas Copco’s Varisco team is now working to train the Physics AI model with expanded simulation data to extend coverage further across the design space. The next development phase will extend the workflow to include full volute geometry alongside impeller optimization, and add net positive suction head (NPSH) analysis. Following the success of the Varisco deployment, Atlas Copco is now looking to extend Physics AI-driven design automation more broadly across the Group, bringing the same approach to additional product lines and engineering teams.
About SimScale
SimScale is the AI-native cloud platform for engineering simulation. Trusted by more than 900,000 users, SimScale empowers engineers everywhere to innovate faster by exploring 1000’s of engineering decisions in seconds. By integrating Engineering AI agents and Physics AI models with computational fluid dynamics (CFD), finite element analysis (FEA), electromagnetic, and thermal simulation in a single cloud platform, SimScale empowers teams to engineer the irreplaceable. For more information, visit www.simscale.com.
About Atlas Copco
Atlas Copco Group enables technology that transforms the future. Founded in 1873 and headquartered in Stockholm, Sweden, the Group innovates to develop products, services and solutions that are key to its customers’ success. Its four business areas offer compressed air and vacuum solutions, energy solutions, dewatering and industrial pumps, industrial power tools, and assembly and machine vision solutions, across a portfolio of more than 80 brands. For more information, visit www.atlascopcogroup.com.
