envelio has developed a GPU-based power flow solver aimed at speeding up distribution-grid time-series simulation work that utilities use for planning and interconnection decisions. The company says the solver cuts annual time-series simulations from days or weeks to less than 30 seconds, and reports performance of up to 20,000x faster than processes relying on conventional desktop grid calculation tools.
In engineering terms, the solver targets the compute bottleneck in running large numbers of power flow calculations across many time steps and scenarios. envelio says the GPU architecture enables massive parallelization, computing across scenarios, time steps, and grid segments simultaneously. The company also positions the approach as “physics-first,” using deterministic, physics-based models rather than AI-based estimation, and it specifically references a “true AC power flow model” intended to capture voltage behavior and reactive power effects in distribution grids.
For data center operators, “speed to power” often hinges on how quickly a utility can evaluate interconnection requests and system impacts under realistic time-series conditions. Faster, higher-volume simulation doesn’t automatically create capacity, but it can change the pace and granularity of studies, especially where hosting capacity is constrained by time-varying thermal or voltage limits rather than a single worst-case snapshot.
The solver is part of envelio’s Intelligent Grid Platform (IGP), which includes a digital twin of the power grid. envelio says the GPU-based solver enhances the IGP’s existing Grid Hub functionality, has been integrated into select time-series and hosting capacity workflows, and is currently available, with expansion to additional use cases planned throughout the year.
“Our new solver reduces annual time-series simulations to less than 30 seconds, achieving performance up to 20,000 times faster than processes relying on conventional desktop solvers,” said Luigi Montana, CEO of envelio. CTO Dr. Fabian Potratz added that the company “opted to still accurately solve the real physics in the grid—but with a new technology that is at least as fast as, if not faster than, AI-based estimation approaches out there.”
Source: envelio
