A complete fluid simulation toolkit spanning traditional CFD and deep learning surrogate models — solvers, datasets and cutting-edge papers.
Updated Nov 3, 2026 9 resources
Introduction
Why this collection
Fluid simulation is one of the most active application areas in engineering Physics AI. This collection organises resources from a toolchain perspective:
Deep learning surrogates — replace the classical solver with a neural network for order-of-magnitude speed-ups.
Differentiable physics — frameworks such as JAX-CFD make CFD end-to-end optimisable, closing the design-control loop.
High-quality datasets — AirfRANS and CFDBench provide validation benchmarks.
It suits engineers and researchers who want to understand both the CFD mechanics and the deep learning methods behind them.
Multiphysics benchmark for computational fluid dynamics covering internal/external flows and thermal convection - includes transient and steady-state data.
See LICENSE fileCFDFluidThermal Convection
💬Editor note: A multi-scenario benchmark spanning internal flows, external flows and thermal convection, with both steady and transient data — a solid basis for cross-condition generalisation tests.
Proposes a neural operator-based surrogate model for CFD-level transient thermal-hydraulic simulation of helical coil steam generators in SMRs, enabling real-time digital twin applications.
Google Research's differentiable incompressible-flow solver in JAX; the whole time-stepping loop is differentiable and JIT-compilable on GPU/TPU. A PNAS 2021 learned-interpolation result reached ~80x/40x speedups at equal accuracy, but the project is now unmaintained - treat it as a research prototype and reference.
NewApache-2.0JAXDifferentiable PhysicsCFD
💬Editor note: A fully differentiable CFD framework written in JAX — it drops straight into a scientific-ML workflow.
Fully auto-differentiable CFD solver built on JAX for CPU/GPU/TPU, focused on end-to-end differentiable fluid simulation that couples seamlessly with neural networks.
NewGNU General Public License v3.0JAXDifferentiable CFDDifferentiable Physics
💬Editor note: Another JAX-based differentiable flow-field framework; complementary to JAX-CFD and useful for comparing two API styles.
Open-source CFD solver as a physics-AI data backbone: principles, case structure, getting started, alternatives, cost accounting and failure boundaries.
Open-source CFD and discrete-adjoint solver with built-in aerodynamic shape optimization; PDE-constrained gradients at a cost independent of design-variable count, a high-fidelity reference for surrogates.
NewGNU Lesser General Public License v2.1CFDAdjointShape Optimization
💬Editor note: An open-source cross-platform CFD solver — the natural classical counterpart to the differentiable frameworks above.
High-performance scientific library from Argonne for large linear/nonlinear systems and parallel infrastructure - the underlying solver engine of top-tier CFD and multiphysics simulation, BSD-2-Clause friendly for commercial use.