Neural turbulence closures improve solution stability and uniqueness
Well-posedness of neural turbulence closures and tangent dissipation
Machine Learning
Summary
Simulating turbulent flows is hard because approximations called closures can cause unpredictable solutions. This paper shows that by designing these closures with certain mathematical properties, you can ensure stable, unique solutions that match real flow behavior better. The authors tested their method on flow through a channel and found that their approach reliably found a single solution while other methods sometimes found many. This work supports more robust turbulence simulations using neural networks.
What this means in practice
- •For computational fluid dynamics engineers: Use neural network closures with enforced tangent dissipation to improve reliability and accuracy in turbulence simulations for engineering designs.
- •For aerospace simulation teams: Integrate dissipation-constrained neural turbulence models to reduce sensitivity and improve uniqueness in high-Reynolds-number flow simulations.
Authors
Zhen Zhang, George Em Karniadakis
Abstract
A neural turbulence closure defines a new boundary-value problem, $R(U)=N(U)+F(U)=0$, with a coupled Jacobian $J(U)=N'(U)+F'(U)$, where $N$ is the original mean-flow operator and $F$ the learned closure. We establish two consequences of global tangent dissipation. For a monotone original operator, a positive uniform margin supplied by the original operator and closure together guarantees existence, uniqueness and a global inverse-sensitivity bound relating a posteriori solution error to the a priori residual. For a general original operator, a dissipative closure cannot worsen tangent dissipation, but this alone does not guarantee uniqueness. Tangent dissipation depends on both diffusion and reaction. We study two complementary ways to promote it: (1) an exact-integral construction enforcing non-negative tangent diffusion while leaving reaction unconstrained, and (2) a penalty on tangent-reaction violations at sampled states. Tangent diffusion enters the Jacobian, and non-negative secant eddy viscosity alone does not control its coercivity. We conduct tests with channel flow at $Re_τ=180$--$5200$, which provides a strongly monotone baseline. Both constrained closures reach accurate solutions in all 50 training-seed/Reynolds-number cases. At $Re_τ=1000$, we conduct tests with 10,000 starts for one fixed network per closure and we find one root for each constrained closure and multiple roots for the other closures. Although this does not prove uniqueness, it provides strong empirical evidence for uniqueness of the tested constrained closures. At $Re_τ=5200$, the construction and penalty reduce the reported inverse sensitivity relative to the original operator by approximately $372\times$ and $11\times$, respectively.