Abstract
Natural convection exhibits turbulent flows which are difficult or impossible to resolve in direct numerical simulations. In this work we investigate a quasilinear form of the Rayleigh-Bénard problem which describes the bulk one-dimensional properties of convection without resolving the turbulent dynamics. We represent perturbations away from the mean using a sum of marginally stable eigenmodes. By constraining the perturbation amplitudes, the marginal stability criterion allows us to evolve the background temperature profile under the influence of multiple eigenmodes representing flows at different length scales. We find the quasilinear system evolves to an equilibrium state where advective and diffusive fluxes sum to a constant. These marginally stable thermal equilibria (MSTE) are exact solutions of the quasilinear equations. The mean MSTE temperature profiles have thinner boundary layers and larger Nusselt numbers than thermally equilibrated two- and three-dimensional simulations of the full nonlinear equations. MSTE solutions exhibit a classic boundary-layer scaling of the Nusselt number Nu with the Rayleigh number Ra of Nu∼Ra1/3. When MSTE are used as initial conditions for a two-dimensional simulation, we find that Nu quickly equilibrates without the burst of turbulence often induced by purely conductive initial conditions, but we also find that the kinetic energy is too large and viscously attenuates on a long, viscous timescale.
Original language | English (US) |
---|---|
Article number | 093501 |
Journal | Physical Review Fluids |
Volume | 6 |
Issue number | 9 |
DOIs | |
State | Published - Sep 2021 |
Funding
We would like to thank C. Doering, who greatly influenced the way many of us think about Rayleigh-Bénard convection. Some of the ideas in this work came from discussions with Charlie at Walsh Cottage during the WHOI GFD summer programs; it is difficult to imagine Walsh Cottage without Charlie's friendly and open scientific style, and enthusiasm for softball. The authors thank G. Vasil, G. Chini, K. Augustson, and E. Kaufman for their valuable feedback and suggestions. We would also like to thank C. Doering and B. Wen for the tabulated DNS and ECS data for Fig. . We thank the dedalus and eigentools development teams. Computations were conducted with support by the NASA High End Computing (HEC) Program through the NASA Advanced Supercomputing (NAS) Division at Ames Research Center on Pleiades with allocation GIDs s2276. L. O'Connor is supported by Northwestern University's Walter P. Murphy Fellowship.
ASJC Scopus subject areas
- Computational Mechanics
- Modeling and Simulation
- Fluid Flow and Transfer Processes