Similarity states of passive scalar transport in isotropic turbulence

Abstract

By simple analytical and large‐eddy simulations, the time evolution of the kinetic energy and scalar variance in decaying isotropic turbulence transporting passive scalars are determined. The evolution of a passive scalar field with and without a uniform mean gradient is considered. First, similarity states of the flow during the final period of decay are discussed. Exact analytical solutions may be obtained, and these depend only on the form of the energy and scalar‐variance spectra at low wave numbers, and the molecular transport coefficients. The solutions for a passive scalar field with mean‐scalar gradient are of special interest, and we find that the scalar variance may grow or decay asymptotically in the final period, depending on the initial velocity distribution. Second, similarity states of the flow at high Reynolds and Péclet numbers are considered. Here it is assumed that the solutions also depend on the low‐wave‐number spectral coefficients, but not on the molecular transport coefficients. This results in a nonlinear dependence of the kinetic energy and scalar variance on the spectral coefficients, in contrast to the final period results. The analytical results obtained may be exact when the similarity solutions depend only on spectral coefficients that are time invariant. The present analysis also leads directly to a similarity state for a passive scalar field with uniform mean scalar gradient. Last, large‐eddy simulations of the flow field are performed to test the theoretical results. Asymptotic similarity states at large times in the simulations are obtained and found to be in good agreement with predictions of the analysis. Several dimensionless quantities are also determined, which compare favorably to earlier experimental results. An argument for the inertial subrange scaling of the scalar‐flux spectrum is presented, which yields a spectrum proportional to the scalar gradient and decaying as k −7/3. This result is partially supported by the small‐scale statistics of the large‐eddy simulations.