Transport modeling of simple fluids and nano-colloids : thermal conduction mechanisms and coarse projection
Author(s)Eapen, Jacob, 1968-
Massachusetts Institute of Technology. Dept. of Nuclear Science and Engineering.
MetadataShow full item record
In the first part of this thesis, the modes of microscopic energy fluctuations governing heat flow in nano-colloids are quantitatively assessed by combining linear response theory with molecular dynamics (MD) simulations. The intrinsic thermal conductivity is decomposed into self and cross correlations of the three modes that make up the microscopic heat flux vector, namely, the kinetic, the potential and the virial. By this decomposition analysis, the interplay between the molecular mechanisms that govern the variation of the thermal conductivity with volume fraction and solid-fluid interaction is examined. For a specific system of nanosized platinum clusters which interact strongly with host liquid xenon, a significant thermal conductivity enhancement is obtained as a result of self correlation in the potential energy flux. The effect saturates at higher volume fractions due to the cross-mode correlation between the potential and the virial flux. A strong solid-fluid coupling also introduces an amorphous-like structural transition and a pronounced cage effect that significantly reduces the self diffusion of the nano-clusters. These attendant structural and diffusive effects, unlike the self correlation of the potential flux, are amenable to experimental observations. The cluster-fluid interface is characterized by large fluctuations in the potential energy which is indicative of an unusual exchange of potential energy among the interfacial fluid atoms. For small nano-clusters, the interfacial layers interact with each other to form a percolating network. The research findings highlight the importance of surface interactions and show that the interfacial thermal resistance emanating from the self correlation of the collision flux is not the limiting mechanism for heat transfer in nano-colloids.(cont.) This thesis also addresses several theoretical concerns regarding the microscopic thermal transport in colloids by using non-equilibrium molecular dynamics simulations (NEMD). The time averaged microscopic heat flux which assumes spatial homogeneity is shown to be applicable to nano-colloidal systems. Further, it is demonstrated that the thermal conductivity from a NEMD simulation is statistically equivalent to that of an equilibrium linear response evaluation only under certain dynamic conditions at the cluster-fluid interface. The concept of interfacial dynamical similarity is developed to establish this equivalence. The proposed thermal conduction model is consistent with several experimental observations such as the anomalous enhancement at small volume fractions with very small nanoparticles (3-10nm), limiting behavior at higher volume fractions, and the lack of correlation of the enhancement to the intrinsic thermal conductivity of the nano-clusters. The model also suggests possible avenues for optimizing the colloids by developing nano-clusters that have functionalized surface layers to maximize the interactions with the fluid atoms. In the second part of this thesis, smooth field estimators based on statistical inference and smoothing kernels are developed to transfer molecular data to the continuum for hybrid and equation-free multiscale simulations. The field estimators are then employed to implement coarse projection, a multiscale integration scheme, for a shear driven flow in an enclosure. This thesis shows that the spatial continuity and smoothness of the microscopically generated coarse variables, geometrically similar initial conditions and the separation of timescales are essential for the correct coarse field evolution with coarse projection.
Thesis (Sc. D.)--Massachusetts Institute of Technology, Dept. of Nuclear Science and Engineering, 2006.Includes bibliographical references (p. 159-166).
DepartmentMassachusetts Institute of Technology. Dept. of Nuclear Science and Engineering.
Massachusetts Institute of Technology
Nuclear Science and Engineering.