Computation of thermal conductivity in nanofluids

Abstract

Molecular dynamics simulation is a popular computational technique that is widely used to simulate and investigate thermophysical properties of nanofluids. Molecular dynamics is frequently used in chemical physics, materials science, and molecular-scale modeling because it offers an atomistic level understanding of the equation of transport coefficient. The most common method is with the use of the Green-Kubo relation, with equilibrium molecular dynamics. The other well known method is by non-equilibrium simulations by either imposing a gradient of temperature by modifying the boundaries or by the use of the so-called thermal force. Experiments suggest nanofluids have enhanced heat transfer coefficient compared to the base fluid. That make nanofluids beneficial in a wide range of industrial and home heat transfer applications, including as engine cooling and vehicle thermal management, fuel cells, microelectronics, heat exchangers in boilers and refrigerators, and solar water heaters. In this thesis we evalute the thermal conductivity of a nanofluid with different molecular dynamics methods. The simulations run under the NVE conditions, for which the total atom number, total volume, and total energy are constant. Using the cell list approach, we create neighbor lists to increase computing efficiency. The Velocity Verlet technique is then used to integrate the equations of motion with a time step of 0.001. Since they are both conserved quantities, the total energy E and the total momentum P are computed to test the code. Throughout all of the simulations, the temperature and density of box are taken to be 0.722 and 0.8442 in dimensionless units, respectively. For the fluid-fluid, fluid-nanoparticle, and nanoparticle-nanoparticle interactions the Lennard-Jones potential is used while for the interatomic interactions in the nanoparticles we used the quantum corrected Sutton-Chen (Q-SC) many body potential. The volume fraction of the nanoparticles is varied from 0.3% to 9%. We used two different reverse non-equilibrium molecular dynamics (rNEMD) methods to evaluate the thermal conductivity of the nanofluid. Namely, the velocity exchange method and dual thermostat method. In the velocity exchange method the total energy and linear momentum are conserved while for the dual thermostat method the total energy and linear momentum are not conserved. The difference between rNEMD methods and the traditional non-equilibrium (NEMD) methods is that a heat flux is imposed on the system and then the temperature gradient is measured. This has two main advantages, first of all the heat flux is not well defined for systems with many-body interactions, secondly when one imposes a temperature gradient by modifying the periodic boundaries, this results in surface effects. However, for the rNEMD simulations the heat flux is imposed by unphysical means, and one must be careful in the interpretation of the results. The velocity exchange method and dual thermostat method results show that for a volume fraction of 0.3% of nanoparticles the thermal conductivity slightly increased while for larger volume fractions the thermal conductivity decreases.