Dissipative Particle Dynamics Research On Complex Heat Convection

Dissipative particle dynamics (DPD) is a particle-based method for the complex flowsimulation in meso-scale. As a coarse-grained version of molecular dynamics (MD)，thecomputational efficiency of DPD is much higher than MD as each DPD particle represents agroup of actual molecules. One of the drawbacks of the classic DPD is that the total energy ofthe system is not conserved in the interaction between particles. It has been remediedindependently by introducing an internal energy variable and a temperature for each particle inthe DPD system. Such a DPD model with energy conservation is known as eDPD in theliteratures.The application of eDPD is still limited. As a powerful mesoscopic simulation method，improvement is needed to extend the application of eDPD to the complex convective heattransfer problems，which is the core content of this research. The meaning of complex refersto three kinds of complex cases，namely，the complex fluids，the complex flow and thecomplex computational domain. The DPD method is an efficient tool for the complex flowsimulation，but at the present，it is not so effectual for all these three cases. The DPD model hasnatural advantage for complex fluids. However，for the complex flow and the complexcomputational domain，more works are required to improve the versatility of the DPD method.The present work is to extend the eDPD to model fluid flow and heat transfer inenclosures with complex boundaries. We propose a numerical strategy for dealing withirregular geometries in DPD system and by which the application of DPD (or any other particlesimulation method) can be extended to mimic hydrodynamics in arbitrarily complexgeometries like ones with moving surface or free surface which cannot be defined bymathematical functions. Then,we take horizontal eccentric annulus as the flowing zone toapply eDPD to the simulation of force，natural and mixed convection.We also take anarbitrarily complex doubly connected region with large structural motions as the computationaldomain，and apply the eDPD method to simulate natural and mixed convection problems. Atlast，we perform a numerical analysis of the Coulomb-driven convection induced by a strongunipolar injection and its effect in augmenting nature convection in a dielectric liquid-filledconcentric annulus.