其他摘要:In previous papers (Idelsohn et al., 2012) the numerical method called Particle Finite Element Method - Second Generation (PFEM-2) was presented. It solves scalar and vectorial transport phenomena such as Navier Stokes equations for transient, laminar and incompressible flows. This method is characterized by a Lagrangian formulation that uses both particles and mesh to solve physics equations. Taking advantage of that, particles can transport physics variables that any model includes, this paper presents an evolution of the method for solving equations that govern the fluid flow coupled with passive-scalar transport equations, such as species concentrations or temperature. Also, the explicit streamline integration strategy used by the method allows to update each variable following its own physic law in an efficient way. This paper places particular emphasis in the thermal coupling through a buoyancy term added to fluid acceleration and transporting at each particle its temperature. Regarding to the implementation, it follows the parallel computing strategy presented above. Finally, results obtained in typical benchmark cases, such as thermal square and cubic cavity, are presented, comparing with bibliographic and experimental data.