Fluid Flow and Heat Transfer Numerical Prediction of Cross Flow Heat Exchanger

Abstract

Heat exchangers design is a major challenge problem in the process industry today. Heat exchanger applies to all equipment used to transfer thermal energy between two streams. Namely equipments in which two process streams exchange heat with each other. One of these equipments -cross flow heat exchanger-, used in numerous industry processes, uses as heat exchanger a normal flow to arrays of circular tubes arranged in staggered or in-line formation. These configurations, for instance, are typical in heat exchanger designs used in fossil-fuel and nuclear power plants and, indeed, over many other sectors of thermal and process engineering. Cross flow heat exchangers are designed based on experimental correlations. Depending on which correlation is used -and on the Reynolds number- , however, the heat exchanger area can vary from 30 to 50 percentage. According to the turbulence modeling technique and other details of the simulation, numerical prediction can be a useful tool for improving designs accuracy, in addition to gives more information about the physics of temperature and velocity field. This study presents the computational modeling of flow and thermal field in a cross flow heat exchanger, with staggered formation, and longitudinal and lateral pitch equal to 2, for Reynolds number about 40,000 and Prandtl number equal to 0,744. The simulation uses a non-periodic computational domain in the longitudinal direction with periodic domains in transverse and along the tube directions. Furthermore, a second computational domain periodic in x,y,z, is used for comparison. Four turbulence modeling techniques are examined: Reynolds-averaged Navier-Stokes (RANS), unsteady Reynolds-averaged NavierStokes(URANS), Detached Eddy Simulation(DES), and a very Large Eddy Simulation(vLES) -in addition to simulations without any turbulence model(WTM)-. On the other hand, RANS and URANS techniques are examined in 2D and 3D. Results from this study show that 3D RANS is superior to 2D RANS, that vLES is the best simulation from a physical point of view, and DES presents the best agreement with experimental data.