Network modelling of transient heat exchanger performance

Abstract

This study investigates the applicability of the thermal-fluid network approach to the modelling of transient heat exchanger performance. Two different solution algorithms, namely the Implicit Pressure Correction Method (IPCM) and the Runge Kutta method with Trapezoidal Damping (RKTD) for the solution of the one-dimensional governing equations in thermal-fluid network problems are presented. The advantages and disadvantages of two types of numerical discretisation schemes used in thermal-fluid network problems are discussed and the discretised one-dimensional governing equations for the staggered grid discretisation scheme used in the IPCM and RKTD method is presented. The RKTD method is used as a time integration scheme for the generalised thermal-fluid network solver Xnet. Several test cases are introduced and the basic primitive elements available in Xnet are compared to the commercial thermal-fluid network code, Flownex (which uses the IPCM), for both steady-state and transient conditions. Two different network topologies are introduced for the discretisation of heat exchangers when a network approach is followed and the thermal-fluid network solver Xnet is applied to a basic parallel and counter flow configured pipe-in-pipe type heat exchanger to investigate the effect on the type of discretisation scheme used. The results obtained are compared to primitive element models in Flownex as well as the composite RX element in Flownex. The extent to which thermal-fluid network solvers are able to predict transient heat exchanger performance are further investigated by modelling a complex shell-and-tube heat exchanger using Xnet and comparing the steady-state and transient results to both a primitive element model in Flownex as well as the composite STX element in Flownex. This contributes to the validation of Flownex’s heat exchanger models by using a different approach than Flownex. The results showed that the explicit method used in Xnet is capable of solving large arbitrary structured thermal-fluid networks with a high level of accuracy. The result of Flownex compares very well with that of Xnet, which proves (verifies) that the solution algorithm is correctly implemented in both codes. Even though the explicit thermal-fluid network code, Xnet, can accurately predict fast transients, a drawback of this method is the large computational time required to simulate transient heat exchangers with large thermal masses.