IIII. J. Hem Maw Transfer. Vol. 32, No. 6, pp. 1141-1153, 1989 Printed in Great Britain 0017-9310/89 S3.00+0.00 0 1989 P~8E,“IOn Press plc Turbofan forced mixer/nozzle temperature and flow field modelling P. KOUTMOS and J. J. McGUIRK Fluids Section, Department of Mechanical Engineering, Imperial College, London SW7 2BX, U.K. (Received 22 September 1988 and in final form 25 November 1988) Abstract-A computational procedure is described for the calculation of the flow and temperature fields in a multilobcd turbofan mixer/nozzle combination. The predictions have been obtained using a finite-volume solution procedure for the steady threedimensional elliptic equations of fluid flow. The procedure allows the calculation of flows within complex geometries using a non-aligned mesh system so that the flow within the lobes themselves may be predicted. Turbulence is modelled using the two-equation k--c eddy viscosity model. Forced mixer performance is shown to be dominated by a periodic array of axial vorticity cells created by the lobe geometry. The calculation of the large secondary velocities associated with this vorticity within and at exit from the lobes provided the necessary boundary conditions at the lobe exit for the predictions of the flow in the mixing duct region. This removes the dependency of previous calculation methods for mixing ducts on measured secondary velocities at the lobe exit plane. The present work demonstrates the capability of the current method to predict the downstream development of the large- scale secondary motions and their strong intluence on the temperature signature at the nozzle exit. Results are also presented to illustrate the ability of the method to predict parameters of importance to mixer designers such as mixer total pressure loss and mixer efficiency. The plausibility of the results obtained illustrates the potential of the method to provide a flexible analysis technique for the complete mixer/nozzle system. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONM INTRODUCTION IN CURRENT and future turbofan engines the use of forced mixers to help mix the hot core and cold bypass streams prior to exhaust through the exit nozzle is likely to be of increasing importance. due to the poten- tial improvements they offer in specific fuel consump- tion, thrust augmentation and jet noise reduction. If the internal cross stream heat transfer in these devices is arranged such that a more uniform total energy profile can be obtained at exit, a small but significant improvement in performance can result with a notice- able reduction in jet noise because of the decrease in maximum nozzle efflux velocities. The level of these gains depends on a balance between the efficiency of temperature mixing achieved and the extra pressure losses incurred in the mixing process. The most popu- lar and successful mixer configuration in use is the lobed mixer (Fig. 1). This increases the interfacial area over which the two streams mix as well as the lateral scales within which the actual mixing takes place (from the incoming boundary layer thickness 6 in undistorted annular mixers to the height of the lobe h >>6, see Werle et al. [l]). Further, due to differing radial deflections, shear between the two streams is also increased and mixing promoted downstream of the lobes. Due to the complexity of the lobe geometry (Fig. 1) the choice of the best configuration depends upon the selection of a large number of parameters. The lobe shape and degree of radial penetration, the plug and tailpipe angle and length, the use of scarfed or scalloped lobes (i.e. cut-outs in the lobe surfaces). Evidently, the traditional approach of identifying an optimum combination of this large number of par- ameters through model scale and then selective full scale testing is both expensive and time consuming. These difficulties have led to several studies into the adequacy and applicability of numerical procedures based on computational fluid dynamics techniques for predicting the flow field and heat transfer behaviour of forced mixers. The majority of earlier efforts have concentrated on calculating the flow in the region between the exit plane of the lobes and the exit plane of the tailplane nozzle, i.e. the mixing duct itself. These computational studies have mainly involved three-dimensional para- bolic versions of the governing equations which require only a single forward marching sweep through the mixing duct downstream of the lobe exit plane. Birch et zyxwvutsrqponmlkjihgfedcbaZYXWVUTS al. [2] and Barton and Birch [3] have used the parabolic analysis of Patankar and Spalding [4], for flows with negligible streamwise diffusion. Kreskovsky et al. [5] and Povinelli and Anderson [6] have calculated mixing duct flows using an alternative method based on the primary/secondary velocity decomposition approach of Briley and McDonald [A. This latter method is more general since it allows the inclusion of a pressure field determined from an a priori potential flow solution within the specified mix- ing duct geometry to introduce some elliptic effects into the streamwise momentum equations. These pre- dictions have by and large been very encouraging 1141