This paper presents both experimental and unsteady RANS investigations of a slot-type casing treatment at a transonic axial flow compressor rotor. Experimental results show that at 60% and 98% of rotor design wheel speeds, approximately 100% and 200% extra extensions of the rotor operation ranges are achieved, respectively. On the other hand, there are about 3.6% and 2.0% drops of efficiencies at 60% and 98% speeds respectively if comparisons are made at the same peak-efficiency mass flow rates of the solid casing case. If comparing the respective peak efficiencies for the solid casing case with those for the treated casing case, there are still about 3.4% and 0.7% drops at 60% and 98% speeds, respectively. As for the unsteady RANS study, an in-house unsteady RANS code has been used to study the casing treatment flow at several operating points, i.e., the peak efficiency and the near stall with regard to the solid casing case at 60% speed and 98% speed, respectively. It is shown that the interactions between the blade passage flow and the casing treatment flow exhibit different manner at two rotating speeds. The flow condition in which the rotor operates, i.e., either the subsonic condition at the 60% speed or the transonic condition with passage shock presented at the 98% speed, is one of the determinate factors that are responsible for the manner the casing treatment works. The loss production due to casing treatment is also particularly discussed.

A CFD code for three-dimensional viscous flows, in particular for those in turbomachinery, has been developed based on Favre-averaged compressible Navier-Stokes equations and one-equation Spalart-Allmaras turbulence closure. The model equation of Spalart-Allmaras turbulence closure is converted into conservative form and discretized in the same manner as that for mean flow equations. A two-dimensional transonic diffuser flow and a two-dimensional transonic nozzle flow which feature pressure-gradient induced separation and shock wave/boundary layer interaction respectively are used to validate the code and application of the Spalart-Allmaras model (hereafter the S-A model) in internal flows. It is shown that the S-A model can give fairly good results compared to the experimental data. Some modifications of model equation are introduced for improving the grid insensitivity of the turbulence model. To validate the applicability of the code to the complex flows in transonic turbomachines, flows through two transonic compressor rotors, NASA Rotors 67 and 37 are calculated, and numerical results are compared with the well documented experimental data. The calculated results agree reasonably well with the experiments, and as expected, the S-A model, which is primarily developed for external flows, can also be effectively applied to internal flows. Discrepancies between the experimental data and calculations and the possible causes are also discussed.

This paper describes a procedure for a rapid and accurate 3D aerodynamic optimization of high performance turbine blades. This procedure has been developed to account for the complicated geometrical aspects and the complex nature of the associated fluid flow, while remaining simple, practical and demanding less computing power. The focus has been placed on the blade geometrical representation using a set of simple algebraic equations (blade parameterization) and on the aerodynamic optimization methodology based on the numerical computations by a N.S. solver. The turbine blade, including thickness distribution and camber line for each section of the blade span and radial stacking line, has been defined by polynomials, allowing investigation of the influence of any single-parameter change on blade performance. An improved response surface method, by incorporating a simulated annealing algorithm (RS-SAM), has been found to improve the accuracy and to strengthen the optimum-searching ability. A multi-objective response surface method (MORSM) has also been included for testing. One example is given here to demonstrate the effectiveness of the procedure.

In order to predict more efficiently the flow in compressor with casing treatment in a integrated manner as well as to facilitate an integrated tool for the design optimization of casing treatment, a mathematical-physical quasi-steady model for the slot-type casing treatment flows were developed based on the understanding of the unsteady flow physics of the rotor-casing treatment combination. The new model has been coded up and validated against a slot-type casing treatment which is examined by experiment and unsteady RANS calculations in the companion paper Part I. The calculations using this model show an encouraging agreement with those obtained from measurements and unsteady RANS calculations, indicating the essence of the flow physics has been captured by the model. Although the model is based on the slot type casing treatment, the principle behind it is rather wide-catching and extensions to other types of casing/hub treatment can also be realized. Since the model equations for the casing treatment flow are solved in the same way as that for conventional steady flows, considerable amount of computer resources can be saved. This enables the model applicable in the current steady RANS-based design system, and overall assessments of a compressor with an applied casing treatment are readily achievable.

The Scale Adaptive Simulation (SAS) models can keep standard RANS capabilities in stable flow regions but resolve turbulent structures in unsteady regions of flow field like LES. This RANS/LES property of SAS model relies on the v. Karman length-scale as a scale determining variable, which allows the model to automatically adapt to the appropriate length-scales in the simulated flows. Although SAS is young and still in developing, it has been proved to be very suitable for the predictions of massive separation flows. In the current study, the SST-SAS model is implemented in an in-house CFD code. First, the test case of decaying homogenous isotropic turbulence is selected for calibrating a constant associated with high wave number damping. Then, the simulation of flow past NACA 0021 airfoil at 60° attack angle is carried out for the validation of this turbulence model in our code. After that, the numerical results of NACA 0021 airfoil at a range of attack angles after stall are also presented for the comprehensive understanding of the SAS model.