Proceedings of the Institution of Mechanical Engineers,Part A: Journal of Power and Energy

Calculations of unsteady turbomachinery flows using conventional time domain methods typically would lead to the use of multiple passages/whole annulus owing to the required direct periodic condition. This makes conventional unsteady solutions extremely time consuming and is one of the major difficulties in applying unsteady calculations in a blading design environment. A single-passage approach to three-dimensional unsteady Navier-Stokes calculations using the Fourier series based shape correction method has been developed. The shape correction method gives a single-passage solution to unsteady flows in blade rows under multiple disturbances with arbitrary interblade phase angles. Therefore the computing time can be dramatically reduced. A case study for a transonic fan rotor subject to long-scale circumferential inlet distortions shows that a factor of 5 speed-up could be achieved. Detailed comparisons in both unsteady and time-averaged flow results between the single-passage and multiple-passage solutions demonstrate the validity and effectiveness of the present single-passage approach. A further calculation of a transonic rotor simultaneously under the influences of both upstream and downstream blade rows also shows the capability of the shape correction method in dealing with multiple disturbances.

The three-dimensional flow field of a transonic high-pressure turbine stage has been studied experimentally in the von Karman Institute short duration compression tube facility. The test programme encompasses two Reynolds numbers (Re_2,c = 1.0 × 10⁶, 1.26 × 10⁶) and three different pressure ratios (P_o1/P_s3 = 2.3, 3.1, 4.1). The steady flow angle, pressure and temperature measurements rely on five-hole pneumatic probes and micro-thermocouples. Fast response pressure and cold wire resistance thermometer probes measure the unsteady pressure and temperature. The experimental time-averaged spanwise distributions are compared with steady state three-dimensional Navier-Stokes stage simulations; zones of secondary flows are identified in the spanwise distributions. The stage exit pitchwise traverses show that the stator exit flow field affects the stage exit flow. Downstream of the stator vane, large time-resolved static pressure fluctuations are identified, attributed to the variation of the origin, the incidence and strength of the stator trailing edge shock. At the stage outlet, large total temperature fluctuations appear owing to the periodic passage of the tip leakage flow. For the high-pressure ratio, the rotor is choked and large pressure and temperature fluctuations are recorded owing to the rotating rotor trailing edge shock system. Zones of high mixing (secondary flows) tend to attenuate these fluctuations.

This paper describes the effect of both an upstream and a downstream vane on the time-resolved surface pressure field around a high-pressure rotor blade. The geometry of the downstream vane considered is a large low aspect ratio vane located in a swan-necked diffuser duct, similar to those likely to be used in future engine designs. Two test geometries are considered: firstly, a high-pressure turbine stage coupled with a swan-necked diffuser exit duct; secondly, the same high-pressure stage but with a vane located in the downstream duct. Both tests were conducted at engine-representative Mach and Reynolds numbers, and experimental data were acquired using fast response pressure transducers mounted on the mid-height streamline of the high-pressure rotor blades. It is shown that the potential field of the downstream vane causes a relatively large static pressure fluctuation on the late rotor suction surface but that the early suction surface and pressure surface are unaffected by the presence of the vane. In addition to the two main interaction mechanisms (upstream vane-rotor and downstream vane-rotor interactions) a third interaction is identified. This is shown to be caused by the interaction, within the rotor passage, between the flowfield associated with the upstream vane and the potential field associated with the downstream vane. This new interaction mechanism is shown to cause static pressure fluctuations on the late rotor suction surface at a frequency corresponding to the difference in the numbers of upstream and downstream vanes.

A time-dependent numerical simulation of a single-stage transonic turbine is used as a direct simulation of stator-rotor interactions. The different parts of the deterministic correlations are computed and compared. The present work aims to provide insight into the influence of these correlations on the steady flow field of the average-passage equation system. This is related to the deterministic models of the literature. Several steady computations are performed with the various correlations. These results are compared with the time-averaged flow field. The impact of boundary conditions is also studied to highlight the importance of the non-uniform circumferential distribution at the so-called ‘mixing plane’. This work establishes that a non-uniform circumferential distribution is a key to ensure the success of steady computation. Furthermore, it demonstrates that full deterministic correlations are necessary to reproduce the time-averaged flow field. Spatial correlations are thus insufficient to model the stator wakes that interact with the rotor. Steady computations using only spatial correlations give less accurate results than those obtained without any correlation. This is a serious flaw of most of the literature's models.

A transonic axial turbine stage is computed by solving the time-averaged Navier-Stokes equations in the turbulent flow regime. The stator and rotor rows are interfaced using a mixing plane. The sensitivity of the stator-rotor coupling to the boundary conditions and mixing strategies is investigated. The mass or area tangential averaging is found to have slight impact on the stator-rotor match. The relative and absolute formulation of the rotor inlet boundary condition is found to alter the match of the two rows. The computations indicate that both relative and absolute boundary conditions should be used to estimate and control part of the uncertainty in the stator-rotor match, which can be greatly reduced by an iterative procedure, based on the stator exit pressure. The area averaging allows a unique matching point to be estimated for both absolute and relative properties.

An unstructured time-accurate simulation is proposed and discussed for the inviscid stator-rotor interaction of a three-dimensional turbine stage. The solver, based on an upwind total variation diminishing finite volume scheme, is briefly described. To allow a time-accurate simulation for the unsteady flow behaviour, two possible approaches are considered. The first proposal consists of a classical explicit multistep Runge-Kutta scheme. The second approach is based on a dual time-stepping strategy, which exploits the implicit time-marching Newton-Krylov method for the inner iteration. In this case the linear solver of the implicit method is the generalized minimal residual method with right preconditioning obtained by an ILU(0) incomplete factorization. The multiblock parallel version of the solver developed for stage investigations is applied to the turbine annular stage of the VKI Institute, developed and tested under the framework of the BRITE EURAM III Project TATEF. Particular concern is devoted to the development of the numerical strategy and validation of the new approach. Memory and central processing unit time costs are both addressed to verify the performances of the numerical schemes proposed.

A design method for profiling the end wall to reduce secondary flow has been reported previously. A profile has been tested in the Durham Linear Cascade and the results confirmed the design method. This paper describes the design and testing of a second-generation end wall, where the profiling is more suited to a real turbine. The new end wall has been tested in the linear cascade and a comprehensive set of measurements have been taken. These include traverses of the flow field upstream and downstream of the blade row, surface static pressure distributions on the end wall and flow visualization. Comparisons have been made with the results with a planar end wall and the earlier profiled end wall. Observed reductions in exit angle deviations are even greater than for the first design, although the loss reduction is not as great. The results verify the design, confirming profiled end walls as a means of reducing secondary flow, kinetic energy and loss. Overall an improved understanding of the effects of end wall profiling has been obtained although further work is required in this area.

Turbine aerothermal analysis is a key issue of the entire engine design as higher performance and longer component life must always be reached. For this purpose, a substantial effort is dedicated to aerothermal methods at Snecma-Moteur. This paper deals with the problem of heat transfer prediction on uncooled turbine blades. The main parameter investigated is the transition process, but consideration is also given to the turbulence model, effect of free stream turbulence on laminar boundary layer and mesh refinement. All the models are implemented in a three-dimensional Navier-Stokes code. Quasi-three-dimensional computations are used to evaluate all models, and then fully three-dimensional computations are performed. Aerothermal predictions are compared with static pressure and heat transfer measurements from the IACA and TATEF Brite-EuRam programmes carried out in the test facilities of DERA, Pyestock and the von Karman Institute.

A study of flow in a stationary model of a two-pass internal coolant passage is presented which focuses on the flow characteristics in the 180° bend region and downstream of the bend, where the flow is re-developing. A stereoscopic digital particle image velocimetry system measured all three velocity components simultaneously to obtain mean velocity and turbulence quantities of the flow field. The coolant passage model consisted of two square passages, each having a 20 hydraulic diameter length, separated by a web of 0.2 passage widths and connected by a sharp 180° bend with a rectangular outer wall. Ribs were mounted on the bottom and top walls of both legs with a staggered arrangement and at 45° to the flow. The rib height and spacing were 0.1 and 1.0 passage heights respectively. The measurements were obtained for a flow condition with a Reynolds number of 50000. The geometry is similar in the straight sections to a previous configuration but has ribs extending into the bend region. The paper presents new measurements of the flow in the straight legs of the passage, comparison of the flow in the bend region with a previous configuration and details of the flow recovery from the bend in the second leg.

Linear and non-linear low Reynolds number k-ɛ eddy viscosity models are applied to the simulation of unsteady transitional flows pertinent to axial turbomachine aerodynamics. Adequate test cases are selected to allow an individual investigation of wake velocity and wake turbulence effects on unsteady transition induced by separation bubbles and wake passing respectively. Separation bubbles under steady main flow conditions are calculated first, demonstrating the applicability of the numerical scheme and the selected turbulence models. Unsteady simulations show that essential features of the experimental investigation can be reproduced, although oscillation characteristics of the separation bubble, such as phase shifts, can only be partly reproduced. Predictions of wake-induced transition on a flat plate agree with the underlying direct numerical simulation study in many important aspects. However, simulation results show deficiencies with respect to the time-averaged transition length and the spatio-temporal evolution of turbulent flow regions.

The evolution of the pressure loss coefficient and the exit flow angle has been investigated for a conventional low pressure turbine rotor blade of Snecma-Moteurs mounted in a stationary cascade arrangement at the von Karman Institute. As the boundary layers along those airfoils could exhibit laminar characteristics over a non-negligible distance, the profile losses are directly linked to the occurrence of transition and separation. The main parameters affecting the boundary layer status are taken into account. The exit Reynolds and Mach numbers respectively range between 190000 and 680000 and between 0.6 and 0.9, the inlet free stream turbulence is 0.8 per cent. The pitch-chord ratio has also been varied from its nominal value. Although general tendencies for the loss coefficient variation with respect to each parameter have been found, it has been demonstrated that a better loss estimation needs to consider the conjunction of all these parameters. The present work also points out the weakness of the traditional correlation methods when the aerodynamic load of the blades is increased.

The present paper deals with the detailed numerical investigation of transonic, turbulent, unsteady two-phase steam flows with internal heat addition after partial condensation of the working fluid. The observed phenomena include homogeneously, homogeneously-heterogeneously and heterogeneously dominated condensation effects. The results concerning the different condensation mechanisms are initially discussed in terms of the turbulent, transonic flow through Laval nozzles. Further, an interesting phenomenon of nucleation and droplet growth inside turbulent boundary layers is presented. Application to transonic turbine flows in the rotor cascade VKI-1 shows the influence of homogeneous nucleation on the turbulent shock-boundary layer interaction. Finally, the different types of condensation models were compared to investigate the two-dimensional transonic two-phase flow in an entire turbine stage including high-frequency rotor-stator interactions.

State-of-the-art centrifugal compressors for turbocharger applications are required to provide broad compressor maps, high pressure ratios and high efficiency levels. Usually these requirements are perceived as contradictory and represent challenging design targets. Various techniques for map width enhancements have been reported since the early 1980s. ABB Turbo Systems Limited has adopted a simple bleed system for internal flow recirculation to a high flowrate, high efficiency, pressure ratio 4.2 centrifugal compressor stage. After initial test runs had proved the effectiveness of the bleed system, computational fluid dynamics (CFD) calculations were performed in order to gain insight into the flow pattern in both the compressor stage and the bleed system. The simulations explain the effectiveness by comparing them with simulations without the bleed channel. Selected streamline and iso-surface plots show the effect of the bleed channel flow on the main flow through the impeller. Four more variants of bleed channels have been modelled and investigated by CFD in order to optimize the bleed slot location and dimension. The most promising configuration, which showed a significant improvement in map width without a loss in efficiency, was subsequently built and tested. This test confirmed the validity of the CFD simulation results.

The present report outlines the numerical simulation of the flow in a hot water circulating pump. The impeller, diffuser and spherical casing of an existing machine are simulated using different computational fluid dynamics programs. Calculation and analysis of both the individual components and their interaction as a coupled system serve as basis for step-by-step improvement. Further, the newly designed impeller is modelled together with the impeller side gap geometry to compute numerically the internal leakage losses and particularly the axial thrust during operation. The measured pump characteristic shows a significant improvement in efficiency compared with the original hydraulics, and the calculated and measured axial thrust agree well over a wide range of operating points.

Two-dimensional particle image velocimetry has been used in order to analyse the flow in the outlet part of a centrifugal pump (impeller and vaneless diffuser). Tests have been carried out in air. The paper first describes the main characteristics of the centrifugal pump and the associated test facilities. An analysis of the main results in design operating conditions of the SHF (Société Hydro-technique de France) impeller is then proposed.

This paper provides the results of a detailed flow investigation within a centrifugal pump equipped with a vaned diffuser. The measurements made with a laser Doppler velocimeter were carried out at two operating points. Unsteady velocity measurements obtained in phase with the impeller angular position gave access to the relative flow within the impeller channels. Measurements were obtained in the impeller and the diffuser at different measuring planes relative to the diffuser vanes. Results are presented as animations reconstituting a temporal evolution of the flow permitting a better comprehension of the complex flow structure existing between the two interacting blade rows. The analysis of the relative flow field indicates the presence of a jet wake structure. The wake is characterized by low relative velocities localized in the suction side-shroud corner. At the design flowrate, the presence of the vanes seems to have a limited effect on the impeller flow structure, except when the suction side of the blades is facing the diffuser vanes. At the impeller discharge, the time-resolved sequences show that the mixing process of the unsteady and periodic flow leaving the impeller is affected by the presence of the diffuser.