Flow Analysis in a Pump Diffuser—Part 2: Validation and Limitations of CFD for Diffuser Flows

1997 ◽  
Vol 119 (4) ◽  
pp. 978-984 ◽  
Author(s):  
F. A. Muggli ◽  
K. Eisele ◽  
M. V. Casey ◽  
J. Gu¨lich ◽  
A. Schachenmann

This paper describes an investigation into the use of CFD for highly loaded pump diffuser flows. A reliable commercial Navier-Stokes code with the standard k-ε turbulence model was used for this work. Calculations of a simple planar two-dimensional diffuser demonstrate the ability of the k-ε model to predict the measured effects of blockage and area ratio on the diffuser static pressure recovery at low loading levels. At high loading levels with flow separation the k-ε model underestimates the blockage caused by the recirculation in the flow separation region and overestimates the pressure recovery in the diffuser. Three steady-state calculations of a highly loaded vaned diffuser of a medium specific speed pump have been carried out using different inlet boundary conditions to represent the pump outlet flow. These are compared to LDA measurement data of the flow field and demonstrate that although the Navier-Stokes code with the standard k-ε turbulence model is able to predict the presence of separation in the flow, it is not yet able to accurately predict the static pressure rise of this highly loaded pump diffuser beyond the flow separation point.

1999 ◽  
Vol 121 (1) ◽  
pp. 119-126 ◽  
Author(s):  
E. Casartelli ◽  
A. P. Saxer ◽  
G. Gyarmathy

The flow field in a subsonic vaned radial diffuser of a single-stage centrifugal compressor is numerically investigated using a three-dimensional Navier–Stokes solver (TASCflow) and a two-dimensional analysis and inverse-design software package (MISES). The vane geometry is modified in the leading edge area (two-dimensional blade shaping) using MISES, without changing the diffuser throughflow characteristics. An analysis of the two-dimensional and three-dimensional effects of two redesigns on the flow in each of the diffuser subcomponents is performed in terms of static pressure recovery, total pressure loss production, and secondary flow reduction. The computed characteristic lines are compared with measurements, which confirm the improvement obtained by the leading edge redesign in terms of increased pressure rise and operating range.


Author(s):  
E. Casartelli ◽  
A. P. Saxer ◽  
G. Gyarmathy

The flow field in a subsonic vaned radial diffuser of a single stage centrifugal compressor is numerically investigated using a 3D Navier-Stokes solver (TASCflow) and a 2D analysis & inverse-design software package (MISES). The vane geometry is modified in the leading edge area (2D blade shaping) using MISES, without changing the diffuser throughflow characteristics. An analysis of the 2D and 3D effects of two redesigns on the flow in each of the diffuser subcomponents is performed in terms of static pressure recovery, total pressure loss production and secondary flow reduction. The computed characteristic lines are compared with measurements, which confirm the improvement obtained by the leading edge redesign in terms of increased pressure rise and operating range.


Author(s):  
M. Bugra Akin ◽  
Wolfgang Sanz

Accurate numerical simulations depend on the correct prediction of all relevant flow phenomena. For many aeronautical devices such as turbomachinery the behaviour of boundary layers, wall shear stress and wall heat transfer are significant for the performance. Turbulence and transition may influence such flow characteristics. Onset and the extent of transition can therefore be of high importance for the design process. In this paper three-dimensional steady-state simulations of a two-stage turbine with and without modeling of transition are performed. The configuration consists of a transonic high pressure turbine stage followed by an S-shaped turning mid turbine frame and a counter-rotating low pressure turbine rotor. The in-house Reynolds-averaged Navier-Stokes solver has been applied to this configuration with the correlation based γ–Reθ transition model developed by Menter et al. which has been added to the SST turbulence model. Also, calculations without transition with the SST and the Spalart-Allmaras turbulence model have been performed. This configuration is also the subject of experimental investigations at the Institute for Thermal Turbomachinery and Machine Dynamics at Graz University of Technology, so that measurement data at two planes are used for the verification of the simulations. After verification and flow analysis, the results are also discussed to evaluate the effect of modeling transition phenomena. An analysis of the overall efficiency of the turbine and the efficiency losses in the blade and vane rows are finally presented.


Author(s):  
T. Ch. Siva Reddy ◽  
G. V. Ramana Murty ◽  
Prasad Mukkavilli ◽  
D. N. Reddy

Numerical simulation of impeller and low solidity vaned diffuser (LSD) of a centrifugal compressor stage is performed individually using CFX- BladeGen and BladeGenPlus codes. The tip mach number for the chosen study was 0.35. The same configuration was used for experimental investigation for a comparative study. The LSD vane is formed using standard NACA profile with marginal modification at trailing edge. The performance parameters obtained form numerical studies at the exit of impeller and the diffuser have been compared with the corresponding experimental data. These parameters are pressure ratio, polytropic efficiency and flow angle at the impeller exit where as the parameters those have been compared at the exit of diffuser are the static pressure recovery coefficient and the exit flow angle. In addition, the numerical prediction of the blade loading in terms of blade surface pressure distribution on LSD vane has been compared with the corresponding experimental results. Static pressure recovery coefficient and flow angle at diffuser exit is seen to match closely at higher flows. The difference at lower flows could be due to the effect of interaction between impeller and diffuser combinations, as the numerical analysis was done separately for impeller and diffuser and the effect of impeller diffuser interaction was not considered.


Author(s):  
Kai U. Ziegler ◽  
Heinz E. Gallus ◽  
Reinhard Niehuis

The interaction between impeller and diffuser is considered to have strong influence on the flow in highly loaded centrifugal compressors. However, the knowledge about this influence is still not satisfying. This two-part paper presents an experimental investigation of the effect of impeller-diffuser interaction on the unsteady and the time averaged flow configuration in impeller and diffuser and the performance of these components. The flat wedge vaned diffuser of the investigated stage allows an independent adjustment of diffuser vane angle and radial gap between impeller exit and diffuser vane inlet. Attention is mainly directed to the radial gap, as it determines the intensity of the impeller-diffuser interaction. In part I it was shown that smaller radial gaps improve diffuser pressure recovery, whereas impeller efficiency is hardly affected. Part II focuses on the reasons for these effects. Measurements with a laser-2-focus velocimeter in the highly unsteady flow field between the impeller exit region and diffuser throat were performed at three different diffuser geometries allowing a detailed flow analysis. Especially the unsteady results show that for a smaller radial gap more impeller wake fluid is conveyed towards the highly loaded diffuser vane pressure side reducing its loading and leading to a better diffusion in the diffuser channel. Concerning the impeller flow, it was found that a smaller radial gap is leading to a noticeable reduction of the wake region at impeller exit. The experimental results are intended to be published as an open CFD testcase under the name “Radiver”.


Author(s):  
S. Anish ◽  
N. Sitaram

A computational study has been conducted to analyze the performance of a centrifugal compressor under various levels of impeller-diffuser interactions. The study has been conducted using a low solidity vaned diffuser (LSVD), a conventional vaned diffuser (VD) and a vaneless diffuser (VLD). The study is carried out using Reynolds-Averaged Navier-Stokes simulations. A commercial software ANSYS CFX is used for this purpose. The intensity of interaction is varied by keeping the diffuser vane leading edge at three different radial locations. Frozen rotor and transient simulations are carried out at four different flow coefficients. At design flow coefficient maximum efficiency occurs when the leading edge is at R3 (ratio of radius of the diffuser leading edge to the impeller tip radius) = 1.10. At lower flow coefficient higher stage efficiency occurs when the diffuser vanes are kept at R3 = 1.15 and at higher flow coefficient R3 = 1.05 gives better efficiency. It is observed that at lower flow coefficients positive incidence causes separation of flow at the suction side of the diffuser vane. When the flow rate is above design point there is a negative incidence at the leading edge of the diffuser vane which causes separation of flow from the pressure side of the diffuser vane. Compressor stage performance as well as performance of individual components is calculated at different time steps. Large variations in the stage performances at off-design flow coefficients are observed. The static pressure recovery coefficient (Cp) value is found to be varying with the relative position of impeller and diffuser. It is observed that maximum Cp value occurred at time step where Ψloss value is lowest. From the transient simulations it has been found that the strength and location of impeller exit wake affect the diffuser vane loading which in turn influences the diffuser static pressure recovery.


Author(s):  
W. N. Dawes

The aim of this paper is to help advance our understanding of the complex, three-dimensional, unsteady flow associated with the interaction of a splattered centrifugal impeller and its vaned diffuser. A time-resolved simulation is presented of the Krain stage performed using a time-accurate, 3D, unstructured mesh, solution-adaptive Navier-Stokes solver. The predicted flowfield, compared with experiment where available, displays a complex, unsteady interaction especially in the neighbourhood of the diffuser entry zone which experiences large periodic flow unsteadiness. Downstream of the throat, although the magnitude of this unsteadiness diminishes rapidly, the flow has a highly distorted three-dimensional character. The loss levels in the diffuser are then investigated to try and determine how time-mean loss levels compare with the levels expected from “equivalent” steady flow analysis performed by using the circumferentially averaged exit flow from the impeller as inlet to the diffuser. It is concluded that little loss could be attributed directly to unsteady effects but rather that the principle cause of the rather high loss levels observed in the diffuser is the strong spanwise distortion in swirl angle at inlet which initiates a strong hub/comer stall.


Author(s):  
Yong Qin ◽  
Ruoyu Wang ◽  
Yanping Song ◽  
Fu Chen ◽  
Huaping Liu

Numerical investigations on the control effects of synthetic jets are conducted upon a highly loaded compressor stator cascade. The influence of forcing parameters including actuation frequency, jet amplitude and slot location are analyzed in detail with the single-slit synthetic jet. Besides, a new slot arrangement is put forward for the purpose of effectively controlling flow separation. Simulation results validate the remarkable effectiveness of the single-slit synthetic jet on controlling flow separation. Owing to the coupling effect between the jet and the main flow, the actuation appears to be most efficient under the characteristic frequency of the main flow passing through the airfoil. Additionally, with the increase of jet momentum coefficient, the control effect is enhanced at first and then decreased, depending on the two aspects: the improvements of aerodynamic performance by momentum injection and the additional flow losses caused by the jet. Compared to other actuator configurations, the segment synthetic jet with three sections can more effectively deflect the end-wall cross flow and thus impede the development of corner vortex, which helps to restrain the accumulation of low momentum fluid towards the corner, emphasizing the importance of slot arrangement. Accordingly, under the optimum condition, the total pressure loss coefficient gains a 15.8% reductions and the static pressure rise coefficient is increased by 5.01%.


1987 ◽  
Vol 109 (4) ◽  
pp. 550-556 ◽  
Author(s):  
Yingkang Zhu ◽  
S. A. Sjolander

The paper presents measurements of the steady aerodynamic characteristics of a series of five radial vaneless diffusers with walls varying from mildly divergent to strongly convergent. The static pressure recovery was determined and the flow was traversed at the inlet and the outlet of the diffuser for a broad range of flow rates in each case. It was found that wall convergence results in a negative (stabilizing) slope in the pressure rise curve for the diffuser. Furthermore, at high flow rates convergence was found to reduce the pressure recovery far less than one would expect and at intermediate flow rates convergence actually improved the pressure recovery. The better-than-expected performance is thought to be closely related to the observed improvement in the flow uniformity at the diffuser outlet when convergent walls are used.


2005 ◽  
Author(s):  
Taku Iwase ◽  
Kazuyuki Sugimura ◽  
Taro Tanno

We designed an axial fan for servers using computational fluid dynamics (CFD) and numerical optimization. The performance of the fan, namely static pressure rise and efficiency, was calculated using commercial CFD software based on an incompressible Reynolds-averaged Navier-Stokes (RANS) solver. An automatic program developed in-house was used to generate the grids for CFD calculation. Numerical optimization—using a simulated annealing algorithm (SA)—was used for determining the optimized shape of the fan. After optimizing the fan, initial and optimized fan designs were made for experiments using rapid prototyping, and their performances, based on such things as efficiency and noise level, were measured. Results demonstrated that the optimized fan design achieved higher efficiency than the initial design. Multi optimization was also developed for maximizing the fan efficiency and minimizing the casing height. An additional finding was that there was a trade-off between the fan efficiency and casing height.


Sign in / Sign up

Export Citation Format

Share Document