High speed centrifugal compressor

Application Challenge 6-08 © copyright ERCOFTAC 2004

Description

Introduction

The SRV2 is a high speed centrifugal impeller, with a very high total pressure ratio (close to 6 which has been designed and investigated in an industrial research project of German and Swiss turbo-compressor manufacturers. The main focus of this work, as outlined in Eisenlohr et al. [2], was to measure and calculate the global flow field structure to provide indications for necessary blade design changes that could lead to a further improvement of the impeller efficiency. High quality performance measurements and extensive laser measurements (Laser 2-Focus L2F) of the flow field in the rotating impeller were carried out at the DLR, and CFD simulations were carried out by the project partners.

The companies involved in this project together with the DLR performed 3D calculations with different codes, involving different grids and using different turbulence models. In addition to comparing the measured and predicted pressure ratio characteristics, the measured and calculated flow fields at design point were compared. This earlier work was unable to come to any final conclusions with regard to the requirements for best practice for these simulations, although some simulations were found to be fairly inadequate. In this current work, more recent simulations with CFX-TASCflow are described, whereby high standards of grid definition and resolution have been applied. In addition, simulations have been carried out for different tip clearance levels, as a common feature of all high speed open impellers including this one is that there is inevitably some uncertainty related to the level of running tip clearance in the machine.

Relevance to Industrial Sector

This test case is a highly relevant and extremely difficult test case for assessing the use of CFD in high speed turbomachinery. The high pressure ratio (of nearly 6) in this centrifugal impeller is at the upper end of typical applications in radial compressors for gas turbines (in helicopter engines and APUs) and is substantially higher than that currently used in radial compressors for turbochargers (typically 3 to 4.5) or other industrial gas compression applications (HVAC and industrial gas processing). The test case is thus representative of extremely highly loaded compressor applications.

The geometry of this impeller is typical of most modern centrifugal compressor stages with splitter vanes and with about 38 degrees of back-sweep at the impeller outlet. The impeller is well-designed and has typical modern values for all of the key global geometrical parameters (axial length, inlet hub diameter, inlet tip diameter, blade inlet angles, blade wrap angle, blade thickness, etc.) which make it highly relevant as a typical industrial test case.

One of the major difficulties in turbomachinery simulations at these high pressure ratios is the compressible nature of the medium being compressed, as this means that the prediction of the flow field becomes strongly dependant on the correct prediction of the performance. A small error in the prediction of performance (losses, efficiency, work input, pressure ratio and temperature ratio) will lead to an error in the predicted density of the gas, and this automatically causes an error in the volume flow leading to a change in the predicted mean flow velocities. This type of error is typical of high speed turbomachinery (in radial machines and in multistage axial compressors and axial turbines). It does not occur in incompressible flows as for these applications the density is constant and the mean flow velocities are not a function of the predicted performance. Because of this, this high speed test case is highly relevant with an added degree of difficulty compared to the low-speed centrifugal compressor test case (see application challenge submitted by Numeca – the NASA low speed centrifugal impeller).

As a result of the sensitivity of the mean flow velocity predictions to small changes in gas density, high speed turbomachinery simulations are also quite sensitive to small changes in geometry, so that an exact representation of the geometry is needed for accurate simulations. This case provides all the relevant geometrical information needed for good simulations, but as in all such high speed radial compressors there remains an uncertain area with regard to the high sensitivity of the results with regard to tip clearance flows. This is a highly relevant application uncertainty for such industrial applications that is well documented in this test case.

For turbo-compressor design, it is primarily the correct prediction of the flow field that is required as a main application of CFD. The accurate calculation of the flow field, with the prediction of the location and extent of separations and shocks, is usually considered more important than a precise prediction of the over-all performance. From the simulations the change in flow-field (such as reduction in the extent of a separated zone or the shift in the position of the shock) due to a change in geometry can be assessed. This is a key element in the design of centrifugal impellers where new versions are often based on modifications of existing impellers that have already been proven in tests. The availability of the detailed flow field measurements with L2F in the SRV2 impeller make this a highly suitable test case by which the competency of CFD for the sector can be judged with respect to such flow patterns.

Design or Assessment Parameters

The design or assessment parameters (DOAPs) for this application challenge are primarily the impeller efficiency, the pressure ratio across the impeller, the Mach number distribution along the flow path and the angle of the relative velocity at the impeller exit.

The definitions are as follows:

isentropic efficiency of impeller:

total pressure ratio across impeller: π_imp = P₀₂/P₀₁

local relative Mach number M_rel = w/√(κRT)

angle of relative velocity β = atan(w_m/w_u)

meridional velocity c_m

Apart from these, the extent of zones of separated flow, the location of shocks and the distribution of Mach number as measured by L2F traverses at different locations are means for the judgement of the CFD calculations although a clearly defined parameter for the quantification of these (other than the local Mach number distributions) is not available.

Flow Domain Geometry

The investigated impeller is un-shrouded with a small clearance gap between the rotating impeller blades and the stationary casing. It features splitter blades, i.e. every second blade is cut back from its leading edge in order to increase the swallowing capacity by removing the blade blockage near to the leading edge. The impeller also has a level of back-sweep that is typical of most modern centrifugal compressor designs. Some general global geometry data are listed in Tab. GEO-A

hub radius at inlet	r_1H	30 mm
casing radius at inlet	r_1S	78 mm
impeller tip radius (exit)	r₂	112 mm
diffuser outlet diameter	r₄	212mm
axial length	l_ax	75 mm
blade height at impeller exit	b₂	10.2 mm
number of full and splitter blades	z	13 + 13
blade angle at impeller inlet, casing	β_1S	26.5°
blade angle at impeller inlet, hub	β_1H	52.9°
blade angle at impeller exit, casing	β_2S	52.0°
blade angle at impeller exit, hub	β_2H	52.0°
wrapping angle at casing	φ_S	68.9°
wrapping angle at hub	φ_H	68.9°
rake angle	φ_Rake	0.0°
blade thickness at impeller inlet, casing	d_B1S	0.56 mm
blade thickness at impeller inlet, hub	d_B1H	1.03 mm
maximum blade thickness, casing	d_BmaxS	1.19 mm
maximum blade thickness, hub	d_BmaxH	2.95 mm

Table GEO-A Characteristic geometrical data

The meridional view of the impeller is shown in Fig. 1.1, the blade angle distribution and the thickness distribution along the normalised meridional length of the hub and tip sections of the impeller are given in Fig. 1.2.

Fig. 1.1 Meridional contour

Fig. 1.2 Blade angle distribution (left) and blade thickness (right) along hub and casing

The coordinates of the flow passage are given in cartesian and cylindrical coordinates in geo1.dat — geo7.dat. Lengths are given in meters, angles in degrees. Note that the fillets at the hub (constant radius of 2mm) have been neglected.

geo1.dat main blade (x,y,z and r,φ; 11 sections with 141 points each)

geo2.dat splitter blade (x,y,z and r,φ;11 sections with 101 points each)

geo3.dat hub contour (r,z)

geo4.dat shroud contour (r,z)

geo5.dat contour of tip clearance (r,z; 0.5mm at LE; 0.6mm at TE)
for simulation case CFD1

geo6.dat contour of 50% increased tip clearance (r,z; 0.75mm at LE; 0.9mm at TE)
for simulation case CFD2

geo7.dat contour of 100% increased tip clearance (r,z; 1.0mm at LE; 1.2mm at TE)
for simulation case CFD3

Flow Physics and Fluid Dynamics Data

The flow is at high Reynolds number and high turbulence levels, so that apart from a short laminar section to be expected close to the leading edge, the flow is turbulent throughout the whole flow passage. It is compressible and the compression is assumed to be adiabatic, i.e., with no heat flux into or from the surrounding walls. The governing non-dimensional parameters (GNDPs) are the stage Mach number Mu₂ and to a minor extent the Reynolds number defined with the impeller exit width and impeller tip speed ( Re_b). The definitions are as follows:

Mu₂ = u₂/√(κRT₀₁)

Re_b = u₂ b₂/ν₁

The working fluid is dry air.

References

The rig and the design data were described in detail in:

[1] "Aerodynamics for a Centrifugal Compressor with Transonic Inlet Conditions", H. Krain, Hoffmann, W., Pak, H., ASME Paper 95-GT-79, 1995

[2] "Analysis of the Transonic Flow at the Inlet of a High Pressure Ratio Centrifugal Impeller", G, Eisenlohr, P. Dalbert, H. Krain, H. Pröll, F.-A.Richter, K.-H. Rohne, ASME Paper 98-GT-24, 1998

The software package TASCflow is described in detail in:

[3] TASCflow User Documentation; AEA Technology Engineering Software Ltd., Waterloo, 1999

Contributors: Beat Ribi; Frank Sieverding - MAN Turbomaschinen AG Schweiz; Sulzer Innotec AG