Numerical Simulation of a Centrifugal Compressor

S.M.Swamy.et.al. Int. Journal of Engineering Research and Application www.ijera.com
ISSN : 2248-9622, Vol. 7, Issue 6, (Part -3) June 2017, pp.46-53
www.ijera.com DOI: 10.9790/9622-0706034653 46 | P a g e
Numerical Simulation of a Centrifugal Compressor
S.M.Swamy*, V.Panndurangadu** and R. Rama Chandra***
* Assistant Professor, Department of Mechanical Engineering, GNITS,Shaikpet, Hyderabad-500104, India.
** Professor, Department of Mechanical Engineering, JNTUA, Ananthapuramu -515002, A.P, India.
*** Professor & Principal, Department of Mechanical Engineering SKD Engineering College, Gooty, A.P.
ABSTRACT
In this paper, the tip clearance effects on flow field of a low speed centrifugal compressor without and with
partial shroud (PS) fitted on to the rotor blade tip at three values of tip clearance, viz. = 2.2%, 5.1% and 7.9%
of rotor blade height at the exit at three flow coefficients, namely, = 0.18, 0.28, and 0.34, was studied. Based
on the theory of computational fluid dynamics (CFD), performance curves and parameter distributions of the
compressor were obtained from the 3-D numerical simulation by using ANSYS CFX 15.0. The centrifugal
compressor in aerodynamic requirement is that edge velocities along the impeller channel passage surfaces like
hub, shroud, pressure and suction surfaces vary smoothly without sudden decelerations, which cause flow
separation leading to losses. Using the periodic boundaries and defined flow conditions at inflow / exit flow and
blade rotations, the turbulent viscous flow between blade channels are computed. The efficiency related
parameters using average quantities, besides flow pattern in terms of velocities, streamlines and pressure
distribution on blade surfaces are graphically interpreted. An attempt is also made to study the influence of
pressure loads on structural deformations in the chosen blade profile. This paper highlights aero-mechanical
features of centrifugal impeller obtained from several numerical simulations, which are expected to provide a
sound basis for further investigations.
Keywords: Centrifugal compressor, Flow coefficient, Partial shroud (PS) attached to the rotor blade tip,
Streamlines, passage wake.
I. INTRODUCTION
The centrifugal compressors have an
extensive variety of applications particularly for
power plants for small aircraft and helicopters, in
process industries, compression of gases and
vapours, because they can provide high-pressure
ratios and large operating ranges with relatively
high efficiencies. Centrifugal compressors are used
primarily for their suitability for handling small
volume flows, but other advantages include a
shorter length than an equivalent axial flow
compressor, less susceptibility to loss of
performance by buildup of deposits on the blade
surfaces and their suitability to operate over a wide
range of mass flow. The gas turbine engines
powering the most passenger and military aircrafts
are also called jet engines. Tip clearance in
centrifugal compressor causes the leakage of high
pressure fluid from pressure surface to suction
surface of the impeller blade, making the flow field
highly complex and effecting the performance. In
centrifugal compressor due to the low efficiency
and thrust to weight ratio, they are limited to short
range applications. However, the inability for spark
ignition engines to work at high altitude warrants
further development to improve the efficiency of
these miniature gas turbines. Hayami (1997) has
found from his experiments that axial movement of
the casing has better efficiency over the movement
of casing in radial and axial directions. Radial
movement of casing increases clearance at inducer,
which reduces the operating range. The tip
clearance studies are conducted to understand the
flow behavior in order to minimise the effect of tip
clearance. Pampreen (1973), Mashimo et al.
(1979), Sitaram and Pandey (1990) have conducted
experimental studies and suggested that by
reducing the tip clearance gap size, the tip
clearance effect can be minimised.
II. MODELING OF COMPRESSOR
BLADES
The major geometrical details of the
impeller are total pressure rise, ∆P: 300 mm WG,
Speed of rotation, N: 2000 rpm, Impeller inlet
diameter at inducer tip, d1t; 300 mm, Impeller inlet
diameter at inducer hub, d1h; 160 mm, Impeller
outer diameter, d2: 500 mm, Blade angle at exit: (a)
At hub: 750
, (b) At mean section:900
, (c) At
shroud:1050
, Blade angle at inlet inducer hub, β1h:
530
, Blade angle at inducer tip, β1t: 350
, Blade
width at the exit, b: 34.7 mm, No. of blades of the
impeller, Nb: 16. All the angles are with respect to the
tangential direction.
RESEARCH ARTICLE OPEN ACCESS

Fig. 1 Schematic diagram of centrifugal compressor setup
Schematic diagram of centrifugal
compressor setup as shown in fig.1.The geometric
model used for the computational simulation of
Centrifugal impeller with above specifications with
3mm thickness throughout the blade for generation
of three dimensional blades was done as shown in
Fig. 2. A single passage of the impeller with inlet at
50mm ahead of the impeller and outlet at a distance
of 35mm downstream of impeller is shown in Fig.
3. Casing is designed with 0.7mm clearance
throughout the blade height. Three tip clearances of
2.2%, 5.1% and 7.9% of trailing edge blade height
are obtained by moving the casing axially. ANSYS
CFX 15.0 version software is used for obtaining the
solution and standard k-ε turbulence model is used
for the closure. The computational domain is kept
at 50 mm ahead of the eye inlet of the impeller in
order to ensure that the inlet boundary conditions
are not affected by the back pressure of impeller
blade. The blade gen layout with construction
features related to wrapper angle, number of
blades, angles distribution and blade thickness for
generation of three dimensional blades was done in
order to carry out structural analysis of the blade
surfaces the software provided export options to
generate computational mesh accounting blade
thickness is also explored in static structural
analysis package. The blade profiles generated in
above software module are imported to template
based multi-block structured computational mesh
software module called turbo grid. The information
related to machine data, profile curves, several
inputs required for width factor periodicity
production, topology freezing for generation of
geometric representation in terms of surface mesh
generation with master and slave control points
being generated. default orientation of master and
slave control points on the blade surfaces are to be
moved to obtain better orthogonally between mesh
points. After these operations three dimensional
computational mesh generation process for
compressor blade with boundary patches, periodic
surfaces are created.
Fig. 2 Centrifugal compressor (without and with PS)
1 Inlet nozzle
2 Suction duct
3 Bullet nose
4 Spacers
5 Position of inlet probe
6 Rotor
7 Diffuser shroud
8 Position of probe at exit 1
9 Position of probe at exit 2
10 Volute casing
11 Diffuser hub
12 Bearing housings
13 Bearing housing cover
14 Delivery duct
15 D. C. motor
16 Throttle cone

Fig. 3 Computational domain of single passage (without and with PS)
III. COMPUTATIONAL MESH FOR
STRUCTURAL ANALYSIS
Most fluid flow simulation technology
uses completely different meshing, loading and
solving methods compared to structural simulation.
And this is where the true power of Workbench
15.0 comes into play. When the user is connecting
up the dialogs within the schematic layouts, many
of these are handled automatically, simply by
dragging a connection between the outputs of one
analysis system and the inputs of another. When
and if there’s mapping to be done, this is handled in
a very simple dialog that makes the process simple,
rather than the often-tortuous methods traditionally
associated with this type of work . There are no
geometry creation tools in the Mechanical
application and so geometry must be attached to
the Mechanical application. The geometry from
either of the CAD Systems and Blade Gen Model
can be connected to simulation model before setup
option is activated. The next step is to generate the
meshing solid body for which several methods are
provided in ANSYS Workbench 15.0 these tools
have the benefit of being highly automated along
with having a moderate to high degree of user
control. Upon startup of the meshing application
meshing options panel displays, which allows you
to quickly and easily set your meshing preferences
based on the physics you are preparing to solve.
For solid bodies, the software uses meshing
techniques based on algorithms and element shape.
One of the technique extensively used are called
patch independent meshing for the part using
tetrahedral element. The default value selects
meshing control parameters and provides a
successful automated mesh. By default, the
application attempts to use auto sweeping for solid
models and quadrilateral element generation for
surface body models. If a model is not sweepable,
the Patch Conforming mesher under tetrahedrons is
used. Then one needs to change the default values
suitably in the Mesh option panel. Patch
independent mesh are defined by max element size
and number of elements. The default value is
calculated from the diagonal length of the bounding
box which is based on the initial seed settings of
mesh object folder. When curvature and proximity
refinement options enabled, the mesh automatically
refined based on geometry. This will result in
larger elements on flat planar faces and smaller
elements in areas of high curvature or within small
gaps. However projection to geometry and
smoothing may push the size even smaller for some
of the elements as observed in blade leading edges.
IV. FLOW SIMULATION IN
CENTRIFUGAL COMPRESSOR
BLADE
To verify the performance of centrifugal
compressor using computational fluid dynamics
commercial codes ANSYS CFX 15.0. We modeled
the geometries with and without partial shroud
using CATIA modeling software procedure. We
imported the airfoil sections at certain heights then
made the surface using multi-section tool in GSD
module of CATIA software. By closing upper and
bottom surface we get the solid section of an one
blade we have to add the partial shroud at tip of the
blade at its trailing edge with the thickness of given
height and 50mm in length along the blade and
2mm width more than actual blade thickness from
its pressure side of airfoil. Next, extraction of fluid
domain total number of blades for this particular
centrifugal compressor are 16 so it’s possible to
simulate only one periodic section of fluid to
interpolate the results for other remaining sections
of fluid domain. This extraction is also based on
equalized divisions of cylindrical fluid domain. We
have extracted fluid domain in CATIA software
itself. Meshing is the next step, in the meshing
nodes on the higher and lower periodic surfaces we
should have equal number of nodes. Match control
option is used for that. Name selection is important
for each surface of non-conformable meshing in

rotating bodies (i.e. with sliding contact notion).
CFX can detect automatically the respective
surfaces names for respective parts of compressor.
For each surface given a meaning full name as
shown in Fig.4 (a), where S1 and S2 representing
stationary inflow and outflow domains
respectively. R1 is rotating domain which
containing blade surfaces (including partial shroud)
inside it. As standard notations are used to
represent the shroud, hub, periodic surfaces, blade
surfaces, and inflow outflow surfaces.
The quality of meshing and number of
elements is very important concern for CFD
simulation this can be seen in below table. Quality
of mashing is measured usually by skewness
present in meshing. It’s a ratio of smallest angle
between edges of element present is actual meshing
by angle between edges in standard element (angle
for standard elements are 60° for triangular mesh
and 90° for rectangular mesh). In other words it’s
showing fraction of deviation of element shape
from standard element shape. In below case
maximum skewness is 0.7977 which is acceptable
this value should be less than 0.9. Number of
elements are 2310116. This much of elements are
require testing mesh independency.
a) b)
Fig.4 a) Meshing Details with Name Selection b) mesh statistics
Fig.5. Frequency of distribution of skewness.

In CFX we have to use turbo mode for this
problem statement, with respective to this
stationary, rotating components are defined. As the
density going to vary (compressible fluid) with
pressure fluid properties air kept as ideal gas.
Reference pressure is 1 atm. Heat transfer is
calculated by total energy because increase in
density causes to increase in temperature and
because of rotational energy it will try to change
the heat transfer rate, in total energy methods only
all this effects will get considered. Shear Stress
Transport Turbulence model was used because of
near wall treatment is required and presence of ad-
versed pressure gradient is present. Inlet and outlet
boundaries are pressure inlet with 0 Pa gauge
pressure and mass flow 0.087 Kg/s at outlet
respectively. Rotating domain interface used as
frozen rotor. Stage for interfaces used for steady
state simulations, rotational interaction between
reference frames is not accounted for – i.e. the
stationary domain “sees” the rotating components
at a fixed position. Physical Time scaling value was
observed to be taken as 10-3
s. Accuracy criteria of
RMS solution was decreased to 10-5
for better
accuracy.
Fig. 6 Extracted Fluid Domain Fig.7 Pressure at outlet, casing, periodic surface and inlet (Clearance 2.2%
without PS)
Fig. 8 Three dimensional mesh for single compressible blade row and streamlines from the inlet of impellor
around the blades
V. RESULTS AND DISCUSSIONS
The impellor blades are rotating with 2000
rpm is subjected to mass flow 0.087 kg/sec with the
working pressure is of order 1 atm. The pressure
loading over stream wise directions for the span of
70% is shown in the Figure. The pressure profiles
on the surfaces near hub mean and tip region
provides details of boundary layer separation with
steep adverse pressure gradients [6]. Pressure drops
rapidly with the increase in mass flow rate and this
will lead to choking at a certain point. On the other
hand, pressure also drops accordingly as the mass
flow rate is reduced from that optimum point and
the stage will quickly start to surge.
Velocity Vectors at Outlet
A velocity vector at the exit of the
impeller is shown in fig. 9. The fluid flows with PS
on tip of the blade velocities are high on suction
surface than pressure surface because of blade
curvature. With increase in tip clearance, the
velocity on both pressure side and suction side is
decreasing. For 2.2% and 5.1% clearance, high
velocity of the fluid above the blade from pressure
side to suction side through tip clearance is clearly
seen.

Fig. 9: Velocity vectors at exit of the blade for 2.2% of clearance for without PS and with PS
Static Pressure Contours at Span 0.7
Pressure contours in blade to blade view,
at span 0.7 is shown in figure 10. The contours
show gradual pressure rise from inlet to outlet of
the compressor due to dynamic action of the
rotating impeller. Gradual increase of static
pressure from inlet to outlet is clearly observed at
all tip clearances. With PS on tip of the blade, low
pressure change is observed. But without PS on tip
of the blade, the pressure at outlet is reduced. High
pressure on pressure side of the blade and low
pressure on suction side of the blade are observed
at all tip clearances. With increase in tip clearance,
reduction in pressure on both pressure side and
suction side is found.
Fig.10: Static pressure contours for 2.2% clearance at mid span without PS and with PS
Velocity Contours at Span 0 .7
Velocity contours in blade to blade
view, at span 0.7 is shown in figure 11. The
contours show low velocity region on suction
side of the blade. With PS on tip of the blade, the
low velocity region is reducing and also velocity
improvement is observed.
Fig. 11: Velocity contours for 2.2% clearance for without PS and with PS
Outlet to inlet pressure ratio at different flow
coefficients is increase in pressure ratio with PS on
tip of the blade is observed at all flow coefficients.
But without PS on tip of the blade, reduction in
pressure ratio is observed at all flow coefficients
because of more fluid leakage through the basic
configuration.
Blade loading curves:
Blade loading curves at design flow
coefficient 0.28 for 2.2% tip clearance at without
PS and with PS were shown in fig. 12. Fluid flow
without PS and with PS on tip of the blade,
significant change in pressure on suction side is
observed. Low static pressure on suction and high
pressure on pressure side of the blade is observed.
With increase in tip clearance, static pressure on
both pressure side and suction side are reducing.

Fig.12: Blade loading for 2.2% clearance at without and with partial shroud
Fig.13: Momentum, Heat transfer and Mass converged residual plot. (Without and with PS)
Fig.14: Static pressure distribution from inlet to outlet over stream wise (without and with PS) for 2.2% of
clearance.
VI. CONCLUSIONS
Tip clearance effects on flow field of a
centrifugal compressor without and with partial
shroud (PS) attached to the rotor blade tip at three
values of tip clearance, viz. = 2.2%, 5.1% and
7.9% of rotor blade height at the exit at three flow
coefficients, namely, = 0.18, 0.28 and 0.34, is
analysed computationally using structured multi
block grid with fine grid in the tip clearance region.
Velocity contours in blade to blade view; at span
0.7 is low velocity region on suction side of the
blade. With PS on tip of the blade, the low velocity
region is reducing then velocity improved. Blade
loading charts at design flow coefficient 0.28 for
2.2% tip clearance for fluid flow without PS and
with PS on tip of the blade, significant change in
pressure on suction side is observed. Low static
pressure on suction and high pressure on pressure
side of the blade is observed. With increase in tip
clearance, static pressure on both pressure side and
suction side are reducing. It is observed that
maximum stress is taking place at the hub surface
trailing edge and minimum stress is noticed on
shroud surface when pressure load obtained from
CFD solution is specified on the blade surfaces. As
mass flow varies at constant rotational speeds the
magnitude of maxima and minima of stresses
decrease. Similarly, the total deformation also
decreases at constant speed at different mass flow
rates. The pressure from inlet to the outlet of the
compressor is increasing gradually along the
stream wise direction due to the dynamic head
developed by the rotating impeller. A drop in static
pressure near streamwise direction is observed for
two cases due to the acceleration of the flow in to
the eye of the impeller.
The pressure contours show that there is a
shifting of low momentum fluid from suction
surface to pressure surface in tip clearance region
due to tip leakage flow. Wake region near suction
side shroud corner is observed. Wake region

growth and movement towards the mid pitch with
increase in clearance is observed.
REFERENCES
[1]. Fowler, H. S., (1971) Aerodynamics Tests
on a Centrifugal Fan Impeller Model with
Swept Back Blades, N.R.C. Report No.
12020.
[2]. Senoo, Y. And M. Ishida, (1975) Behaviour
of Severely Asymmetric Flow in a Vaneless
Diffuser, Trans. ASME, Jl. of Engg. for
Power, 97, 375-382
[3]. Moore, J., (1973), "A Wake and an Eddy in
a Rotating Radial-Flow Passage", Part 1:
Experimental Observation", Part 2:"Flow
Model", Trans. ASME Journal of
Engineering for Power.pp.205-219.
[4]. Eckardt, D., (1976) Detailed Flow
Investigations within a High Speed
Centrifugal Compressor Impeller, Trans.
ASME, Jl. of Fluids Engg. 98, 390-402.
[5]. Adler, D., (1980) Status of Centrifugal
Internal Aerodynamics, Part II: Experiments
and Influence of Viscosity, Trans. ASME, Jl.
of Engineering for Power, 102, 738-746
[6]. Shivaprasad, B. G. and B. R. Ramaprian,
(1977) Turbulence Measurements in
Boundary Layers along Mildly Curved
Surfaces, ASME Paper 77-WA/FE-8.
[7]. Krain, H. and D. Eckardt, (1978) The Flow
Field in a High Speed Centrifugal
Impeller-A Comparison of Experimental and
Theoretical Results, First International
Conference on Centrifugal Compressor
Technology, IIT Madras.
[8]. Johnson, M.W., and Moore, J., (1980), "The
development of wake flow in a centrifugal
compressor" ASME Journal of Engineering
for Power, Vol. 102, pp. 383-390.
[9]. Johnson, M. W. and J. Moore, (1983)
Secondary Flow Mixing Losses in a
Centrifugal Impeller, Trans. ASME, Jl. of
Engg. for Power, 105, 24-32.

Numerical Simulation of a Centrifugal Compressor

More Related Content

What's hot (15)

Similar to Numerical Simulation of a Centrifugal Compressor (20)

Recently uploaded (20)

Numerical Simulation of a Centrifugal Compressor