A P P L I C A T I O N B R I E F S F R O M F L U E N T

EX143

Backward-Inclined Centrifugal Fan

FLUENT is validated in this example of flow through a backward-inclined centrifugal fan. A range of fan
flow rates was studied. The steady-state multiple reference frames (MRF) model along with the realizable
k-e turbulence model are shown to suitably capture several performance characteristics of the fan when
compared to available test data.

The fan under consideration All solutions were

is a backward-inclined performed using second
centrifugal fan with a order discretizations for
conventional rotor design. all equations and the
The fan was tested in a standard SIMPLE
laboratory (in accordance pressure-velocity
with ANSI/AMCA 210-85, coupling scheme. The fan
ANSI/ASHRAE 51-1985 geometry is shown in
procedures) by mounting Figure 1. It consists of a
the outlet of the fan to the top rim (shroud), 15
inlet of a wind tunnel. Air blades, and back plate.
from the ambient was The rotor is installed in a
permitted to enter the fan scroll-type casing, which
through the inlet orifice Figure 1: The collects the flow from the
from all directions. The fan geometry rotor and discharges it
pressure rise and flow rate through a rectangular
were measured in the Therefore, the inlet-rotor domain outlet. An inlet orifice is
wind tunnel using conventional was modeled using the moving also employed to help direct the
techniques (static pressure taps reference frame model (constant flow into the rotor with minimal
and flow nozzles). Performance rotational speed), while the casing loss.
data for the fan, which included domain was assumed to be
pressure rise, shaft power, and stationary. The
sound pressure level, were effects of turbulence
collected for a nominal operating were modeled using
speed and a range of flow rates. the realizable k-
Due to minor variations in speed turbulence model
and air temperature, all data were with non-equilibrium
corrected to the nominal rotational wall functions. The
speed and standard atmospheric working fluid (air)
density (0.075 lbm/ft3). was assumed to be
incompressible with
One of the goals of this study was constant properties
to determine the efficacy of the (density = 0.075
Figure 2: The
steady-state MRF approach in lbm/ft 3, viscosity = surface mesh
simulating centrifugal fans. 1.2x10-5 lb/ft-s). used

Copyright 2001 Fluent Inc. EX146 Page 1 of 3

 * Power Coefficient: coefficients over-predict the data
= P/(N3D5) throughout the range of flow
* Efficiency: rates, with a maximum error of
= / about 12%.

where Q is the The efficiency comparisons are

volumetric flow rate shown in Figure 5. These reflect
through the fan, N is the the discrepancies in the power
fan speed in rev/sec, D is coefficient results; however, the
the rotor diameter, p is point of peak efficiency is
the pressure rise across correctly predicted, and the errors
the fan, P is the power in most of the efficiency values
drawn by the fan, and are well within 10%.
is the air density. The
power is computed from The pressure distribution on the
Figure 3: Pressure rise coefficient vs. flow coefficient the CFD results by rotor and fan casing for an
multiplying the torque on intermediate flow rate is shown in
The computational grid for the Figure 6. The pressure rise
centrifugal fan was created through the fan is clearly seen
using GAMBIT. The in this figure, as are the radial
geometry for the fan rotor and pressure gradients in the
casing was obtained in the casing.
form of IGES geometry files.
This geometry was used as the Figure 7 shows the relative
basis for constructing flow velocity vectors at a cutting
domain volumes within plane midway between the
GAMBIT. The surface mesh bottom of the rotor and the top
created is shown in Figure 2. edge of the rotor exit for the
This was used to generate the intermediate flow rate shown
final hybrid, unstructured in Figure 6. The flow is fairly
mesh containing 543,028 cells, steady and uniform at this
with tetrahedral elements in flow rate, as well as at higher
the inlet-rotor domain, and flow rates. At lower flow
Figure 4: Power coefficient vs. flow coefficient
hexahedral elements in the
casing domain. the fan blades by the
angular speed of the fan
Solutions were obtained for a rotor. In Figure 3, the
range of flow rates so that fan pressure rise coefficient
performance data were generated is plotted as a function of
and compared with the available flow coefficient, and the
test data. The following non- FLUENT predictions are
dimensional parameters were in excellent agreement
used to characterize the fan with data.
performance:
The power coefficient
* Flow Coefficient: results are shown in
= Q/(ND 3) Figure 4. Even though
* Pressure Rise Coefficient: the trend is correctly
= p/(N2D2) predicted, these Figure 5: Efficiency vs. flow coefficient

Copyright 2001 Fluent Inc. EX143 Page 2 of 3

 To summarize, rate and the point of peak
performance efficiency, were correctly
calculations for a predicted. These results suggest
backward-inclined that a steady-state MRF approach
centrifugal fan were can be used effectively for
carried out using the computing centrifugal fan flows.
FLUENT 5 CFD solver While the present calculations are
on an unstructured, reasonable approximations to the
hybrid mesh. The mean flow field within the fan, it
numerical results were is to be expected that, as the flow
found to be in good begins to breakdown at very low
agreement with the flow rates, the flow will become
available test data. highly unsteady. Consequently,
Moreover, important the steady-state MRF approach
performance trends, will not provide acceptable
such as the variations in accuracy, and an unsteady
pressure rise and (sliding mesh) solution will be
Figure 6: Static pressure contours
efficiency versus flow required.
for the intermediate flow rate case

rates, however (not shown), the

flow appears to be stalled in some
of the rotor passages as a result of
localized recirculation zones. In
these conditions, the flow is not
well suited to the MRF
formulation, and needs to be
properly simulated with the
sliding mesh model.

Figure 7: Velocity vectors on the mid-plane for the intermediate flow rate case

Copyright 2001 Fluent Inc. EX143 Page 3 of 3