3. Results and discussion
3.1. Simulations setup
The fluid phase was treated as air with constant density
1.205 kg/m3 and viscosity 1.82 105 kg/m s. A uniform velocity
profile (Uin) of 10.67 m/s was prescribed at the inlet, yielding a Reynolds
number based on cyclone’s body diameter of 21,900. At the
overflow outlet the static pressure was prescribed and all the
cyclone walls were considered as no-slip boundaries.
Monodisperse polystyrene latex particles with density 1050 kg/
m3 and diameters ranging from 0.5 to 6.0 lm were injected with
the same velocity as the fluid in the cyclone inlet, Stokes number,
given by Eq. (1), and the particle relaxation time are shown in
Table 2 as functions of the particle diameter. For each diameter,
a total of 1012 particles were fed into the cyclone, which were uniformly
distributed at the inlet boundary. At the solid surfaces, particles
were reflected considering a perfectly elastic collision.
Particles that crossed the overflow face were considered escaped
and deleted from the calculation. As in [15] particles that touched
the cone bottom were considered collected.
3.2. Effect of different bend position and radii in the outlet duct on
cyclone pressure drop and cut off diameter
In this first study, the velocity and pressure profiles showed, in
general, no clear tendency. Actually, in most cases, contrary to [19],
only small differences in the average profiles were observed.
Regarding the RMS profiles, some cases presented considerable differences
for the peak values (located at the cyclone center), but in
general, only small differences were obtained outside the central
region. Based on this information, most of the obtained profiles
are omitted in the present work