Abstract:
For more than a century, cyclones have carved out their place as a simple static device to
separate gas-solid particles mixture in chemical process industry. However, their use as heat transfer
equipment is recent one. Cyclones are suitable for heating solid particles where direct contact with
hot gases is allowed. Their use as raw meal preheater in cement industry has already proved their
worth in saving energy. They hold much promises to be used as gas-solids heat exchanger as well as
separator in many allied industries like fertiliser, polymer powder, pharmaceutical and food stuffs in .
near future. The above latent potential of cyclones merits study on gas to solids heat transfer in
cyclone.
Despite its considerable potential of being used as heat recovery equipment, research works
leading to the sizing, design and performance prediction ofthe cyclones as heat transfer equipment
are hardly available in open literature, mainly because of its high patent value. It appears that
Yen et al. (1990) are the pioneer investigators in this field followed by Raju et al. (1994). Both the
groups of researchers have done experimental work on heat transfer from air to sand inside acyclone
and have proposed correlations for the prediction of heat transfer coefficient. Although both the
investigators have addressed many aspects ofthis highly complex problem, the correlations proposed
by them do not predict data ofother investigator within acceptable limit oferror. The above lacuna
seems to be obvious, as during the above period hardly any substitutive experimental data were
available neither to examine the validity of their correlations and nor to improve it. This calls for
further refinement ofthe existing correlations to develop a more meaningful and widely applicable
correlation.
The cyclones are known for their complex flow fields. Addition of particles in to it and heat
transfer between hot air to moving particles makes it all the more complex. Thus, the phenomena of
gas to solid particles heat transfer inside the cyclone is highly complex and integrated/coupled. Since
the movement of particles inside a cyclone may be approximated as a combination of fluidised bed
and moving bed, it naturally calls for the study of associated phenomena taking place in fluidised,
moving and packed beds to draw inferences and analogies.
In the above backdrop, the present experimental studies have been planned to generate
reliable data for gas to solids heat transfer in cyclone and to propose improved correlations for the
estimation of heat transfer coefficient, holdup of solid particles in the cyclone, ratio of pressure drops
for solid particles laden air to that of solid free air at same inlet air velocity across cyclone, and outlet
temperatures of gas and solids.
To achieve the above objectives, a well instrumented experimental setup has been designed,
fabricated and commissioned to simulate the process of heating of solid particles by hot air inside a cyclone. A tangential entry reverse flow cyclone fabricated with mild steel sheet is used in the
present study. The cyclone has internal diameter equal to 100 mm, length of straight cylindrical
section equal to 200 mm and conical section equal to 200 mm with rectangular inlet of size 20 mm x
50 mm and a circular outlet of 50 mm diameter. Hot air at about 200 °C (higher than that used in
earlier investigations) is used as gas phase whereas the locally available river sand at ambient temperature of about 30 °C is used as solids phase to exchange heat.
Before the experiments for the present study are initiated, the reliability and integrity of the
experimental setup has been established by conducting experiments to measure separation
efficiencies for solids laden air and pressure drops for solids free air and comparing the measured
values with those estimated from well-known correlations. The separation efficiencies as predicted from the models of Leith and Licht (1972), Mothes and Loffler (1988) and Clift et al. (1991) are
n
found in agreement with the separation efficiencies obtained experimentally in the present
investigation. The pressure drops for dust free air obtained in the present work compares well with
those predicted from the correlation proposed by Fie (1987).
After the integrity is established, experimental runs have been planned to study the effect of
variations ofinlet air velocity (9 to 23 ms*1), solids feed rate (0.5 - 8g s"1) and particles size (163 to
460 u.m) on the holdup of solid particles, heat transfer coefficient and pressure drop for solids laden
air. In all one hundred experimental runs for 5 inlet air velocities, 5 solids feed rates and 4 particles
sizes have been conducted.
To gain an insight into the mechanism of heat transfer inside the cyclone, temperature profile
inside the cyclone has been measured. A study of temperature profile along with air and particles
flow profile inside the cyclone shows that the maximum temperature occurs at a certain radial
distance from the wall towards the vertical axis of the cyclone and a temperature gradient exists on
either sides of it. Air in the outer vortex moves co-currently with the solid particles from top to
bottom of the cyclone. Also radially inward flow of air bypasses air from outer vortex to inner
vortex. Thus, air transfers heat to solid particles confined to the wall as well as to the air in the inner
vortex through air bypass. In addition, variation of exit solids temperature, LMTD, thermal
effectiveness, heat transfer rate and heat transfer area with the operating parameters have also been
studied.
Heat transfer area in a cyclone is the total surface area of the particles inside the cyclone at a
given instant, i.e., holdup of particles inside the cyclone. Since the holdup is a function of operating
parameters, heat transfer area in a cyclone varies with operating parameters and particles size. This is
unlike in conventional heat exchangers where area of heat transfer is constant for a given heat
exchanger and is independent of operating parameters. Thus, the variation of the holdup of solid
particles with different operating parameters has also been studied. It has been found in the present
study that the holdup varies linearly with solids feed rate, increases with increasing inlet air velocity
in
and decreases slightly with increasing particles size. A correlation for the estimation of holdup has
been developed using regression analysis based on Marquardt Levenberg algorithm for the present
dataof solids to air loading ratio ranging from 0.05 to 0.87 and cyclone Froude number ranging from
82 to 552 as:
( Mh ^
VACADc/7sy
f-\( 2 \0M
-0.00096
m. VCi
vmJUDcy
The proposed correlation predicts the holdup of solid particles in the cyclone within an error
band of+15 % to -10 % for the present data and within an error band of+15 % to -25 % for the data
of other investigators. The proposed correlation seems to be better than existing correlations as the
correlation given by Yen et al. (1990) predicts the present holdup in a error band of+200% to -50%
and that given by Raju et al. (1994) within +75% to -40%.
Following Gelperin and Einstein (1971), the heat transfer resistance resides in the air film
surrounding the particles as the Biot number (< 0.1) for the present study is much less than 0.25.
Therefore, overall heat transfer coefficient is considered as film heat transfer coefficient surrounding
the particles. The average heat transfer coefficient between air and solid particles in the cyclone has
been computed based on heat gained by solid particles, co-current log mean temperature difference
based on inlet and outlet temperatures of air and solid particles and surface area of the solids holdup
inside the cyclone. The heat transfer coefficient is found to be almost independent of air velocity,
increases with increasing particles size and increases with increasing solids feed rate. The rate of
increase of heat transfer coefficient with solids feed rate reduces progressively with increasing solids
feed rate and approaches to almost zero at high values of solids feed rate. This leads to approaching
of heat transfer coefficient to an asymptotic value, which depends upon particles size. Following the
form of correlation proposed by Gnielinski (1981) for the prediction of heat transfer coefficient in the
packed bed, the proposed correlation has been obtained after regression analysis using present data of
IV
solids to air loading ratio from 0.05 to 0.87, Fedorov number from 5.63 to 17.3, Reynolds number
from 46.8 to 307, Prandtl number from 0.68 to 0.69 and Nusselt number from 0.16 to 2.72 as:
Nup =0.0047Fe145Fm0375(2 +0.664Re0/ Pr033)
The proposed correlation predicts the heat transfer coefficient for the composite data within
an error band of+25 % to -15 % and appears to be better than existing correlations as the correlation
proposed by Yen et al. (1990) predicts the present heat transfer coefficient in a error band of+100%
to +20% and that proposed by Raju et al. (1994) within +10% to -70%.
Computation of temperatures of air and solids at the cyclone outlets based on basic design
equation is computationally intensive as it involves iteration. Therefore, a method for the
computation of the outlet temperatures of air and the solids has also been proposed based on thermal
effectiveness determined on the lines of conventional 1-1 co-current heat exchanger using proposed
correlations for the heat transfer coefficient and the holdup, as given below:
o _ _J_ai___[_ae_ . _Ti _ -NTU(l+R)|
T.; - T.; 1+ R ai si
where, R=-^-^- and NTU= hA
ms Cps ma Cpa
The method predicts the outlet air temperature within an error band of+2.5% to -1.5% and
the outlet solids temperature within ±6%. The method seems to be useful for the practising engineers
because it is computationally less strenuous.
The pressure drop is a major performance parameter for a heat exchanger as it governs the
operating cost. The pressure drop across cyclone for solids laden air is studied in terms of ratio of
pressure drop for solids laden air to that of solids free air at same inlet air velocity. Within the range
of parameters investigated, the pressure drop ratio increases with increasing particles size but reduces
with increasing solids feed rate. The proposed correlation predicts the pressure drop ratio within an
error band of ±10 %for the present data of solids feed rate from 0.5 to 8 g s"1, ratio of particles size
to cyclone diameter from 0.00163 to 0.0046 and pressure drops ratio from 0 to 1.0.
AP< 0.34 + °66
AP
1 + 3.4E-7 p
vDcy
m!64
The proposed correlation for the pressure drops ratio appears to be better than existing
correlations given by Briggs (1946), Smolik et al. (1975) and Baskakov et al. (1988) for the data of
the present investigation.
Cyclone performance curves have been devised to help the plant engineers and the design
engineers in understanding the effect of operating parameters on the heat recovery. The cyclone
performance curves are plotted for different inlet air velocities between thermal effectiveness of air
and pressure drop ratio for different solids feed rates and particles sizes as parameters. Cyclone
performance curves suggest that to maximise heat recovery, solids to air loading ratio should be large
and to minimise pressure drop across cyclone, small size particles should be used