Sie sind auf Seite 1von 7

Food Control 12 (2001) 7783

www.elsevier.com/locate/foodcont

Steady state modelling and simulation of an indirect rotary dryer


rquez, Dio
Edgardo R. Canales *, Rodrigo M. Bo genes L. Melo
Department of Chemical Engineering, University of Concepci
on, P.O. Box 160 C, Correo 3, Concepci
on, Chile
Received 2 February 1998; received in revised form 20 June 2000; accepted 20 June 2000

Abstract
A steady state model for a sh meal indirect pilot plant rotary dryer with steam tubes is developed. The model is based on mass
and energy balances and heat transfer constitutive equations. Simulated and experimental results are presented. The moisture
condition of sh meal predicted by the model agrees satisfactorily with experimental values obtained in the pilot plant dryer within
the range of operating conditions explored. 2000 Elsevier Science Ltd. All rights reserved.

1. Introduction heating devices are below 170C (saturated steam at 6


bar gage as heating medium). In this way, sh meal
Over the years, Chile has consolidated its position in deterioration is avoided and its quality is improved.
the international market as one of the main shing The objective of the present work is to develop the
countries. A large part of the sh harvest is destined to mass and energy balances for an indirect pilot plant
the production of sh meal. During this process, great rotary dryer with steam tubes, as well as to quantify the
care is required since the material is submitted to severe mechanisms of heat and mass transfer, in order to ob-
thermal conditions that could induce a series of unfa- tain a steady state model for the dryer.
vorable deterioration reactions. Then the importance of
the drying operation in the quality of the product and in
the economic aspect of the elaboration of sh meal. 2. Steady state modelling
Conventional direct rotary dryers have been used in
sh meal factories, because heat and mass transfer rates The dryer of the pilot plant is of the indirect type; its
are accomplished very eectively. However, extended main characteristic is the use of both jacket and tubes as
exposure of the product to combustion gases at high heat transfer areas with high pressure saturated steam as
temperature (800C) induced a series of deterioration the heat source. The dryer is made of stainless steel, with
reactions that aected the nal quality of the meal. a total length of 3.050 (m), internal diameter of the
Adsorption of such toxic substances in the gases as NOx jacket of 0.497 (m) and 36 tubes of 0.019 (m) of diameter
can cause the formation of nitrosamines with cancerous each. See Fig. 1.
properties. Some countries are restricting the use of By means of steady state heat and mass balances
drying with combustion gases for certain nutritional within and element of the dryer, a mathematical model
products (Wilkes, Ponlsen & Green, 1991). is developed assuming that there are no radial temper-
Indirect rotary dryers with steam-heated surfaces ature and moisture proles.
(tubes, coil, jackets) at low temperature have been in- The steady state model of the indirect dryer is based
troduced in almost all sh meal plants in the last 10 on the following assumptions: solids and vapours ow
years. The drying process occurs in an atmosphere of through the dryer in plug ow. Thus the axial dis-
slightly superheated steam provided from the vapouri- placement velocity of the solid is constant. Solid parti-
sation of the water content of the wet solid particles, and cles are of uniform size. There are no heat losses to the
some little entrainment of air. Typical drying tempera- surroundings. There are negligible temperature gradi-
tures are about 9698C, and surface temperature of ents in the tubes and jacket walls. There is negligible
axial dispersion of solids in the dryer.
In order to simplify the modelling of the drying
*
Corresponding author. Tel.: +56-41-204534; fax: +56-41-247491. process, the operation is divided in two zones, as can be

0956-7135/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 9 5 6 - 7 1 3 5 ( 0 0 ) 0 0 0 2 7 - X
78 E.R. Canales et al. / Food Control 12 (2001) 7783

Nomenclature R rate of drying (kg water/kg meal s)


Aj area of heat transfer of the jacket (m2 ) Re Reynolds number
Asv supercial area of the solids by volume of S cross-sectional area of the dryer (m2 )
dryer (m2 /m3 ) s slope of the dryer (m/m)
At area of heat transfer of the tubes (m2 ) Tv temperature of the vapour (K)
CD drag coecient tR residence time of solids in the dryer (s)
cv heat capacity of vapours (J/kg K) Ts temperature of the solid (K)
cl heat capacity of the liquid water (J/kg K) Tw tubes and jacket temperature (K)
cs heat capacity of the solid (J/kg K) Vdryer volume of the dryer (m3 )
D dryer internal diameter (m) vv axial velocity of vapours (m/s)
dp particle diameter (mm) vs linear speed of solids (m/s)
F feed ow of sh press cake (kg/s) vt terminal velocity of particles (m/s)
Fs ow of dry solids (kg/s) W moisture of the sh meal (kg water/kg
f volumetric fraction of solids in the dryer dry solid)
hsv coecient of heat transfer vapoursolid Wf nal moisture of the sh meal (kg water/
(W/m2 K) kg dry solid)
hjv coecient of heat transfer jacketvapour wf nal moisture the sh meal (kg water/ kg
(W/m2 K) wet solid)
htv coecient of heat transfer tubesvapour Wi initial moisture of the sh press cake
(W/m2 K) (kg water/kg dry solid)
kv thermal conductivity of vapours (W/m K) wi initial moisture of the press cake (kg water/
L total length of dryer (m) kg wet press cake)
N rotation speed of the dryer (rpm) Y molar fraction of the steam in vapour phase
Nu Nusselt number (mol steam/mol vapour)
P ambient pressure of drying (atm) z local length of dryer (m)
p partial pressure of the steam in vapour phase DHvap enthalpy of vaporization of water at the
(atm) temperature of the solid (J/kg)
Pr Prandtl number lv viscosity of vapours (kg/m s)
psat saturation pressure of liquid water (atm) qv density of vapours (kg/m3 )
Psteam pressure of the saturated steam of tubes and qs density of solid particles (kg/m3 )
jacket (kgf /cm2 ) w sphericity of particles

seen in Fig. 2. In the rst zone occurs the heating of the The pressure of the generated vapour due to water
sh press cake, up to the temperature of drying. It is evaporation from the solids is considered homogeneous
assumed that during this stage there is no water evap- throughout the dryer, thus the temperature of the va-
oration from the solids, thus the moisture content of the pour is uniform along the dryer.
solids remains constant. In the second zone, the drying Since the dryer is open to the atmosphere at the load
of the sh press cake occurs. Here, water is evaporated and exhaust zones, there is a small quantity of air
at constant temperature. In this part of the dryer, the
water content of the solids reduces until it reaches the
required level.

Fig. 1. Schematic of the indirect rotatory pilot dryer. Fig. 2. Temperature and moisture proles along the dryer.
E.R. Canales et al. / Food Control 12 (2001) 7783 79

Fig. 3. Solids movement inside the dryer.

Fig. 4. Schematic of heat transfer in the dryer.


present in the interior of the dryer. Drying occurs in an
environment almost entirely composed of slightly su-
perheated steam at ambient pressure. dTs hsv  Asv  S
 Tv Ts :
dz Fs  cs cl  Wi
2.1. Movement of solids
The mass ow-rate of dry solids is determined from the
The axis of the dryer is inclined at a small angle with feed of press cake and its initial moisture,
respect to the horizontal. The particles advance through Fs F =1 Wi . The specic surface of solid particles is
the dryer in a series of solid falls and, between falls, determined from its diameter and the volumetric frac-
sliding at the bottom of the dryer, see Fig. 3. The solids tion of solids within the dryer, which is experimentally
fall is formed in the following way: the particles are measured
captured by longitudinal ns in the low part of the dryer
and carried to the upper part. The solids fall vertically 6
Asv  f:
from the edge of the n and descend through the vapour wdp
to the bottom of the dryer. In the fall the vapour may
drag the particles. The dryer is inclined with respect to From data published by B
orquez (1996), the sphericity
the horizontal, so the radial ascend and the vertical fall factor is correlated as
of the particles produces the advance of the solids along
the dryer. Marshall and Friedman (1949) proposed the 1
w :
following correlation for estimating the solids residence 1:101 414dp
time in a rotary dryer:
The heat balance is integrated from the initial solid
kL
tR temperature as boundary condition at z 0, and up to
N Ds
0:9
the length at which the solids reach the saturation
from which the axial displacement of solids is calculated temperature, as will be next explained, point at which
as vs L=tR . Constant k is determined experimentally drying begins. The boundary between the heating zone
for each particular dryer arrangement. and the drying zone is so determined.

2.2. Mass and energy balances 2.2.2. Zone II: Drying of the sh press cake
2.2.2.1. Energy balance in the solid (Parodi, 1992). When
Consider the schematic presented in Figs. 1, 2 and 4. stating the energy balance, it is considered that the va-
The model will be developed for the heating and drying pour gives sensible heat to the solids. It receives sensible
zones separately. heat from the tubes and jacket, and latent heat by
vapourisation of the water at the solids temperature, see
2.2.1. Zone I: Heating of solids Fig. 4. The heat transfer from the steam to the vapour is
In this zone, the mass balance is not needed, since accomplished in parallel form through tubes and jacket
there is no variation in the moisture of the solids. The walls, while the heat transferred to solids is accom-
energy balance establishes that heating of the solid along plished by means of the vapour only. Interactions be-
the dryer is caused by the rate of heat transfer from the tween solids and heating walls are neglected as heat
vapour transfer mechanism.
80 E.R. Canales et al. / Food Control 12 (2001) 7783

hsv  Asv  Vdryer  Tv Ts where the drag coecient CD is found from well-known
htv  At hjv  Aj  Tw Tv Fs  DHvap  Wi Wf : relationships.
Tubesvapour heat transfer coecient: Heat transfer
It is assumed that the wall temperature of the heating occurs between tubes that rotate around the axis of the
surfaces corresponds to the saturation temperature of dryer and vapours that move slowly along the dryer. For
the condensing steam within tubes and jacket, ow against circular cylinders, the heat transfer coe-
Psteam psat Tw , because the heat transfer coecient by cient is calculated from correlations presented by Bejan
steam condensation is very high as compared to the (1993). Characteristic length is the outside diameter of
tubesvapour heat transfer coecient. This energy bal- tubes, and characteristic velocity is the linear speed of
ance in dierential form also holds in Zone I, heating of rotation of tubes.
solids, but with the last term excluded. Heat transfer coecient between vapours and jacket
As mentioned before, drying occurs at ambient is calculated from the correlation of Seider and Tate
pressure in a vapour phase with a small fraction of (1936) for internal ow along a circular cylinder, and the
air. The supercial temperature reached by the solid characteristic parameters are the internal diameter of the
particles corresponds to the saturation temperature of dryer and the axial velocity of vapour calculated from
water vapourising at its partial pressure the ow of vapourised water

p P  Y psat Ts : Wi Wf
v v Fs :
qv S
(a) Heat transfer by conduction.
The heat transferred to solids by conduction is as-
sumed to be negligible as compared to convective 2.2.2.2. Mass balance in the solid. The decrease of solid
transfer, on the grounds that the surface area available moisture along the dryer is due to the rate of drying of
for conductive transfer, namely particle contact with the particles
tubes, ns, and jacket, and also particle-to-particle col- dW R
lisions, is small. Convective transfer is assumed to occur :
dz vs
over the entire surface area of solid particles. For a
This equation is solved beyond Zone I, from the initial
particle size of 1 mm, the estimated exposed surface area
solid moisture and up to the end of the dryer.
of all particles is about 75 m2 . Heating surface areas of
(a) Rate of drying. Period of constant drying rate
all tubes and jacket account for 11.7 m2 . Eective sur-
(Perry, 1988; Treybal, 1988).
face area for conductive heat transfer should even be less
The press cake that enters the dryer is previously
than this last gure.
cooked and pressed, presenting a brous structure with
(b) Heat transfer by convection
free water entrained that would permit an extended
Vapoursolids heat transfer coecient: The heat
period of constant rate of drying, that can be repre-
transfer coecient between an isothermal surface of a
sented as
falling sphere and a free isothermal ow of uid, as
proposed by Whitaker (1972), is used. hvs  Asv
R  Tv Ts :
qv  DHvap
Nu 2 0:4Re1=2 0:06Re2=3 Pr0:4 ;
Drying occurs at constant saturation temperature Ts ,
where and the rate of water vapourisation corresponds to the
rate of heat transfer from vapour to solid particles.
dp
Nu hsv ;
kv
q
Re dp vt v ; 3. Results and discussion
lv
lv
Pr cv : The simulation of the indirect rotary dryer is ac-
kv
complished for dierent operating conditions, a base
This correlation applies in a wide range of operative case is taken and the output variables (nal moisture of
conditions (3:5 < Re < 76; 000 and 0:71 < Pr < 380). sh meal, temperature of solids and vapours, heat
The terminal velocity of falling particles is found from transfer coecients) are plotted against the operation
the force balance between the apparent weight of par- variables (ow and moisture of sh press cake, particle
ticles and the drag force exerted by the vapour on its diameter, pressure of the saturated steam, speed of ro-
surface tation, slope of the dryer).
The simulation is carried out for the following con-
4 d p g q s qv ditions as base case: speed of rotation of dryer, N 38
v2t  ;
3 CD qv (rpm); initial moisture of sh press cake, Wi 0:48
E.R. Canales et al. / Food Control 12 (2001) 7783 81

Table 1
Comparison between experimental and simulated nal moistures in the drying of sh press cake

Run Experimental values Model wf (%)

F (kg/s) Psteam (kg/cm2 ) wi (%) wf (%)

A 0.02 5.0 49.0 17.5 11.0


B 0.03 3.0 48.0 32.0 26.0

(kg water/kg wet solid); molar fraction of steam in va- Even when the model could be corrected considering
pour phase, Y 0:95 (mol steam/mol vapour); slope of some of the aspects discussed above particle size dis-
the dryer, s 0:02 (m/m); volumetric fraction of solids tribution, fouling factors, dispersion mechanisms, and
in the dryer, f 0:02; particle diameter, dp 1 mm; average falling velocity of particles so to approximate
feed of press cake, F 0:023 kg=s; saturated steam predicted results to experimental values in Table 1, the
pressure, Psteam 3:5 kgf =cm2 . purpose of this work was to develop a simple model that
Upon comparing nal moistures predicted by the would help to explain the eect of dierent conditions of
model with experimental values obtained in the pilot operation on dryer responses (Cartes, 1996). These are
dryer, it is seen in Table 1 that observed moistures are next discussed.
higher than predicted ones. These dierences can be Fig. 5 shows the eect of feed rate of press cake that
partially explained because an important fraction of the enters the dryer on the nal moisture of the meal. The
material fed to the pilot dryer is in the form of lumps moisture trend is to increase with the increase of the ow
with a diameter that greatly exceeds the diameter of of press cake. As the ow of press cake becomes larger,
uniformly sized particles as considered in the model. the quantity of solids circulating in the dryer is aug-
This leads to a decrease in the overall rate of drying in mented, provoking an inecient drying. And a decrease
the actual process, because of the smaller supercial in the ow of press cake reduces the quantity of wet
exposed area, and also a longer water diusion path solids inside the dryer, producing a better drying. This
from the interior of the lumps. The results is a nal can be explained considering the drying capacity of the
product with greater water content in its interior, as was dryer, i.e., the heat transfer surface of the equipment is
experimentally proven. designed to evaporate a prescribed amount of water,
Additionally, fouling produced in tubes and jacket of and a higher feed of wet solids would result in a higher
the dryer by the wet material decrease the coecients of moisture content in the exit meal.
heat transfer between the heating surfaces and the va- Fig. 6 shows the eect of particle diameter on the
pour, and as a consequence in the rate of drying. The nal moisture of meal. A clear trend to reduce the
model did not consider fouling eects at this stage. moisture when the diameter decreases is observed, for
Discrepancies between model predictions and exper- example, for a diameter of 1 (mm), the moisture is 0.14
imental results may also be attributed to phenomeno- (kg water/kg dry solid), while for a particle diameter of
logical simplication of the model that overestimate the 2 (mm), the moisture reaches 0.23 (kg water/kg dry
rate of drying. First, the actual dryer certainly presents a
degree of axial mixing of solids as the particles fall from
the tip of the ns and slide along the dryer. This be-
haviour was observed during start-up and shut-down of
the equipment. Unfortunately, there is no available in-
formation for eective dispersion coecients of sh
meal in rotary dryers, and the dispersion mechanisms of
heat and mass transfer were not considered in the
model. As pointed out by Himmelblau and Bischo
(1968), the plug ow model predicts the best perfor-
mance, whereas dispersion reduces process eciency.
Secondly, it is assumed that particles fall at its terminal
velocity, while in fact they must accelerate their motion
from rest until they reach settling equilibrium, and in the
average their velocity of falling becomes less than that of
free sedimentation. As a consequence the vapoursolids
heat transfer coecient becomes somewhat overesti- Fig. 5. Simulated nal moisture of sh meal with respect to feed of sh
mated and so the rate of drying. press cake.
82 E.R. Canales et al. / Food Control 12 (2001) 7783

Fig. 6. Simulated nal moisture of sh meal with respect to particle


diameter. Fig. 7. Simulated nal moisture of sh meal with respect to steam gage
pressure.

solid), representing a variation of 64% in the quantity


of water content in the meal. The increase in particle drying remains almost constant. In spite of the fact that
diameter beyond 3 (mm) does not provoke a drastic temperatures of solids and vapour do not stay constant,
increase in moisture, and approaches an asymptotic and its temperature dierence increases mildly with the
value of 0.32 (kg water/kg dry solid). This is due to the increase of air (see Fig. 9), the vapoursolid coecient
fact that the exposed evaporating surface decreases of heat transfer decreases also mildly with the increase
with larger particle diameter. Calculation results of air content, and therefore the rate of drying stay
show that the supercial area of solids changes from 130 constant.
(m2 /m3 of dryer) for a particle diameter of 0.5 mm, to Fig. 8 presents the eect of the saturated steam
40 (m2 /m3 of dryer) for a particle diameter of 4 mm, and pressure in tubes and jacket on the exit temperature of
remains almost unchanged for larger particles. Further- solids and vapour. The temperature of solids is inde-
more, the vapoursolid heat transfer coecient also de- pendent of pressure of the saturated steam, and corre-
creases with particle size, from about 105 (W/m2 K) to a sponds to the saturation temperature at the system
constant value of 40 (W/m2 K), when the particle diam- pressure. On the other hand, the temperature of the
eter increases to 4 mm. As a consequence, the reduction vapour is established from a heat balance between the
of transferred heat originates an augmented moisture of heating surfaces and the hot solids. The model predicts
nal product as the particle size becomes larger. well the behaviour of this variable, since upon increasing
The eect of pressure of the saturated steam through the pressure of the saturated steam the temperature of
tubes and jacket on the nal moisture of meal is pre- vapours increases.
sented in Fig. 7. As expected, an increase in the pressure
of the saturated steam decreases the moisture of the
meal. This increase in pressure of the steam rises the
temperature of tubes and jacket, enlarges the tempera-
ture dierence available for heat transfer and generates a
boost in the rate of drying. For example, for a pressure
of the saturated steam of 3.0 (kgf /cm2 ), the meal reaches
a nal moisture of 0.17 (kg water/kg dry solid), while for
a pressure of the saturated steam of 4.0(kgf /cm2 ), a
moisture of 0.12 (kg water/kg dry solid) is reached, rep-
resenting a decrease of water content of 29% in the meal.
The heat transfer coecient between tubes and va-
pours ranges from 23 to 30 W/m2 K, and is much less than
the heat transfer coecient of condensing steam inside
the tubes typical values are 300010,000 W/m2 K so
the assumption that heating surfaces are at the saturation
temperature of the steam becomes very realistic.
The eect of air content in the vapour on the moisture Fig. 8. Simulated solids and vapour temperature with respect to steam
of meal is not meaningful, due to the fact that the rate of pressure.
E.R. Canales et al. / Food Control 12 (2001) 7783 83

principles such as plug ow of both solid particles and


water vapours, and drying occurring at constant rate.
The model neglects axial dispersion of solids, tempera-
ture and moisture gradients within the solid particles,
and particle size distribution. Steady state heat and mass
balances are written rst for a heating zone of solids,
and next for a drying zone.
Predicted values of nal moisture of sh meal given
by the model agree fairly well with measured experi-
mental values, but observed nal moistures are higher
than model predictions. The discrepancies are attributed
to the simplifying assumptions, and a corrected model
should include more realistic phenomena, namely: size
distribution of solids, fouling resistance to heat transfer
in heating surfaces, unsteady settling velocity of falling
particles, and increased solidvapour heat transfer co-
Fig. 9. Simulated solids and vapour temperature with respect to the ecient for particles of irregular shape.
molar fraction of steam in the vapour.
The model proves to be a useful tool to predict dryer
behaviour. Final moisture of sh meal is the main re-
Fig. 9 shows the behaviour of the temperature of sponse of interest, and decreases strongly with smaller
solids and vapours with respect to the molar fraction of particle diameter, with lower feed of press cake to the
steam in the vapour. The temperature that reaches the dryer, and with higher temperature of the heating me-
solids, as was already mentioned, corresponds to the dium.
saturation temperature of water at its partial pressure, The model can even be improved if axial dispersion of
and only depends of the quantity of air present in the solids is taken into account. This parameter must be
vapour. It can be observed in Fig. 9 that as the solids are determined performing appropriate experiments in the
inmersed in an atmosphere thoroughly of steam, their pilot dryer.
temperature approaches the boiling temperature. The
temperature of vapours also increases with an increase
in the steam content, because of the increase of the co- References
ecients of heat transfer, generated by higher values of
the physical properties. Bejan, A. (1993). Heat transfer (1st. ed.). New York: Wiley.
Also from Fig. 9 it can be said that the model predicts B
orquez, R. (1996). Modeling of protein sh nutrient retention during
vapour temperatures which are higher to measured values drying and storage. Ph.D. Thesis, University of Concepci on.
Cartes, H. (1996). Steady state modeling and simulation of a pilot
observed in this type of dryers. This discrepancy may be plant indirect rotary dryer. Chemical Engineer Thesis, University
explained as follows: the heat transfer coecient has been of Concepci on.
estimated from correlations for ow around a sphere, but Himmelblau, D., & Bischo, R. (1968). Process analysis and simulation.
for particles of irregular shape, the heat transfer coe- Deterministic systems. New York: Wiley.
cient increases by a factor of 1.62.3 due to a higher Marshall, Jr., & Friedman, S. (1949). Chemical Engineering Progress,
45, 482573.
degree of turbulence around the particle (VDI-Warme- Parodi, C. (1992). Steady state modelling of an indirect rotary dryer.
atlas, 1991). Consequently, the temperature of vapours Chemical Engineer Thesis, University of Santiago.
should drop in order to maintain the heat balance. Perry, R. (1988). Chemical engineers' handbook (5th. ed.). Mexico: Mc
As an additional result, the model predicts a mild Graw-Hill, 5(20), 1015.
decrease of temperature of vapours with the increase in Treybal, R. (1988). Mass transfer operations. Mexico: Mc Graw-Hill.
VDI-Warmeatlas (1991). VDI-Verlag, Sechste Auage, Gh1 1, Dus-
the ow of press cake, explained by the increase of the seldorf.
quantity of solids circulating inside the dryer. Whitaker, S. (1972). AIChE Journal, (18), 361371.
Wilkes, K., Ponlsen, K. P., & Green, E., (1991). Direct drying of foods
with natural gas as fuel. Institution of Chemical Engineers, Trans. I.
4. Conclusions Chem. E., 69(C), 182188.

A simple mathematical model of a pilot plant indirect


rotary dryer has been developed, based on fundamental