You are on page 1of 17

DEVELOPMENT OF A NOVEL VACUUM PYROLYSIS REACTOR WITH

IMPROVED HEAT TRANSFER POTENTIAL


Vacuum pyrolysis- Improved heat transfer

C. ROY, J. YANG, D. BLANCHETTE, L. KORVING and B. DE CAUMIA


Institut Pyrovac Inc.
1560, avenue du Pare Beauvoir, Sillery (Quebec) Canada GIK 7P4

Abstract
A novel reactor has been developed in our laboratory which addresses the heat
transfer limitations usually encountered in vacuum pyrolysis technology.
Conventional pyrolysis reactors such as multiple hearth furnaces, rotary kilns and
screw type reactors exhibit overall heat transfer coefficients ranging from I 0 to 60
W·m- 2K 1, depending primarily on the type of feedstock treated. The new reactor
design includes a novel feedstock transport and agitation system which produces a
forced exchange between the feedstock particles heated at the surface of the heating
plate and the colder particles located at the core of the packed particle bed. Thus,
the heat transfer between the reactor and the pyrolyzed material is dramatically
increased. The other novelty of this reactor is the use of an indirect heating system
involving commercial molten salts (Hitec®). Experimental and theoretical studies
have been undertaken in order to tentatively correlate the reactor design parameters
and the heat transfer coefficient. A new heat transfer model, based on Schliinder's
heat transfer model, has been developed to model the heat transfer in the bed of
particles as a function of the mechanical movement of particles created by the
agitation. This model is validated by comparing theoretically calculated overall heat
transfer coefficients in a batch reactor with experimentally measured values for gravel
feedstock in the same reactor. The model is then used to predict overall heat transfer
coefficients for various feedstocks in the novel continuous reactor. Coefficients
ranging from 70 to 250 W·m- 2K 1 are obtained with this new system.
Keywords: Agitation, heat, transfer, model, Hitec®, packed, particles, surface,
vacuum, pyrolysis.

1 Introduction

Over the last ten years, a vacuum pyrolysis process has been developed at Universite
Laval in Quebec, Canada. This thermal decomposition process enables a large
variety of solid and semi-liquid wastes to be transformed into useful products.
Vacuum pyrolysis is typically carried out at a temperature of 400-500°C and a total
pressure of 2-20 kPa [1]. These vacuum conditions allow the pyrolysis products to

351
A. V. Bridgwater et al. (eds.), Developments in Thermochemical Biomass Conversion
© Springer Science+Business Media Dordrecht 1997
be rapidly withdrawn from the hot reaction chamber, thus preserving the primary
fragments originating from the thermal decomposition.
A major limitation of vacuum pyrolysis technology is heat transfer. Previous
studies have shown that the rate of heat transfer is essentially the rate limiting step for
pyrolysis reactions [2]. Conventional pyrolysis reactors such as multiple hearth
furnaces, rotary kilns and screw type reactors exhibit overall heat transfer coefficients
ranging from 10 to 60 W·m· 2K 1 [3], depending on the type of feedstock treated. The
low thermal conductivity of the feedstock materials partially explains why the heat
transfer fluxes are low.
The multiple hearth furnace which has been used in the previous studies of this
laboratory [4] can be used to provide an example of the factors which limit the heat
transfer in this type of reactor. Since the heat source in the multiple hearth furnace is
the external wall, the heat transfer is primarily effected by radiation. The plates on
which the feedstock is placed inside the reactor are heated by radiation from the
external wall and thus heat the feedstock from below. The feedstock is also directly
exposed to the radiant heat source (the external wall) from above, as there is some
distance between the different plates inside the reactor. One limitation of this
radiative heat transfer mechanism is the view factor between the external wall and the
plates inside the reactor. This factor is only slightly greater than 0.5 at the point
where the plate meets the wall (assuming that the plate is a black body), and
decreases considerably as one moves away from the external wall and approaches the
middle of the plate in the center of the reactor, especially when the distance between
neighbouring plates is small. Another limitation arises from the type of agitation
system used in this reactor. Each plate inside the multiple hearth furnace is equipped
with a series of radial metal bars which transport the feedstock as they move over the
plates. The feedstock particles tend to accumulate along these bars, thus leaving
almost 50% of the heated plates uncovered by feedstock and rendering the heat
transfer inefficient, especially since there is very little mixing of the hot and cold
particles. Finally, since the plates are only heated by radiation, the heat transfer
coefficient is limited both by the unfavourable view factor and the minimal
conduction between the heated plates and the feedstock. Both of these factors could
be greatly improved, especially for small feedstock particles, by surface-to-particle
contact conduction and radiation.
The new vacuum pyrolysis reactor, described herein, uses an indirect heating
system involving commercial Hitec® molten salts which flow inside the heating
plates, thus heating the feedstock on these plates both by conduction and radiation.
In addition, the novel transport system in the reactor efficiently agitates the
feedstock, thus greatly improving the heat transfer potential.
The importance of the influence of agitation on heat transfer has been investigated
widely in previous studies [5-8]. These studies mainly dealt with the following two
aspects: the contact heat transfer resistance between the heating surface and the
feedstock particles, and the heat transfer in the feedstock as a function of the flowing
dynamics of the particles. An important contribution was made by Schliinder [7]
who developed a theory to quantitatively describe the surface contact heat transfer
between a spherical particle and a heating surface. This theory was later further
developed by Malhotra and Mujumdar [9] to describe the surface contact heat

352
transfer of other particle shapes. The particle flowing dynamics have been studied by
several authors [10-11]. However, there are still no general equations which
correlate the particle movement with the actual design of the agitation system. The
new heat transfer model proposed herein assumes that the heat transfer in a moving
bed is only a function of the heat exchange at the heating surface which depends on
the surface-to-particle contact resistance and the renewal of the heated bottom layer,
as described in the theory section below. The influence of particle mixing in the bulk
of the particle bed is ignored as agitation renders the interparticle heat transfer vety
efficient and therefore this heat transfer resistance component is negligible.

2 The new vacuum pyrolysis reactor

In our laboratory a semi-continuous horizontal pilot plant reactor with a length of 3


m, a diameter of 0.6 m and a capacity of about 100 - 400 kg· h-t, depending on the
feedstock treated, has been built (see Figure 1). The moving bed vacuum pyrolysis
reactor uses a novel transport and agitation device which produces a forced exchange
between the feedstock particles heated at the surface of the heating plate and the
colder particles located at the core of the packed bed of particles. As a result, the
heat transfer between the reactor and the pyrolyzed material is dramatically increased.
The patents for the transport system have been claimed and hence further details
cannot be given for the moment (12). In the reactor, the feedstock is conveyed over
two horizontal heating plates, one on top of the other. The plates are 35 em wide
and together they have an effective length of 4.1 m. They are composed of
rectangular tubes which are welded together in the longitudinal direction.
Commercial molten salts (Hitec®) flow countercurrently with the feedstock through
the interior of these tubes, thus indirectly heating the feedstock which is conveyed
over the heating plates. The salts leaving the reactor are collected in a tank which is
equipped with a vertical pump to circulate the molten salts through the system, to a
heater to be reheated and then back to the reactor. The Hitec® molten salt is an
eutectic mixture of potassium nitrate, sodium nitrite and sodium nitrate which can be
used as a heat transfer medium between 150°C and 540°C. Below 150°C lies the
melting point of the mixture (at 142°C) and above 540°C the salt will degrade too
rapidly for commercial applications. The vapour pressure of the salt is negligible in
the operating range. At 500°C, the Hitec® salts have a density of 1700 kg·m-3 , a
specific heat capacity of 1549 J-kg- 1K 1 and a thermal conductivity of0.61 W·m-1K 1•
During pyrolysis, the feedstock is heated under vacuum to a temperature of about
450-500°C. The pyrolysis vapours are rapidly evacuated from the hot reactor
chamber by means of a vacuum pump which maintains a total pressure of 2-20 kPa in
the reactor. The vapours are condensed in two spray towers. After pyrolysis, the
solid residues leave the reactor under vacuum and are cooled with water in a
quenching vessel. The sludge formed is directed towards a screen where the char is
sieved and separated from the water. The cooling water is then recycled to the
wetting vessel. The non-condensable gases can be burned to provide a significant
portion of the energy necessary to heat the molten salts. The amount of heat which

353
can be provided by burning the non condensable gases will depend on the nature of
the feedstock. In applications involving plastics for example, three times more heat is
available in the pyrolysis gas for each kg of feedstock treated than the actual amount
required to achieve the pyrolysis reactions. For automobile shredder residues or auto
fluff, this ratio is approximately two. Another example is contaminated soil
containing 15% moisture where the "pyrolysis gas heat content per kg of feedstock I
heat of pyrolysis per kg offeedstock" ratio is approximately 0.6. An induction heater
can also be used as a make-up heat source to enable the molten salt to reach the exact
set point temperature in the reactor.
In order to determine the best design and operation parameters for the new
reactor, theoretical and experimental work has been carried out to investigate the
correlation between the heat transfer coefficient and the design of the agitation
system. The proposed new heat transfer model was validated using a smaller batch
scale agitated bed reactor.

feedstock to scrubbers &


vacuum pump

heating pyrolysis
bed
plate vapours &
gases

residues transport &


L __ _• to cooling unit agitation
system
flow of
molten salts
salt
holding
tank

electrical heating
Figure 1. Pilot plant vacuum pyrolysis reactor

3 Theory

3.1 The surface renewal model


The proposed surface renewal model is based on the heat transfer model which
Schli.inder [13] developed for the indirect contact vacuum drying of solid particles.

354
In Schlunder' s model, particle movement due to the agitation was described as a
series of time periods during which the particle bed is assumed to be static, followed
by a rapid mixing of the particles. Thus the average heat transfer coefficient during
the static period can be calculated using the Penetration Theory, which results in the
following equation:

Uws
(I)

In the equation above, CXws is the surface-to-particle contact heat transfer coefficient,
abed is the overall heat transfer coefficient in the agitated bed, p, Cp, and "A are the
average density, specific heat capacity and the thermal conductivity of the particle
bed and 'l:R is the time period of the static period. In order to relate the time scale of
the mixing device to the time scale of the time periods of the static bed, Schliinder
introduced the following relationship into the formula above:

(2)

Together these formulas yield the following formula:

(3)

where Ntherm is given by:

2
N = Uwstmir
(4)
tlwrm
pCp A

In these formulas tmix is the characteristic mtxmg time of the agitation device
(normally the time necessary for one tum of the device). Nmix is defined as a "mixing
number" whose value is a function of the geometrical and mechanical properties of
the agitation system, i.e. the agitation design parameters. This parameter must be
obtained from curve fitting of the measured abed as a function of the characteristic
time of agitation, lmix· Schliinder could not give a physical explanation for the value
of this Nmix·
In order to improve Schlunder's model and to find the direct correlation between
heat transfer and the mechanical properties of an agitation system, the heat transfer
mechanism in an agitated bed has been analyzed. As shown in Figure 2, two heat
exchange phenomena occur in an agitated bed: the wall-to-particle heat exchange at
the heating surface and the interparticle heat exchange in the bulk. Particle agitation
improves the heat transfer in the following two ways: (i) the hot particles are

355
periodically removed from the heating surface, thus maintaining a large temperature
difference between the particle bed and the heating surface and (ii) the hot particles
are directed towards the bulk of bed, resulting in a solid convection which greatly
improves the interparticle heat exchange. If the particle bed is not very thick, perfect
mixing of the hot particles in the bulk should be obtained and the heat transfer
resistance is considerably reduced. In this case, the heat transfer in an agitated bed
can be simplified as a periodical wall-to-bed contact heat transfer problem, which is a
function of the wall-to-bed contact heat transfer coefficient, the period of agitation
and the quantity of particles removed after each agitation. The resulting new heat
transfer model is based on the following assumptions:

1. The characteristic time of the interparticle heat transfer is shorter than the time
required for the agitated hot particle to return from the bulk to the heating surface.
2. The overall heat transfer in the bed of particles is determined by the heat transfer
between the bottom layer of particles and the heated plate.
3. Each agitation removes p percent of the bottom layer particles from the heating
surface; Pis defined as the bottom layer renewal efficiency.
4. The agitation results in a distribution of the residence time of the particles at the
heating surface. The fraction of particles with a residence time ktmix is PO-Pt 1;
tmix is the time that is necessary to complete one agitation.
5. During the residence time of a specific fraction, it is assumed that this fraction
remains static (see Figure 2.a). Thus the heat transfer coefficient of this fraction
can be calculated using the Penetration Theory.
6. The overall heat transfer coefficient is calculated by summation of the heat transfer
coefficients of each fraction.

interparticle
heat exchange

wall-to-bed
contact heat
transfer

a) during the static period b) during the instantaneous


0 < t < tmi.x mixing, t =tmi.x

Figure 2. Heat transfer mechanism in an agitated bed of particles

These assumptions result in the following equation:

356
00

abed= LBkak (5)


k~l

where abed is the overall heat transfer coefficient in the bed, and Bk and ak are the
quantity and the heat transfer coefficient of fraction k. The variables, Bk and ak, are
calculated by the following equations:

(6)

(7)

In this equation, p, Cp and A are physical properties determined by the nature of the
feedstock. Thus, the only parameters which depend on the reactor design are the
surface-to-particle contact heat transfer coefficient, aw,, and the bottom layer renewal
efficiency, p. As p is directly influenced by the reactor design, this formulation
allows a relationship to be established between the reactor design and the overall heat
transfer in the particle bed.
In comparison with Schliinder's model, the major advantage of this formulation is
that the parameter Nmix, which can only be obtained by curve-fitting, has been
replaced by the experimentally measurable parameter p. The bottom layer renewal
efficiency, p, has been defined as the particle fraction removed after each agitation.
Thus, Phas a direct physical meaning and can be easily obtained by a simple cold
mechanical run using tracers.

3.2 Surface-to-particle contact resistance (aws)


The so-called surface-to-particle contact heat transfer is defined as the heat transfer
from the heating surface to the bottom layer of the particles. It consists of three
components: the thermal conduction through the gas wedge between the particles
and the heating surface (awp), the thermal conduction through the void space between
the particles (awg) and the heat transfer by radiation (ar). Therefore the contact heat
transfer coefficient (Uws) can be expressed by:

<Xws = <Xwp +<Xwg + <Xr (8)

According to Schliinder' s theory, a minimal thermal contact resistance exists between


a particle and a contacting surface, which is symbolized by Uwp· A formula for the
calculation of Uwp has been developed for spherical particles [13]. It is expressed as:

<Xwp =~A ·4A,g


-
dp
l{2(cr +Sr) + l } ln{
dp 2
( dp
cr +S,)
}- l l (9)

357
where <!>A is the area fraction of contact points. The heat transfer through the voids
of gas (awg) can be formulated as:

(10)

The radiation between a particle and the heating surface (a,) is expressed as [14]:

0.2268 ( T ) 3 (11)
ar = ~-0.264 100
E

Since the equations above were developed for spherical particles, they are still not
available to determine the surface-to-particle contact heat transfer coefficient in this
work where the materials treated often have an irregular shape. The importance of
these equations is that they indicate which parameters have a significant influence on
the surface-to-particle heat transfer resistance. In this work, the surface-to-particle
contact resistance is determined experimentally. The surface-to-particle heat transfer
coefficient for irregular shaped particles is currently being further studied in our
laboratory.

4 Experimental

4.1 Equipment
A batch scale reactor was developed and used as a tool to validate the new heat
transfer model at a total pressure slightly lower than atmospheric pressure. The
transport and agitation devices of both the laboratory and full scale reactors are
equivalent. The heat transfer tests were carried out in order to determine the effect
of the agitation on the rate of heat transfer. Once this influence is known, the results
obtained at near-atmospheric-pressure can be corrected for vacuum conditions. The
vacuum will only influence the thermal parameters, i.e. the surface-to-particle contact
resistance, aws, and the conductivity in the bed Abed, and will not affect the agitation
mechanism. Thus, the heat transfer relationship will remain the same even if the
reactor pressure changes. So, the values of aw. and Abed can be measured under
vacuum and entered into the developed formulas so as to determine the overall heat
transfer coefficient under the vacuum conditions found in a continuous feed reactor.
The batch reactor, shown in Figure 3, consists of a well insulated tank which
contains molten salt and is equipped with heating elements in order to be able to heat
the salt to the required temperature. A second insulated tank can be placed in the
bath of molten salts. The feedstock enters the reactor through the feedpipe. The
agitation mechanism transports and agitates the feedstock in a circular direction. The
center of the reservoir is kept free of feedstock by a scraping mechanism. The
diameter of the feedstock tank is 107 em and the effective heat transfer area is 0.82

358
m2 • The total mass of salt in the salt-tank is 285 kg. During a pyrolysis run, the
vapours formed are evacuated by a blower. The air tightness of the system and the
capacity of the pump are able to maintain a total pressure slightly lower than ambient
atmosphere which is sufficient to study all relevant heat transfer phenomena. For the
tests described in this paper the blower was not used.

suction of feedpipe
pyrolysis
vapours/ gases
insulation trapdoor

electrical circular transport


molten salts & agitation device
heaters
Figure 3. Batch vacuum pyrolysis reactor

Once the surface-to-particle contact resistance, CXws. had been measured, the batch
reactor was used to measure the bottom layer renewal efficiency, p for the circular
agitation device and finally the overall heat transfer coefficient in the particle bed, abed
(see section 4.2). The semi-continuous reactor was used to measure the bottom layer
renewal efficiency, p, for the novel transport and agitation device (see section 4.2.2).
The measured value of p was then used to predict the overall heat transfer coefficient
in the new reactor, abed (see section 5.4).

4.2 Methodology

4.2.1 Determination of the surface-to-particle contact resistance ( CXws)

The surface-to-particle contact heat transfer coefficient, CXws, is measured with the
apparatus shown in Figure 4. This apparatus consists of a stainless steel heating plate
with a diameter of 400 mm, the particles to be heated, and a glass wool insulation
plate with a thickness of 30 mm. The temperature of the heating plate, Tw, is
increased from 30 to 200°C. The particles have an irregular shape. Their
approximate dimension is 15 mm x 10 mm x 4 mm with a mean volume diameter dp
of 5 mm. The average of the maximum particle diameter and the volume average

359
diameter is 10 mm. A thermocouple is placed in the center of the particle to measure
the particle temperature, Tr

Ti
2~ Tp

3 ~ Tw
Figure 4. Measurement ofthe surface-to-particle contact heat transfer coefficient.
I. heating plate, 2. particles, 3. insulation material.

A computer data acquisition system is connected to record the temperatures, Tw,


1i and Tp. As the radial dimension of the particle bed is much larger than the axial
dimension, the heat transfer in the radial direction can be ignored. Thus, an energy
balance equation is developed for one single particle, which gives:

(12)

where rn, Cp, p and A are the mass, specific heat capacity, density, thermal
conductivity and surface area of the particle. The temperature of the heating plate,
the particles and the insulation plate are represented by T,., 1i and Tp, respectively.

4.2.2 Determination of the bottom layer renewal efficiency CP)

Tests with tracers have been carried out in order to determine the bottom layer
renewal efficiency for the transport and agitation system in the batch reactor and in
the semi-continuous pilot plant reactor. For both reactors the test-method was the
following. For a small part of the bed the bottom layer of the bed was replaced by a
monolayer of coloured particles. To determine p, the length of a transport unit was
passed over the bed. For the batch reactor this was equivalent to one tum of the
agitation and transport device; for the continuous system this would be a fraction of a
tum depending on the number of transport units on the chain. Then several samples
of the bottom layer were taken at various places in the bed. From these samples the
total mass of coloured particles remaining in the bottom layer was calculated and
divided by the original mass of coloured particles to give the fraction p.
For the batch reactor, P was determined for 10 mm gravel particles at seven
different turning speeds (1, 3, 5, 10, 19, 26, 38 rpm). The tests with the transport
and agitation device of the semi-continuous reactor were performed using gravel with
different particle sizes (5, 10 and 14 mm diameter) at five different speeds of the
transport and agitation device (0.10 m·s-t, 0.11 m·s-t, 0.16 m·s-t, 0.21 m·f 1 and 0.28
m-s- 1). For each test the bed height was 2.5 em. More research is being carried out
to determine the influence of the design of the agitation device and the mechanical
feedstock properties on the Pvalue [15].

360
4.2.3 Determination of the overall heat transfer coefficient in the agitated bed, abed

Using the batch reactor, four heat transfer tests were performed with gravel particles
with a diameter of 10 mm. The conditions for each of these tests are listed in Table 1
and the thermal and physical properties of the feedstock are given in Table 2.

Table 1. Operating conditions for runs performed with the batch reactor

Test Feedstock Turning speed (rpm) Load (kg)


G1 gravel 10 55.0
G2 gravel 19 55.3
G3 gravel 26 55.3
G3 gravel 38 53.5

About 15 minutes before the test, the heating of the salt (in the outer tank) was
stopped. Immediately before the beginning of the test, the agitation mechanism was
started and the feedstock was placed in the feedpipe. The test was then started by
opening the trapdoor to allow the feedstock to fall into the inner feedstock tank. At
that moment the temperature of the salt was approximately 520°C for each test.
Several thermocouples were placed in the salt directly on the heat transfer surface
and at different distances from this surface. During the tests both the temperature of
the feedstock and the salt were recorded as a function of time.

Table 2. Physical and thermal properties of the feedstock

Physical I Thermal Property Gravel Gravel Gravel


5mm 10mm 14mm
Average particle size* (mm) 4.9 9.2 14.7
Apparent density (kg·m- 3) 1260 1190 1100
True density (kg·m- 3) 2590 2640 2664
Specific heat capacity (Hg" 1K 1) 900 900 900
Thermal conductivity (W·m- 1K 1) 0.2 0.2 0.2
* Average of the maximum particle diameter and the volume average diameter

5 Results and Discussion

This section will describe how the new heat transfer model, based on the Schliinder
heat transfer model, was validated. First, the results of the measured surface-to-
particle contact resistance, CX:ws. and bottom layer renewal efficiency, p, will be
described. This will be followed by a description of how these two values were used
to predict and then experimentally validate the value of the overall heat transfer
coefficient, abed in the batch reactor using gravel feedstock. Finally, the measured
values of Pwill be used to predict the overall abed in the semi-continuous reactor for
various feedstocks.

361
5.1 Surface-to-particle contact resistance (aws)
The surface-to-particle contact heat transfer coefficients of gravel have been
measured. The measurements were carried out under atmospheric pressure. When
the measured temperatures Tw, T; and Tp are introduced into Equation 12, the contact
heat transfer coefficient ctws is obtained. The llws of the gravel is 250 W·m- 2K. The
contact heat transfer does not depend on the material, but only depends on the
particle dimension and shape, as well as on the fluid property between the heating
surface and the heated particles (see Equation 9).

5.2 Bottom layer renewal efficiency(p)


Intuitively, it could be foreseen that the bottom layer renewal efficiency, p, would be
a function of the moving speed of the transport unit, the design of the transport unit,
the bed height and the physical properties of the feedstock. The values of pthat were
measured in the batch and semi-continuous reactor are summarized in Tables 3 and 4.

Table 3. Bottom layer renewal efficiencies for the circular agitation device in the
batch reactor

Turning speed Characteristic time, Bottom layer renewal


(rpm) tmix (s) efficiency, P
1 46.2 0.67
3 19.4 0.72
5 11.8 0.73
10 6.1 0.56
19 3.1 0.51
26 2.0 0.46
38 1.6 0.47

Table 4. Bottom layer renewal efficiencies for the transport and agitation system of
the semi-continuous reactor

Moving speed Bottom layer renewal efficiency, p


(m·s- 1) Gravel, 5 mm Gravel, 10 mm Gravel, 14 mm
0.10 0.72 0.62 0.47
0.11 0.66 0.63 0.44
0.16 0.67 0.66 0.59
0.21 0.74 0.61 0.55
0.28 0.69 0.51 0.50

The value of the bottom layer renewal efficiency should ideally be as close to 1.0
as possible, for example a value of 0.9. However, a value of p greater than 0.5 is
very satisfactory, since not much will be gained if the value of p is increased from 0.5
to 0.9. It can be shown that this increase will improve at best the rate of heat transfer
by a factor of 1.3. However, the value of p should not be too low, since this will
greatly affect the rate of heat transfer. It can be shown that a decrease of p from 0.5

362
to 0.1 will decrease the rate of heat transfer by a factor of 2.2. The results in Table 3
indicate that values of Pgreater than 0. 7 could be obtained in the batch reactor at low
turning speeds. The results also suggest that an increase in turning speed leads to a
decrease in p. The bottom layer renewal efficiency is significantly influenced by the
particle size, as indicated by the study carried out with the semi-continuous reactor
(Table 4).

5.3 Overall heat transfer coefficient in the batch reactor («bed)


In order to validate the heat transfer model which correlates the heat transfer
coefficient in an agitated bed, «bed, with the surface-to-particle contact heat transfer
coefficient, aw,., and the bottom layer renewal efficiency, p, heat transfer coefficients
in the batch type agitation reactor have been measured with gravel feedstock and
compared with the predicted values. Table 5 presents the experimental and
theoretical results.
The theoretical results were calculated using Equation 5 and the thermal
properties p, Cp and A. listed in Table 2. The measured surface-to-particle heat
transfer coefficient, O:ws (250 W m- 1 K 1), and the bottom layer renewal efficiency, p
(see Table 3), were used for the prediction. The value of p only changes slightly with
the agitation speed.

Table 5. Measured and calculated overall heat transfer coefficients in the batch
reactor for various feedstocks

Test Feedstock Turning Heat transfer coefficient, (W·m- 2K- 1)


abed
speed (rpm) Calculated Measured
Gl gravel 10 95 81
G2 gravel 19 112 117
G3 gravel 26 119 124
G4 gravel 38 132 156

The results in Table 5 indicate that the experimental measurements and the model
predictions are in good agreement. Even though the model seems to overpredict the
heat transfer coefficient at low turning speeds and underpredict this coefficient at
high turning speeds, the difference between the experimental and theoretical values is
less than 21%. This difference may be due to the fact that the model ignores the
interparticle heat exchange (it was assumed that the agitation device renders this heat
transfer resistance component negligible).

5.4 Predicted overall heat transfer coefficient in the semi-continuous reactor


The most important characteristic of the novel vacuum pyrolysis reactor is the
significant increase in heat transfer between the reactor and the feedstock. The new
heat transfer model proposed in this work enables a quantitative prediction of the
overall heat transfer coefficient as a function of the design and operation parameters.
As shown in Equations 5-7, the most important design parameters are aw,., p and the
agitation speed. When the design and operation parameters are fixed, the heat

363
transfer coefficient will be a function of the thermal properties of the feedstock. Table
6 provides the thermal properties of various pyrolysis feedstocks and Table 7
presents the predicted heat transfer coefficient in the new pyrolysis reactor for
different feedstocks. These values do not take into account the radiation effect above
the bed.

Table 6. Feedstock physical and thermal properties used to predict overall heat
transfer coefficients

Feedstock Density Heat capacity Thermal conductivity pCpA. (·10 4 )


(kg·m- 3) (Hg-'K') (Wm- 1K 1)
Gravel 1360 700 0.2 31.9
Charcoal 230 1000 O.D7 1.6
Wood 240 1450 0.07 2.4
Rubber 600 1600 0.1 14.4

Table 7. Predicted overall heat transfer coefficients for various feedstocks

Feedstock Predicted heat transfer coefficients (W·m- 2K 1) for varying


surface-to-particle contact resistance values
CXws= 250 CXws= 500 CXws= 750 CXws= 1000
W·m- 2K 1 W·m- 2K 1 W·m- 2K 1 W·m- 2K 1
Gravel 143 200 232 252
Charcoal 70 82 87 90
Wood 81 97 104 107
Rubber 122 161 181 193

The values of P and lmix used in the prediction were 0.7 and 1.7 second,
respectively. As the tracer-tests show (see Table 3), P= 0.7 is a reasonable value for
the agitation system. The moving speeds of the transport and agitation system of the
semi-continuous reactor can reach 0.3 m-{ 1, a reasonable speed for full scale
applications. With the current configuration of the mechanism this will give a
characteristic mixing time of approximately 5 seconds. For larger scale reactors, the
design of the transport and agitation system can easily be optimized to provide three
times as much agitation, resulting in a mixing time of 1.7 s.
It is difficult to select one value of the surface-to-particle contact heat transfer
coefficient, CXw,, since CXws depends on both the reactor heating temperature and the
feedstock particle diameter and shape (see Equations 8-11). Thus in Table 7, values
of CXws ranging from 250 to 1000 W·m- 2K 1 have been used, which include the most
probable contact heat transfer coefficients in agitated reactors. The predicted results
in Table 7 show that the overall heat transfer coefficient in the new vacuum pyrolysis
reactor varies from 70 to 250 W·m- 2K\ and mainly depends on the value of the
thermal properties of the feedstock, i.e. the type of materials treated.
These predictions were made with a proposed heat transfer model. The
predictions do not take into account phase changes and pyrolytic reactions which will

364
take place when feedstock such as rubber and biomass is pyrolyzed. Work is
ongoing to study potential modifications of this model in cases where substrates such
as biomass or rubber are used as feedstock.

6 Conclusions

A novel vacuum pyrolysis reactor has been described whose overall heat transfer
coefficient abed ranges from 70 to 250 W·m- 2K\ depending on the feedstock treated.
This new reactor uses commercial molten salts (Hitee®) which flow through the
heating plates in the reactor to indirectly heat the feedstock which is conveyed over
the heating plates. The novel transport system agitates the feedstock, thus greatly
improving the heat transfer potential in the reactor.
A new heat transfer model based on Schliinder' s heat transfer model has been
proposed. This model suggests that the two most important heat transfer parameters
are the surface-to-particle contact resistance, !Xws, and the bottom layer renewal
efficiency,~. Using a simple apparatus, !Xws was measured. Then~ was measured in
the batch and semi-continuous reactor and found to depend mainly on the feedstock
particle size, although its value did slightly decrease at high moving speeds. The
values of !Xws and ~ were then used to calculate the overall heat transfer coefficient,
abed, in the batch reactor using gravel feedstock. These theoretical values were in
vety good agreement with measured values of abed in the batch reactor (at most a
20% difference). Having validated the proposed heat transfer model, the measured
values of~ and the previously obtained values of !Xws were used to predict the overall
heat transfer coefficient in the semi-continuous pilot plant reactor for various
feedstocks. The validated new heat transfer model will help optimize reactor design
parameters and diminish the number of experiments required to determine the optimal
parameters, thus facilitating the process of achieving optimal heat transfer in an
industrial scale reactor.

Acknowledgements
This project has been supported by Pyrovac International Inc., Hydro-Quebec and the
Ministere des res sources naturelles du Quebec. The collaboration of the Universite
Laval and Ecotechniek b.v. has been appreciated. The authors wish to thank Mrs.
Carelle Malendoma and Dr. Annette Schwerdtfeger for their precious help in the
preparation of this manuscript.

Notation
A surface area, m 2
B quantity of the agitated fraction k of the first layer of particles
Cp specific heat capacity, J-kg- 1K 1
d diameter, m
m mass, kg
Nmix experimental mixing number according to Schliinder
N,herm dimensionless group variable according to Schliinder

365
Sr sum of particle and surface roughness
T temperature, K
tmix characteristic time of the agitation device, s

Greek letters
a heat transfer coefficient, W·m- 2K 1
p bottom layer renewal efficiency
E average emissivity of wall and particle
A. heat conductivity, W·m- 1K 1
Ag continuum heat conductivity of the gas, W·m- 1K 1
p density, kg·m- 3
a modified mean free path of gas molecules, m
¢A surface coverage factor

subscripts
insulation WS wall to surface
r radiation wp wall to particle
p particle wg wall to gas

7 References

1. Roy, C., Labrecque, B. and de Caumia, B. (1990) Recycling of scrap tires to oil
and carbon black by vacuum pyrolysis. Resour. Conserv. Recycl., Vol. 4, pp. 203-
213.
2. Yang, J., Tanguy, P. A. and Roy, C. (1995) Heat transfer, mass transfer and
kinetics study of the vacuum pyrolysis of a large used tire particle. Chern. Eng.
Sci. Vol. 50, No. 12, pp. 1909-1922.
3. Perry, R.H. and Green, D. (1984) Perry's chemical engineers' handbook, sixth
edition, McGraw-Hill Book Co., New York, pp. 11.48- 11.49.
4. Labrecque, B. (1987) Etude du transfert de chaleur par radiation thermique dans
un reacteur de pyrolyse sous vide des vieux pneumatiques. Memoire de Maitrise,
Universite Laval, Quebec.
5. Lehmberg, J., Hehl, M. and Schugerl K. (1977) Transverse mixing and heat
transfer in horizontal rotary drum reactors. Powder Tech., Vol. 18, pp. 149-163.
6. Wes, G.W.J., Drinkenburg, A.A.H. and Stemerding, S. (1976) Heat transfer in a
horizontal rotary drum reactor. Powder Tech., Vol. 13, pp. 185-192.
7. Schlunder, E.U. (1985) Vacuum contact drying of free flowing mechanically
agitated particle material. DRYING'85, (Eds) R. Toei and A.S. Mujumdar,
Hemisphere/Springer-Verlag, New York, pp. 75-83.
8. Malhotra, K. and Mujumdar, A.S. (1989) Indirect heat transfer and drying in
mechanically agitated granular beds - an annotated bibliography. Drying Tech.,
Vol. 7, pp. 153-171.

366
9. Malhotra, K. and Mujumdar, A.S. (1990) Effect of particle shape on particle-
surface thermal contact resistance. J. of Chern. Eng. ofJapan, Vol. 23, pp. 510-
512.
lO.Weidenbaum, S.S. (1956) Mixing of solids, advances in chemical engineering,
Vol. I, Academic Press, New York, pp. 209-324.
ll.Cooke, H., Bridgwater, J., and Scott, A.M. (1976) Powder mixing- a literature
survey. Powder Tech., Vol. 15, pp. 1-20.
12.Roy, C., Blanchette, D. and de Caumia, B. (1996). Horizontal moving bed
reactor. International patent pending.
13.Schliinder, E. U. (1984) Heat transfer to packed and stirred beds from immersed
bodies. Chern. Eng. Process., Vol. 18, pp. 31-53.
14.Waoka, N. and Kaguei, S. (1982) Heat and mass transfer in packed beds.
Publisher: Gordon and Breach Science Publishers, New York.
15.Malendoma, C. (1996) Physical and mechanical parameters which influence the
transport and agitation of a bed of particles. Institut Pyrovac Inc., Internal report.

367

You might also like