Biomass Pyrolysis Heat Transfer in an Electric Furnace Reactor
Brad Hartwell, Bob Snow
April 29, 2008
EGEE 520
TABLE OF CONTENTS
ABSTRACT……………………………………………………………………………………… 1
1. INTRODUCTION…………………………………………………………………………..… 2
2. MODELING IN COMSOL…………………………………………………………………… 2
2.1
G
OVERNING EQUATIONS …………………………………………………….… 3
2.2
F
ORMULATION……………………………………………………………………. 4
2.3
S
OLUTION………………………………………………………………………….. 5
3. VALIDATION………………………………………………………………………………... 6
4. PARAMETRIC STUDY...……………………………………………………………………. 8
5. CONCLUSION….……………………………………………………………………………9
REFERENCES…..……………………………………………………………………………… 10
ABSTRACT
Pyrolysis of biomass produces larger amounts of bio-oil than other chemical conversions.
Common ways to pyrolyze biomass in a laboratory setting is to utilize a reactor that is being
supplied heat that is then transferred to the particle. This paper models the heat transfer upon a
spherical sample of biomass. The objective is to model the effect of particle size and internal
temperature of the reactor on the heat transfer to the particle. The effect of nitrogen flow on heat
transfer is also addressed but is not the focus since it is only being utilized to ensure a nonoxygenated environment. Conduction and convection heat transfer processes are modeled and
coupled to the nitrogen flow behavior within the reactor. The biomass particle was modeled as a
“pine” sphere in COMSOL. The heat flow into the center of the particle was shown to decrease
as the size of the sphere increased and increased as the temperature surrounding the particle
increased. This paper presents a good case of heat transfer into a sphere from a reactor using
COMSOL. The model can be extended to look at different biomass feed stocks, the shrinking of
the biomass particle and the use of different gases that are pumped into the system.
1
1. INTRODUCTION
As the world’s population continues to increase, demand for energy resources increases.
Meanwhile the fossil fuel reserves such as coal, natural gas and petroleum are depleting at an
ever increasing rate. The world is now turning to alternative fuels to help preserve these fossil
fuel reserves and help reduce greenhouse gas emissions.
Unlike fossil fuels, biomass is a renewable fuel source that is also considered to be carbon
neutral [1]. Increased use of biomass derived fuels will help slow the depletion of fossil fuel
reserves while reducing the carbon dioxide emissions of the world. While most byproducts of
biomass have relative low energy concentrations, biomass oils tend to have the highest energy
concentration of the byproducts [1]. In general, the best chemical conversion process to obtain
the largest fraction of oil is pyrolysis [2]. Pyrolysis is the heating of a substance in an oxygenfree environment so that thermal decomposition occurs, leaving byproducts of energy dense
hydrocarbons. The oils obtained from pyrolysis can be utilized as is to offset the need of heating
oil or can be refined into biodiesel [3-5], which can limit the use of petroleum for transportation.
This paper focuses on modeling the heat transfer of a spherical sample of pine in a cylindrical
electric furnace reactor (r = 15 cm). The objective is to model the internal temperature of the
particle to better understand the heating rate of the biomass particle. The effects of particle size
and internal temperature of the system on the solid-gas heat transfer are addressed. The paper
does not take into account the shrinking or the mass loss of the particle that will occur once it
reaches pyrolysis temperatures. However, others have modeled these changes of a biomass
particle undergoing pyrolysis as well as studying the kinetics behind the process [6-13].
2. MODELING USING COMSOL
COMSOL is computer modeling software that can solve a variety of physical, mechanical, and
thermal systems using the appropriate theoretical equations to model a system. COMSOL is
useful because you can couple several sets of theoretical equations together in order to best
demonstrate a specific system. It is a basic simulation package that is somewhat limited to the
behavioral systems that it lists in the start-up menu. The program had all of the behavioral
2
systems necessary to carry out a heat transfer analysis of biomass pyrolysis, and the
methodology for using COMSOL is described below.
2.1 GOVERNING EQUATIONS
The pyrolysis heat transfer model will be governed by both Navier-Stokes flow around a sphere
as well as conduction and convection heat transfer equations. The fluid flow behavior is the
primary function because the conduction convection equation will require a velocity input from
Navier-Stokes’ analysis. The following were the necessary assumptions for this model: steady
state laminar fluid flow, homogeneous biomass sample, and conservation of mass.
The Navier-Stokes equation for fluid flow around a sphere is:
Where μ is nitrogen’s dynamic viscosity, ρ is its density, and P is the pressure gradient driving
the fluid flow.
The pressure gradient is somewhat ambiguous, so it is typically an
experimentally measured parameter. It should be noted that the viscosity and density of nitrogen
are functions of temperature, creating a link between the conduction and convection behavior.
The partial differential equation conduction and convection is:
Where k is thermal conductivity, ρ is density, and Cp is heat capacity. It should be noted that the
heat capacity, viscosity and density of nitrogen are functions of temperature, thus, coupling the
two equations together. The Navier-Stokes outputs a velocity for the conduction-convection
equation to use as an input, and the temperature output of the heat transfer equation is the new
input for the Navier-Stokes, ending the loop of their interaction.
3
2.2 FORMULATION
The formulation of the problem was done using two different governing equations coupled
together in a COMSOL model. This model utilized the steady state “Incompressible Navier
Stokes” and “Convection and Conduction heat transfer” equations. The biomass particle is
suspended within a reactor of constant temperature of 1000 K. The particle was modeled as a
spherical piece of pine in COMSOL. To account for the flow of nitrogen within the reactor to
ensure pyrolysis and not combustion the +z velocity was set to 1.25 cm/s.
Initially the problem was set up as a simple 2-D model but was later converted into an axisymmetrical 2D model due to the convenient symmetry of the system. The model incorporated
two modes of heat transfer, convection and conduction. The temperature of the outside wall was
set to 1000 K which in turn conducted the heat from the flowing nitrogen gas to the biomass
particle. The nitrogen flow accounted for the convection heat transfer, where heated nitrogen
would be replenished continuously around the particle, maintaining a large heat gradient.
The boundary conditions of the system were set to properly demonstrate the reactor.
As
previously stated, the outside wall was set to a constant temperature of 1000 K and the bottom
and top walls were set to be insulated. Both the biomass sphere surfaces as well as the reactor
walls are treated as no slip surfaces for appropriate fluid flow behavior. The bottom horizontal
boundary is treated as an inlet with a flow velocity of 1.25 cm/s. The top boundary is treated as a
non-stress boundary. The biomass was given an initial temperature of 298 K, and the Nitrogen
gas surrounding it was set initially to 1000 K.
N2 (g)
Wood (Pine)
ρ
C
K
(kg/m3)
(J/kg/K)
(W/m/K)
1.251
2079.25
0.024
560
420
0.12
Table 1: Preliminary input parameters
The values in Table 1 are for ambient temperature and pressure, and were used for preliminary
solution running. In the formulation for “Conduction and Convection,” the velocity’s used were
those derived from the “Incompressible Navier Stokes” equation. The following functions for
nitrogen gas were input after preliminary solutions were converging in order to more accurately
4
represent the fluid flow and the thermal behaviors. These relations are best fit functions for
known data at various temperatures and ambient pressure. All temperatures are input as Kelvin.
C = 4e-12 T3 - 6e-8 T2 + 0.0003 T + 0.9641
2.3 SOLUTION
The pyrolysis model utilizes two behavioral systems; the first representing heat flux via
convection and conduction (CC), and the second being Navier-Stokes (NS) which represents
fluid flow around a spherical biomass particle. Making these two systems communicate with
each other was a tough preliminary issue to resolve. It was discovered later that there is a
velocity functionality in the CC system which allows a linkage between CC and the NS flow
velocities into the heat transfer system by defining the CC velocities as the output velocities from
the NS system. After much exploration, our original gas flow velocity proved to be too great for
the model to converge on a solution. Reducing this velocity by a factor of 10 sufficed in
allowing the system to converge to a solution. To prove the new velocity’s validity, it was
converted to a flow rate of 63 mL/min which fit well into the range observed in biomass
pyrolysis literature [14-17]. The next issue observed was that when using a constant heat flux
provided in literature [1, 18-20], the gas in the system was accumulating too much heat and
reaching unrealistic temperatures (upwards of 90,000 K). These high temperatures were also
immensely decreasing the time necessary for pyrolysis completion.
The furnace boundary
condition was then switched from a constant heat flux, to a constant temperature of 1000 K.
Instead of supplying a constant amount of heat, actual furnace reactors are thermostatically
controlled to maintain a specific output temperature.
It is typical for the center of a woody biomass particle undergoing furnace pyrolysis to reach
temperatures between 700 and 900 K [1, 21]. Typical heating times for a comparable sized
5
spherical particle (r = 2 cm) to completely pyrolyze is between 12 to 15 minutes [10, 15]. Our
model has demonstrated validity by producing a temperature of 700 K in the center of our
particle after a heating time of 14 minutes (See Figure 1). Both of these values fit within their
respective realistic ranges.
Figure 1: COMSOL temperature solution of system at various times
5. VALIDATION
In order to validate the solution obtained the velocity of the nitrogen flow was set to zero which
would make the simulation purely conduction. Since the problem is symmetrical we can utilize
line plots. Differences between the conductive fluxes in the vertical plane at the center of the
particle and just off to the side of the particle are shown in Figure 2a and Figure 2c. Since the
6
biomass particle transmits heat much slower than the nitrogen atmosphere the conduction flux
decreases and becomes near zero in the center, while when off to the side of the particle the flux
increases steadily until it reaches the center and retreats symmetrically. A comparison between
Figure 2a and Figure 2b shows the differences between the systems when there is a flow of
nitrogen and when there is not. Figure 2d shows the conductive flux on the horizontal axis also
going through the center of the sphere. As the wall of the reactor is approached along the
horizontal axis it is noticed that the conductive flux gradient approaches zero. This phenomenon
is due to the constant temperature being kept at the wall and shows that COMSOL is calculating
the heat flux correctly.
Figure 2: Conductive flux along A) z-axis (r=0); B) z-axis r=0 for cond-conv
C) z-axis (r=0.021); D) r-axis (z=0).
7
6. PARAMETRIC STUDY
This studies main two goals were to investigate the solution differences when altering the
particle radius, as well as changing the furnaces’ heating wall temperature. Figure 4 below
demonstrates the former investigation. The most obvious aspect of this investigation is that after
the same reaction time (10 minutes), the smaller particles have significantly higher internal
temperatures.
Therefore, as you increase the particle’s radius, larger pyrolysis times are
required. This indicates that it might be more advantageous to have smaller biomass particles in
a large scale pyrolysis reactor. This would mean that biomass grinding may be necessary for
pyrolyzing wood.
Investigating the effect of the furnace wall temperature (see Figure 3) on the solution shows that
after 10 minutes of reacting time and increasing the furnace wall temperature 200 K, there is
only a 100 K temperature difference in the center of the particle. There also seems to be a larger
internal heat gradient both within the particle as well as in the nitrogen gas for the higher wall
temperature case. This is most likely a result of the larger temperature gradient between the wall
and the nitrogen gas, and in turn a larger gradient between the nitrogen gas and the initial
temperature of the particle.
8
Figure 3: Cross-sectional solution (along r-axis at z=0) of particle after 10 min. for furnace wall
temperatures of a) 1000 K and b) 1200 K
Figure 4: Effect of particle size on solution after 10 minutes
7. CONCLUSION
A heat transfer model of a pine spherical biomass particle undergoing pyrolysis in an electric
furnace was done using COMSOL. The heat flow into the center of the particle was shown to
both decrease as the size of the sphere increased, and increase as the temperature surrounding the
particle increased. These results imply that biomass should be grinded into fine pieces to
optimize a large scale pyrolysis furnace reactor. Although the time of reaction is decreased with
higher furnace temperatures, further investigation should be carried out on the most overall
efficient furnace temperature for pyrolysis. Further modeling investigations should involve
simple thermodynamic kinetics to model thermal degradation of the particle coupled with heat
9
transfer behavior. This paper presents a good case of heat transfer into a sphere from a reactor
using COMSOL. The model can be extended to look at different biomass feed stocks and the use
of different inert gases flowing through the system.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
Demirbas, A. and Ia, Effect of temperature on pyrolysis products from biomass. Energy Sources
Part a-Recovery Utilization and Environmental Effects, 2007. 29(4): p. 329-336.
Lede, J., et al., Properties of bio-oils produced by biomass fast pyrolysis in a cyclone reactor. Fuel,
2007. 86(12-13): p. 1800-1810.
Yaman, S., Pyrolysis of biomass to produce fuels and chemical feedstocks. Energy Conversion and
Management, 2004. 45(5): p. 651-671.
Dominguez, A., et al., Production of bio-fuels by high temperature pyrolysis of sewage sludge
using conventional and microwave heating. Bioresource Technology, 2006. 97(10): p. 11851193.
Park, E.S., B.S. Kang, and J.S. Kim, Recovery of oils with high caloric value and low contaminant
content by pyrolysis of digested and dried sewage sludge containing polymer flocculants. Energy
& Fuels, 2008. 22(2): p. 1335-1340.
Babu, B.V., A.S. Chaurasia, and Ch, Heat transfer and kinetics in the pyrolysis of shrinking
biomass particle. Chemical Engineering Science, 2004. 59(10): p. 1999-2012.
Chaurasia, A.S. and B.D. Kulkarni, Most sensitive parameters in pyrolysis of shrinking biomass
particle. Energy Conversion and Management, 2007. 48(3): p. 836-849.
Cai, J.M. and R.H. Liu, Parametric study of the nonisothermal nth-order distributed activation
energy model involved the Weibull distribution for biomass pyrolysis. Journal of Thermal Analysis
and Calorimetry, 2007. 89(3): p. 971-975.
Senneca, O., Kinetics of pyrolysis, combustion and gasification of three biomass fuels. Fuel
Processing Technology, 2007. 88(1): p. 87-97.
Zhang, X.D., et al., Study on biomass pyrolysis kinetics. Journal of Engineering for Gas Turbines
and Power-Transactions of the Asme, 2006. 128(3): p. 493-496.
Thipkhunthod, P., et al., Describing sewage sludge pyrolysis kinetics by a combination of biomass
fractions decomposition. Journal of Analytical and Applied Pyrolysis, 2007. 79(1-2): p. 78-85.
Zhang, J., et al., Product analysis and thermodynamic simulations from the pyrolysis of several
biomass feedstocks. Energy & Fuels, 2007. 21(4): p. 2373-2385.
Shao, J., et al., Pyrolysis characteristics and kinetics of sewage sludge by thermogravimetry
Fourier transform infrared analysis. Energy & Fuels, 2008. 22(1): p. 38-45.
Miao, X.L., et al., Fast pyrolysis of microalgae to produce renewable fuels. Journal of Analytical
and Applied Pyrolysis, 2004. 71(2): p. 855-863.
Di Blasi, C., et al., Pyrolytic behavior and products of some wood varieties. Combustion and
Flame, 2001. 124(1-2): p. 165-177.
Chen, G., et al., Biomass pyrolysis/gasification for product gas production: the overall
investigation of parametric effects. Energy Conversion and Management, 2003. 44(11): p. 18751884.
Miao, X.L., Q.Y. Wu, and Wd, High yield bio-oil production from fast pyrolysis by metabolic
controlling of Chlorella protothecoides. Journal of Biotechnology, 2004. 110(1): p. 85-93.
Zhang, Y., et al., Peculiarities of rapid pyrolysis of biomass covering medium- and hightemperature ranges. Energy & Fuels, 2006. 20(6): p. 2705-2712.
10
19.
20.
21.
Dominguez, A., et al., Conventional and microwave induced pyrolysis of coffee hulls for the
production of a hydrogen rich fuel gas. Journal of Analytical and Applied Pyrolysis, 2007. 79(1-2):
p. 128-135.
Ischia, M., et al., Pyrolysis study of sewage sludge by TG-MS and TG-GC-MS coupled analyses.
Journal of Thermal Analysis and Calorimetry, 2007. 87(2): p. 567-574.
Becidan, M., O. Skreiberg, and J.E. Hustad, Experimental study on pyrolysis of thermally thick
biomass residues samples: Intra-sample temperature distribution and effect of sample weight
("scaling effect"). Fuel, 2007. 86(17-18): p. 2754-2760.
11