Pennsylvania State University



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

1. GOVERNING EQUATIONS …………………………………………………….… 3

2. FORMULATION……………………………………………………………………. 4

3. SOLUTION………………………………………………………………………….. 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 non-oxygenated 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. 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 oxygen-free 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 [pic][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 [pic][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 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:

[pic]

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:

[pic]

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.

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 axi-symmetrical 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.

|  |ρ (kg/m3) |C (J/kg/K) |K (W/m/K) |

|N2 (g) |1.251 |2079.25 |0.024 |

|Wood (Pine) |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 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

[pic]

[pic]

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 [pic][14-17]. The next issue observed was that when using a constant heat flux provided in literature [pic][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 [pic][1, 21]. Typical heating times for a comparable sized 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.

[pic]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 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.

[pic]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).

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.

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

[pic]

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 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. 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.

2. 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.

3. Yaman, S., Pyrolysis of biomass to produce fuels and chemical feedstocks. Energy Conversion and Management, 2004. 45(5): p. 651-671.

4. 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. 1185-1193.

5. 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.

6. 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.

7. 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.

8. 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.

9. Senneca, O., Kinetics of pyrolysis, combustion and gasification of three biomass fuels. Fuel Processing Technology, 2007. 88(1): p. 87-97.

10. 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.

11. 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.

12. Zhang, J., et al., Product analysis and thermodynamic simulations from the pyrolysis of several biomass feedstocks. Energy & Fuels, 2007. 21(4): p. 2373-2385.

13. 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.

14. 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.

15. Di Blasi, C., et al., Pyrolytic behavior and products of some wood varieties. Combustion and Flame, 2001. 124(1-2): p. 165-177.

16. 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. 1875-1884.

17. 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.

18. Zhang, Y., et al., Peculiarities of rapid pyrolysis of biomass covering medium- and high-temperature ranges. Energy & Fuels, 2006. 20(6): p. 2705-2712.

19. 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.

20. 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.

21. 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.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download