Heat Transfer in a Pulsed Fluidized Bed of Biomass Particles

Subscriber access provided by UB + Fachbibliothek Chemie | (FU-Bibliothekssystem)

Article

Heat transfer in a pulsed fluidized bed of biomass particles Dening Jia, Xiaotao Tony Bi, C. Jim Lim, Shahab Sokhansanj, and Atsushi Tsutsumi Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.6b04444 • Publication Date (Web): 15 Mar 2017 Downloaded from http://pubs.acs.org on March 16, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Industrial & Engineering Chemistry Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Heat transfer in a pulsed fluidized bed of biomass particles Dening Jiaa, Xiaotao Bia*, C. Jim Lima, Shahab Sokhansanja,b and Atsushi Tsutsumic

a. Department of Chemical and Biological Engineering, the University of British Columbia, 2360 East Mall, Vancouver, BC, V6T 1Z3, Canada b. Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA c. Collaborative Research Center for Energy Engineering, Institute of Industrial Science, The University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505, Japan

*Corresponding author: Xiaotao Bi, email: [email protected]

ACS Paragon Plus Environment

1


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 59

Abstract Bed-to-surface heat transfer coefficients of various biomass particles were measured in a pulsed fluidized bed. Effects of flow rate, pulsation frequency, particle size distribution, fines and vibration on heat transfer were investigated. Higher gas flow rates generally yielded higher heat transfer coefficients. Natural frequency was found to be optimum as it offered ample bed movement and internal solid circulation without too much inactivity between pulsation cycles. Heat transfer was also heavily influenced by the interaction between gas convective and particle convective heat transfer, which was verified by the proposed heat transfer model. Two mechanisms, one of which treated the flow-on and flow-off period within a pulsation cycle individually, the other utilized the actual bubble rise velocity obtained via high-speed imaging were identified to account for the significantly different flow behavior below and above natural frequency. Good agreement was observed between experimental data and modelled results.

Keywords: Fluidized bed; pulsation; heat transfer; biomass; multiphase flow


2

Page 3 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. Introduction Biomass is a promising energy source that can be not only used directly to generate heat and electricity, but also converted into solid, liquid or gas fuels for different applications. Types of biomass include industrial waste, industrial co-products, municipal solid waste, food waste, agricultural and forest residues, among which harvesting residues (e.g. tree tops and branches removed from the forest) and wood residues (wood chips, shavings and sawdust) are favored in particular over the fact that food security is not compromised when compared to biomass resources such as corn.1 Research by Mabee and Saddler 2 has shown that second-generation feedstock could satisfy half of Canada’s need for transportation fuel, reducing the amount of carbon emissions by replacing fossil fuels. Many biomass thermal conversion technologies are readily available such as torrefaction, pyrolysis and gasification. In order to process biomass materials from different sources that may differ widely on physical and chemical properties, a versatile reactor that allows flexibility in terms of feedstock properties and operating conditions is greatly favored. It is not surprising that fluidized bed is often selected for biomass conversion over fixed bed reactor for its high heat and mass transfer rate, temperature uniformity and the ease of scaling-up. Biomass in general, however, cannot be easily fluidized due to its irregular shape, high moisture content, low bulk density and wide particle size distribution. Many biomass fluidized beds employ sand or other inert bed particles, in which sand serves not only as the bed medium to stabilize the gas-solid fluidization, but also as a heat carrier for thermal conversion processes such as combustion and gasification.3, 4 However, the quality of the solid product is often compromised due to the increased ash content introduced from attrition of bed materials, which also poses a threat to biomass-


3


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 59

operated boilers and turbines as the increased ash content in fuel pellets may lead to sintering, fouling, and eventually unscheduled shutdowns. The fluidization quality of biomass particles is often affected by their cohesive nature.5 As a result, a variety of approaches such as specially designed distributors, baffles and internals, gas pulsation, electric field, vibration, mechanical agitation as well as sonication can be adopted to improve the fluidization quality. In the case of gas pulsation, fluidization medium (such as air) is supplied periodically, where the bed expands as gas is turned on and contracts while the gas is cut off. This is often achieved by the periodic control using a solenoid valve or a butterfly value.6, 7 Periodic switch of fluidization gas to different part of bed can also be treated as a form of gas pulsation, which have been reportedly achieved by the use of rotating gas distributors,8, 9 spinning discs with a partially hollow section,10 as well as hollow cylinders placed in the bed with horizontal slots on opposite sides.11 Massimilla, et al. 12 investigated gas pulsation in both 2D and 3D fluidized beds and observed three types of bed behavior at different pulsation frequencies, including the intermittent fluidization at low frequencies (1.2–2.7 Hz), piston-like fluidization at medium frequencies (2.7–4.8 Hz) as well as an apparently “normal” fluidization at high frequencies (>4.8 Hz). Similar bed behavior was discovered by Wong and Baird 6 in a 102 mm diameter fluidized bed with the pulsation frequency ranging from 1–10 Hz. A piston model was also proposed to predict the natural frequency of such systems. Kobayashi, et al. 13 studied bed pressure profiles and particle velocity profiles in a pulsed fluidized bed with glass beads of four different sizes and determined the maximum “flow-on” time in a pulsation cycle corresponded to the amount of time required for the first bubble to reach bed surface. Wang and Rhodes 14 studied gas pulsation via computational fluid dynamics (CFD) where a wide range of frequencies and gas flow rates were applied. Effects of pulsation on overcoming


4

Page 5 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


defluidization was observed between 3 and 15 Hz. Wu, et al.

15

successfully replicated the

experimentally observed bubble pattern in a pulsed fluidized bed based on simulation using the discrete element model (DEM). Compared to conventional fluidized beds with continuous supply of fluidization medium, fluidized beds with pulsed gas flow are able to process a wider range of particle sizes. The enhanced gassolid contact in pulsed beds also leads to improved heat and mass transfer rates between gas and solid particles, as well as dense phase and heating/cooling surfaces. Kobayashi, Ramaswami and Brazelton

13

reported a 300% increase in product uniformity and 100% increase in processing

capacity when air was pulsed in a commercial mixer of minerals. Many researchers also worked on pulsed fluidized bed drying, including agricultural products, forest residues and chemicals. Increasingly, fluidized beds have also been used for biomass torrefaction,16, 17 pyrolysis18 and gasification.19 Pulsed fluidized beds have shown promises in overcoming difficulties in fluidization caused by the unique properties of biomass materials such as high moisture content, irregular shape, wide particle size distribution and low bulk density. A pulsation frequency near the natural frequency of the fluidized bed system was found to be effective in improving mass transfer rate and better gas-solid mixing

20-22

. However, the design and scale-up of pulsed fluidized bed reactors for

thermochemical operations are still held back by the limited understanding of the heat transfer behavior of biomass particles under pulsed gas flow. Few studies could be found in the literature on heat transfer in pulsed fluidized beds, in terms of the improvement in heat transfer as a result of pulsation. Zabrodskii and Bokun 23 studied bed-tosurface heat transfer in fluidized bed with pulsed gas flow with the pulsation frequency up to 2 Hz.


5


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 59

It was discovered that as soon as minimum fluidization was reached, slightly increasing gas flow rate resulted in higher heat transfer rates in pulsed fluidized bed compared to a fluidized bed without gas pulsation. Less airflow was required to reach the same maximum heat transfer coefficient as in a conventional fluidized bed. Kobayashi, Ramaswami and Brazelton

13

experimentally investigated the effects of pulsation on heat transfer in a gas-fluidized bed. Pulsation resulted in an attenuation of the bubbling phenomenon, shorter particle residence time on the heating probe surface, and a lower bubble volume fraction, compared with a conventional fluidized bed with steady gas flow. Increases in heat transfer were attributed to active particle mixing below minimum fluidization.24 Zhang

25

studied the hydrodynamics and heat transfer

between silica sand of different sizes ranging from Geldart A to Geldart B/D and a horizontal heating probe in a pulsed fluidized bed operated in bubbling regime and concluded that pulsation could significantly increase heat transfer by improving the fluidization quality, where higher standard deviation of pressure fluctuations was observed in the pulsed bed than the conventional ones. The bed-to-surface heat transfer between a vertically mounted heating probe and pulsed fluidized bed was studied by Nishimura, et al. 26 where the flow-on time of 0.5 or 1.0 s was used. At lower gas velocities, the prolonged flow-off periods resulted in a pressure build-up and subsequently the formation of large bubbles, which improved the contact efficiency between the heating surface and the dense phase, as well as the heat transfer coefficient. There are limited studies on heat transfer of biomass in fluidized beds. Sun, et al. 27 investigated heat transfer of crushed pellets being co-fired with coal in a circulating fluidized bed, in which up to 25% (w.t) of biomass was mixed with coal. The heat taken away from the fluidized bed by bedto-surface heat transfer was measured by the evaporation of water. The overall heat transfer coefficient was also divided into different components and calculated using known correlations,


6

Page 7 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


including radiation heat transfer, gas convective and particle convective heat transfer. Kalita, et al. 28

measured the bed-to-surface heat transfer coefficients of biomass-sand mixtures in a circulating

fluidized bed where up to 20% (w.t) of sand was mixed with the biomass particles. By blending in more biomass, heat transfer decreased. Moreover, heat transfer coefficient was also found to increase with an increase in suspension density and operating pressure. An optimum with high heat transfer rate and uniform circulation rate was also identified. CFD simulation performed by Papadikis, et al. 29 revealed the influence of particle size on bed-to-surface heat transfer of biomass particles, with the maximum heat transfer coefficients for both 350 and 550 μm particles being obtained. However, experimentally obtained heat transfer coefficients of pure biomass particles in the absent of inert bed particles are hardly researched. The lack of data on the bed-to-surface heat transfer coefficients of pure biomass particles in fluidized beds, and the lack of mechanistic understanding on the heat transfer behavior in pulsed fluidized bed have been the major motivations for this work, where the heat transfer behavior of various biomass particles was measured, in combination with a mechanistic heat transfer model for predicting bed-to-surface heat transfer coefficient in pulsed fluidized beds.

2. Experimental apparatus and procedure 2.1 Pulsed fluidized bed The experimental setup of the pulsed fluidized bed system is shown in Fig. 1. The pulsed fluidized bed (PFB) was initially designed to study the fluidization characteristics and electrical charging behavior of fly ash30 and later modified for biomass drying.21 The total height of the Plexiglas


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 59

column was 1.0 m with a rectangular cross-section of 0.15 m×0.10 m. There were seven measurement ports on the side of the column, and the first one was located 25 mm above the distributor plate, with the rest being 102 mm apart. Windbox of the PFB was securely mounted on a vibratory base (Eriez 48A, Eriez Manufacturing Co., USA) that offered a constant vibration frequency of 60 Hz. The amplitude was fully adjustable from 0% to 100% with a maximum vertical displacement of 0.381 mm. To reduce pressure drops across the grid and possible dampening of gas pulsations, a perforated aluminum plate with triangular-patterned 3.2 mm diameter orifices and 40% opening area ratio was selected as distributor instead of the commonly implemented 10– 12% opening area ratio. To avoid particles falling into the windbox, a fine mesh was sandwiched between the distributor plate and a bottom plate that had slightly larger orifices with the same orifice patterns as the top plate. Gas pulsation was generated by a solenoid valve (8212G034120/60, ASCO Valve, USA) that was controlled by a computer program, in which the “ON” and “OFF” time of the solenoid valve could be controlled independently. Besides the pulsating stream mentioned above, another line of air (the fluidizing stream) was also installed so that particles could remain mobilized should the oscillating flow rate of air in the pulsating stream become too low. Both streams were equipped with rotameters and needle valves (FL6000-V Series, Omega, Canada) in order to control the flow rate. A surge tank was placed on the pulsating stream to ensure stable air supply. Dehumidified high-pressure building air was the fluidization medium, which was fed to both streams after being regulated by a pressure regulator (AR-40-N04H-Z, SMC Pneumatics, Canada). The system was also equipped with an in-line gas heater (AHP-7561, Omega, Canada) so that temperature of the fluidizing air could be regulated by a programmable temperature controller (CN4316, Omega, Canada).


8

Page 9 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


2.2 Bed-to-surface heat transfer measurement A heating probe was assembled and installed in the fluidized bed column, shown as HT2 in Fig. 1, to measure the heat transfer coefficients between an immersed surface and the bed, or the bedto-surface heat transfer coefficient. Detailed design of the probe is illustrated in Fig. 2. Ni–Cr heating wire was wrapped tightly around a hollow polytetrafluoroethylene (PTFE) tube in a spiral fashion and inserted into a copper shell (50.8 mm in length) with slightly larger diameter (15.9 mm OD). The gap between the PTFE core and copper shell was filled with thermally conductive and electrically insulating cement (OB 600, Omega, Canada). To prevent air bubbles from being trapped in the tight gap that would interfere with heat conduction, the filling process was completed in vacuum on a vibrating table. Both the end cap and top cap were made of PTFE so that heat loss by means of axial heat conduction from both ends could be reduced. T-type thermocouples (T4 and T5 in Fig. 1) were cemented at the midpoint of the copper surface on opposite sides of the heater to measure surface temperature, Ts. A B&K Precision power supply (Model No.1785B, USA) with adjustable voltage (0-18 V) provided DC power to the heating wire. The probe was mounted in the fluidized bed column vertically within a rigid and insulating acrylic tube. The heat transfer coefficient was calculated according to an energy balance, given as,

h

VI  Ql As (Ts  Tb )

(1)

in which V and I are the voltage and current read from a DC power supply. Ql is the estimated heat loss. Ts and Tb represent surface temperatures of the probe and the fluidized bed, respectively. As is the vertical surface area of the copper shell where heat transfer takes place. In order to determine


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 59

the heat loss Ql, experiments and computational fluid dynamics (CFD) simulations were conducted where the probe was used to measure heat transfer coefficient between the heating surface and surrounding air, with measured temperatures of the top cap and end cap as boundary conditions. Results indicated a heat loss of less than 5%. As a result, heat loss was considered negligible and Eqn. (1) is further simplified to,

h

VI As (Ts  Tb )

(2)

Heat transfer coefficients at different operating conditions were measured. Parameters investigated including average gas flow rate, pulsation frequency, pulsation duty cycle, particle size as well as biomass species.

2.3 Data Acquisition Pressure drop across the bed was measured by a pressure transducer (PX164-0105DV, Omega, Canada), in addition to the absolute pressure of the column (PX163-005BD5V) and windbox (PX142-005D5V). Relative humidity of air at the inlet and outlet were also monitored (HM141 and HMT335, respectively, Vaisala, Finland). T-type thermocouples were installed at different locations to measure the temperatures of the dense phase and freeboard. All analog signals were collected through a number of data acquisition devices (DAQs), while being displayed and saved onto a computer equipped with LabVIEW (Version 2015, National Instrument, USA). Instantaneous voltage and current readings of the DC power supply were directly exported to LabVIEW through a TTL serial interface provided by the manufacturer. To capture and analyze the transient behavior of the pulsed fluidized bed, a high-speed camera was


10

Page 11 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


placed in front of the column (240 frames per second at 1280 × 720 pixels, f/2.0 aperture). Details of the DAQs and sampling frequencies are listed in Table 1. The long response time of thermocouples limited the measurements of heat transfer coefficient to be steady state only. The surface temperature of the probe was set to be 20 °C higher than the bed temperature such that the temperature difference (Ts−Tb) was significant enough to reduce error in Eqn. (2) for calculating h, while ensuring the probe surface temperature would not become too high to greatly alter properties of surrounding air. For each operating condition, at least 5 min was allowed for the system to stabilize so that temperature fluctuations on the probe surface were kept within 1 °C. After reaching such a steady state, one minute of data was recorded at 20 Hz that would yield approximately 1200 data points.

2.4 Biomass Particles Douglas fir sawdust used in most of the experiments was generously donated by Tolko Industries Ltd. (Vernon, Canada), along with pine sawdust. Switchgrass was also studied. True density of the biomass was measured by a nitrogen gas pycnometer (Model MVP-D160-E, Quantachrome Instruments, Boynton Beach, FL, USA), and the bulk density was measured by a graduated glass cylinder (25 ml capacity, 22.2 mm diameter, 0.5 ml scale intervals). All biomass materials were sieved through a mesh with 3.15 mm openings to remove particles too big for fluidization. Since switchgrass contained more fines and needle-shaped particles, the sphericity was lowest among all samples. Sphericity was measured in the same manner as Tannous, et al. 31 where high-resolution images of particles were captured by a scanner, imported and processed in a MATLAB image processing toolbox (Toolbox version 7.10.0499, MATLAB version R2010a). Sphericity of a


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 59

particle was determined by the ratio of diameters of the inscribed and circumscribed circles.32 Particle size distribution of all three species is shown in Fig.3, with the rest of the properties including minimum fluidization velocity summarized in Table 2. It should be noted that all properties were measured with particles containing 7% moisture (dry basis).

3. A heat transfer model for pulsed fluidized bed 3.1 Particle convective heat transfer Different approaches have been taken to model and predict heat transfer coefficient in fluidized beds. Many attempted to link heat transfer coefficient with gas flow rate and other properties of the fluid and particles. In most cases, empirical correlations were established between Nusselt number (Nu), fluid Prandtl number (Pr) and Reynolds number (Re). Borodulya, et al.

33

investigated the heat transfer between an immersed vertical tube and a fluidized bed, and arrived at the following correlation,

Nu p 

hc d p kg

 0.74(Ar)0.1 (

+0.46(Re p Prg )

Molerus and Wirth

34

 p 0.14 c pp 0.24 ) ( ) (1   )0.67 g c pg

(1   )0.67

(3)



looked into multiple dimensionless groups which were related to heat

transfer in fluidized beds, analyzed the importance of such groups at different flow regimes (Archimedes number) and proposed a semi-empirical correlation for an intermediate gas flow rate with 102 < 𝐴𝑟 < 105 ,


12

Page 13 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1

1   U  U mf  p c pp    3 0.125(1   mf ) 1  33.3  3 (U  U mf )     U k mf g     hlt   2 0.5 kg  U mf   g    p c pp k g  2 1 (U  U mf )   3 1  0.28(1   mf )   U  U mf 2c pp    p   g   k g g   

 g +0.165 Pr1/3    g  p

1/3

  

  U  U mf  1  0.05    U mf

  

1

   

   

(4)

1

where lt is the turbulent flow regime length scale, given by,   lt    g ( p  g )g 

   

2/3

(5)

In terms of mechanistic modelling, the packet theory proposed by Mickley and Fairbanks 35 may be one of the most appropriate for fluidized bed heat transfer. An analogy is made with the surface renewal theory for mass transfer36 where heat transfer surface is contacted alternately by gas bubbles and the dense phase (or packets). Heat transfer is achieved by transient conduction between packets and heating surface through the gas film, from the brief period of time when the packet resides on the surface. The instantaneous heat transfer coefficient between an isothermal heating surface and the packet is,

hi 

mf c p ,mf keff t

(6)

in which keff, cp,mf, ρmf are the effective conductivity (W/(m·K)), heat capacity (J/(mol·K) and density (kg/m3) of the dense phase, respectively. The time-averaged heat transfer coefficient could be calculated if the residence time distribution of the packet is made available. In slugging regime


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 59

where the upward velocity of the dense phase is shown to be (U − Umf),37 the residence time on a short vertical surface of Length L (m) could be expressed as,

  L / (U  U mf )

(7)

whereas in bubbling fluidized beds the upward motion of packets is mostly attributable to the motion of gas bubbles, specifically the wake and drift of the bubbles. Assume the upward velocity of the bubble wake is close to the bubble rise velocity UB (m/s), the residence time of the packets is given by,

  L / UB

(8)

When the heat transfer surface is longer than the distance travelled by the packet before it is interrupted by a rising bubble, the characteristic length l (m) of the contact time should replace the surface length L. It has been shown that l can be taken as about half of the bubble diameter dB (m). In a fluidized bed with a multi-orifice distributor, bubble size at a given bed height z (m) can be calculated based on the semi-empirical correlation proposed by Mori and Wen 38,

d B  d Bm  (d Bm  d B 0 )exp(0.3z / Dt ) d Bm  1.64  A(U  U mf ) 

dB0

0.4

1.38  A(U  U mf )   0.2   g  N or 

(9)

(10)

0.4

(11)

Dt (m) is the diameter of the fluidized bed column, A is the cross-sectional area of the bed, and Nor is the number of orifices on the distributor plate. Bubble rise velocity can be estimated by Grace 39

,


14

Page 15 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


U B  0.71 gd B  (U  U mf )

(12)

The average heat transfer coefficient between the dense phase and the heat transfer surface can then be obtained through integration, assume l < L,

hp  



0

hi



dt  2

keff mf c p.mf U B l

(13)

where cp,mf and ρmf can be estimated respectively by cpp (heat capacity of the solids) and 𝜌𝑝 (1 − 𝜀𝑚𝑓 ). Since heat transfer surface is alternately contacted by dense phase and bubbles, the fraction of bed that is occupied by bubble must be taken into consideration for the calculation of particle convective heat transfer,

hpc  hp (1   B )

(14)

The effective conductivity keff (W/(m·K)) is generally given as a sum of two components, kes and kef,

keff  kes  kef

(15)

in which kef represents conductivity due to fluid flow, and kes is the thermal conductivity of fixed beds of particles. kes has been thoroughly studied, and correlations can be found in Yagi and Kunii 40

, as well as Kunii and Smith 41. A numerical solution was also proposed by Deissler and Boegli

42

, kes  k p    k g  k g 

0.280.757log10  0.057log10 ( k p / k g )

(16)


15


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 59

kes values calculated from above equation are in good agreement with that predicted by Kunii and Smith 41 and experimental data by Dietz 43. kef on the other hand is given by Ranz 44,

kef  0.1 g c pg d pU mf

(17)

3.2 Gas convective heat transfer For system with large particles, the contribution of gas phase to overall heat transfer should not be neglected. This is mostly due to the fact that in systems with larger particles heat penetrates fewer layers of particles at the heat transfer surface. Kimura, et al. 45 showed that the void fraction near the wall was larger than that at the core of a fixed bed, indicating that the influence of gas properties is no longer negligible. In fact Baskakov, et al.

46

demonstrated that gas convection became

dominant when particles were larger than 3 mm in fluidized beds. For a vertical cylinder immersed in a fluidized bed, Gabor

47

showed the integral mean gas

convective heat transfer coefficient over the entire length to be,

hs 

4keff  g c pgU

L



keff Ds

(18)

where Ds (m) is the diameter of the cylinder. It has been verified experimentally that bed-to-surface heat transfer is heavily influenced by gas thermal conductivity, as well as the volumetric heat capacity of solids (𝜌𝑝 𝑐𝑝𝑝 ). Better agreement results have been obtained if a thin gas layer of thickness dp/6 is assumed at the wall in series with the dense phase,47, 48

hw  6kg / d p

(19)


16

Page 17 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Many researchers considered the simple approach of factoring the wall resistance along with the bed resistance,49, 50 such that the overall gas convection component hgc (W/(m2·K)) to be given as,

hgc 

1 1/ hs  1/ hw

(20)

Therefore, the overall heat transfer coefficient h is simply the sum of particle convective and gas convective component,

h  hgc  hpc

(21)

3.3 Modeling heat transfer in pulsed fluidized beds A large number of studies could be found on fluidized bed heat transfer, and mechanistic heat transfer modelling,23, 35, 51 with majority of them intended for spherical particles in fluidized beds with continuous gas supply and only a limited number of works extended to other irregular shapes of particles or pulsed gas flow. Kobayashi, Ramaswami and Brazelton 13 related pulsed fluidized bed heat transfer to particle residence time on heat transfer surface as correlated with parameters derived from visual observation and failed to address the unique phenomenon caused by the oscillatory gas flow and the mechanism behind the improvement in heat transfer. Therefore, a better model is still needed. Phenomenally, pulsed fluidized beds exhibit different behavior at different pulsation frequencies. At lower pulsation frequencies distinct “ON” and “OFF” periods where vigorous gas-solid mixing occurs in the “ON” period, followed by the motionless “OFF” period. As pulsation frequency increases above the natural frequency of the system, “OFF” periods become less discernable. However, velocity and pressure still vary greatly in each cycle. Therefore, it seems reasonable to


17


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 59

take different approaches to model heat transfer in pulsed fluidized beds operating above or below natural frequency.

3.3.1 Model for low pulsation frequency (Model 1) For fluidized beds operated below the natural frequency where intermittent fluidization occurs, the “ON” period could simply be treated as a continuous fluidized bed while the “OFF” period to be treated as a fixed bed with motionless particles and negligible gas flow. The overall heat transfer coefficient should be a combination of the two. In this case, the average superficial gas velocity in “ON” periods can be expressed as,

U ON 

U 

(22)

̅ (m/s) is the average superficial gas velocity based on the entire cycle of gas pulsation, here 𝑈 ̅𝑂𝑁 (m/s) represents the average superficial gas velocity in the “ON” period, with β being whereas 𝑈 the time fraction of the “ON” period. It is noteworthy that β increases with pulsation frequency. The value of β can be determined by high-speed imaging, as well as the differential pressure ̅𝑂𝑁 in Eqns. (9) – (12) so that the bubble size and bubble rise velocity signals. Substitute U with 𝑈 can be calculated. It is noteworthy that above equations were derived using Group B round particles. Attentions have been paid to dry and round particles of narrow size distributions, with limited studies on irregular shaped particles. When applying those equations and correlations to irregular particles such as biomass that possesses high moisture content, low shape factor or sphericity, wide particle size distribution and low bulk density, cautions need to be exercised. Kruggel-Emden and Vollmari 52


18

Page 19 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


found that the minimum fluidization velocity of biomass particles calculated from correlations deviated from the experimental data by 20% to 170%. It was also discovered that particle shape excreted strong impact on the flow behavior. Until better correlations for irregular shaped particles become available, cautions should be exercised when Eqns. (9) – (12) are applied to estimate bubble size and bubble rise velocity in fluidized beds of irregular shaped particles on Geldart B/D boundary. The overall heat transfer coefficient of the “ON” period hON that consists of a particle convective and gas convective component can be evaluated from Eqns. (13) – (21) by simply replacing U with ̅𝑂𝑁 . Heat transfer coefficient of the “OFF” period hOFF (W/(m2·K)) equivalent to that of a fixed 𝑈 bed reactor with essentially no airflow,

hOFF  kes / l

(23)

Finally, the overall heat transfer coefficient can be calculated by,

h1   hON  (1   )hOFF

(24)

By clearly identifying the flow characteristics at pulsation frequencies below the natural frequency of the pulsed fluidized bed, the “ON” and “OFF” periods were treated as a fluidized bed and a fixed bed, respectively. Since “ON” period behaves as a continuous fluidized bed with bubbling frequency identical to the pulsation frequency, correlations developed by Mickley and Fairbanks 35

(packet theory for fluidized bed heat transfer), Mori and Wen

bubbling fluidized bed), Grace

39

38

(bubble size estimation for

(bubble rise velocity for fluidized bed) and Gabor

47

(gas

convective heat transfer in fluidized bed) were applied. The “OFF” period resembles a fixed-bed reactor with very low fluid flow, so correlation developed for fixed-bed by Deissler and Boegli 42


19


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 59

was used. It should be emphasized that no correlation derived for fixed-bed reactor was used in the fluidized-bed regime, and vice versa.

3.3.2 Model for high pulsation frequency (Model 2) Above natural frequency only one set of voids (slugs or bubbles) are formed above the distributor in each pulsation cycle, as the gas is cut off before a new set could be formed. The bubble frequency becomes increasingly close to pulsation frequency as pulsation frequency increases. Bubble size decreases with bubble frequency. In light of the oscillating velocity profile, initial gas velocity at the beginning of each “ON” period when bubbles are formed (Uin, m/s) is adopted as the velocity from which bubble size and bubble rise velocity are calculated. In order to obtain Uin, high-speed camera footage was deconstructed into full-resolution uncompressed images frame by frame, resulted images were imported into ImageJ (ver. 1.46r) so that the motion of particles and gas voids (bubbles or slugs) could be tracked. Previous study discovered at the beginning of each cycle, large amount of gas entered the fluidized bed where a horizontal layer of slugs was formed.21 By tracing the movement of the slugs, Uin is shown to be,

U in 

N

j

j 1

0

t N

(25)

where λj (m) is the distance of a void travelled between two adjacent frames, and t0 (s) is the time interval between two adjacent frames of footage. It should be noted for each case Uin was taken from the average of N samples. Since the footage was videotaped at 240 frames per second, 𝑡0 ≈


20

Page 21 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


4.17 × 10−3 s. Substitute UB with Uin in Eqn. (13) and U in Eqn. (18) and the overall heat transfer coefficient h2 (W/(m2·K)) could be calculated. The bed-to-surface heat transfer in fluidized bed is divided into particle convective heat transfer and gas convective heat transfer in this work. The former is achieved by particle packets in contact with the heating surface, the latter is contributed by gas bubbles passing through the heating surface. Particle convection and gas convection are modelled separately as the probe surface is considered to be contacted by packets and bubbles alternately. However, the rising bubble and adjacent particles, such as the ones in its wake and cloud may exchange heat with the bubble phase, therefore affecting the bed-to-surface heat transfer between the heating surface and the bubble by altering the thermal properties of the gas. Such an interphase heat transfer, however, may only affect bed-to-surface heat transfer in a limited region when the particles and bubbles are contacting the heating surface in close sequence, and therefore it was neglected in the current heat transfer model.

4. Results and discussion 4.1 Experimental results 4.1.1 Effect of gas flow rate Previous study of Zhang 25 showed that splitting gas supply into a pulsating stream and a steady stream instead of keeping only the pulsating stream could dampen the ‘shock’ or additional acceleration introduced by pulsed gas flow and undermine the positive effect of pulsation on overcoming channeling, slugging and defluidization in fluidized beds of biomass. Therefore, only


21


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 59

the pulsating stream was used in this study. Experimentally obtained minimum fluidization velocity (Umf) of Douglas fir sawdust at 7% moisture content (dry basis) was used in the following ̅ in relation to Umf. The average superficial figures to express the average superficial gas velocity 𝑈 ̅ was read from the rotameter and adjusted according to pressure and temperature. gas velocity 𝑈 The effects of gas flow rate on heat transfer coefficient at different pulsation frequencies f (Hz) are clearly illustrated in Fig. 4. In addition to f=0.5, 1, 2 and 5 Hz, special attention was paid to f=1.25 Hz due to its proximity to the natural frequency fN=1.29 Hz of the system calculated according to the theory proposed by Hao and Bi 53. The fluidized bed was considered analogous to a secondorder mechanical vibration system and pressure waves represented the output response of such a dynamic system to an external excitation,

fN 

1 2

g H mf

(26)

in which Hmf (m) is the static bed height. It is evident that higher gas flow rate leads to better heat transfer, as the heat transfer coefficient at all frequencies investigated showed the same upward trend. Both f=1 and 1.25 Hz exhibited the highest heat transfer coefficient amongst all, roughly 104 W/(m2·K). However, at f=1.25 Hz the maximum was reached at a lower gas flow rate of 1.0Umf instead of 1.25Umf for 1 Hz, indicating less air is required to reach good gas-solid mixing around natural frequency. This is likely attributed to the resonance effect that occurs in fluidized beds just like that of a mechanical system where higher amplitude of pressure fluctuations is present near the system’s natural frequency. Heat transfer coefficients at f=2 Hz had the same trend but the values were lower than that at 1 Hz and 1.25 Hz, presumably resulted from weaker gas pulsations and lower amplitude of pressure


22

Page 23 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


fluctuations from shortened opening time of the solenoid valve in a pulsation cycle that might have reduced quality of gas-solid mixing. A maximum of 98.6 W/(m2·K) was observed at 1.5Umf. At f=5 Hz, gas pulsations were too weak to lift the entire bed at 0.8–1.2Umf. As a result, channeling and bypassing took place and significantly hindered gas-solid contact and heat transfer. For ̅=1.25–1.5Umf, improved fluidization and better heat transfer was observed. Shown in Fig. 4, the 𝑈 maximum heat transfer coefficient was 96.9 W/(m2·K). For f=0.5 Hz, fluidization was achieved even at a flow rate as low as 0.7 Umf, as a result of strong pulsations. h increased from 89.1 W/(m2·K) at 0.7Umf to 92.6 W/(m2·K) at 1.5Umf. The lower values and flattened profile of heat transfer coefficient at such a low pulsation frequency may be related to the long intermittence of fluidization in the “OFF” period where the bed remained still. Despite higher flow rate and greater amplitude of pressure fluctuations were achieved in the “ON” period with increasing airflow, the effect was then offset by the stagnant “OFF” period. Interestingly, at higher pulsation frequencies investigated (𝑓 ≥ 1 Hz) heat transfer coefficient reached a maximum and plateaued. This may be resulted from the compensation between gas convective and particle convective heat transfer components that co-existed for the particles investigated, as pointed out by Molerus and Wirth 34. As flow rate increases, particle convective component phases out as the bed voidage is getting higher, heat transfer slowly switches to gas convection, which is consistent with the asymptotical increase of heat transfer coefficient.

4.1.2 Effect of pulsation frequency Fig. 5 demonstrates the effect of pulsation frequency on bed-to-surface heat transfer coefficient in the pulsed fluidized bed. The highest heat transfer coefficients among all three flow rates were


23


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 59

seen between 1–1.5 Hz. This is consistent with the findings in the previous section. Generally, higher gas flow rate promotes solid renewal on probe surfaces and gas convective heat transfer. ̅=1.5Umf showed overall higher heat transfer coefficients than 1.3Umf and 1.1Umf. As a result, 𝑈 At f=0.5 Hz pulsed fluidized bed was only fluidized intermittently, and both the “ON” and “OFF” periods were as long as 1 s. Significant pressure build-up occurred during the “OFF” period, followed by the highest amplitude of peak pressure drops as well as standard deviation of pressure fluctuations of the “ON” period, leading to high instantaneous gas flow rate and large yet infrequent bubble generations. A horizontal layer of small bubbles formed above the distributor could be seen once “ON” period initiated. As revealed by Ozkaynak and Chen 54, smaller bubbles in shallow beds could easily miss the heating tubes, resulting in low heat transfer coefficients. As the smaller bubbles rose they quickly coalesced into large bubbles and slugs, carrying many particles upward to the freeboard, which explains the severe entrainment at 0.5 Hz. The timeaverage h remained low due to the less frequent bubble generation of the “ON” period and the ̅=1.1Umf, 1.3Umf and 1.5Umf at 0.5 Hz, the respective heat transfer prolonged “OFF” period. For 𝑈 coefficients were 86.8, 90.6 and 93.9 W/(m2·K). As pulsation frequency increased from 0.5 Hz to 1 Hz, both the time fraction and absolute length of “OFF” periods diminished, and the fluidized bed gradually shifted from a seemingly intermittent fluidization regime to a more continuous one. As bubble generation is deeply linked with gas pulsation, higher pulsation frequency usually leads to smaller and more frequent bubbles, which in return promotes solid circulation and particle renewal on the heat transfer surface. For 1.3Umf an increase of h as much as 10% could be seen in Fig. 5.


24

Page 25 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Maximum heat transfer coefficients for all three gas flow rates were achieved between 1 and 1.5 ̅=1.5Umf. Interestingly this frequency range also Hz, with the maximum being 103.5 W/(m2·K) at 𝑈 coincides with the natural frequency of fN=1.29 Hz, predicted by Eqn. (26). Since fluidized beds could be seen as a second-order mechanical vibration system

53

, when the external excitation

(pulsed gas flow) approaches the natural frequency, resonance effect takes place, leading to enhanced gas-solid contact, higher heat transfer rates between the packets and heating surfaces. Around the natural frequency, the oscillation of the bed becomes sustained by the input pulsation with lowest dissipation, overcoming the cohesive forces between biomass particles and preventing channeling and partial defluidization. Further increasing pulsation frequency beyond fN may lead to faster bubble generation, which will likely benefit the particle convective heat transfer as packets are able to come into contact with the heating surface and get renewed more frequently, leading to a shorter residence time. Meanwhile, gas convection will take a toll due to the reduced bubble rise velocity in spite of an increase in gas holdup. The interaction between gas convective and particle convective heat transfer and their similarity in magnitude might have contributed to the rather flat profile of heat transfer coefficient at 1.5Umf as the dominating heat transfer mechanism switched from one to the other. On average, h=100.1±0.4 W/(m2·K) between 2 and 5 Hz. At 1.1Umf and 1.3Umf, operating above fN could no longer generate gas pulsations powerful enough to lift the entire bed. Consequently, channeling and bypassing occasionally took place, which may have reduced the gas-solid contact efficiency and undermined the bed-to-surface heat transfer. For 1.3Umf heat transfer coefficient dropped from 96.4 W/(m2·K) at 2 Hz to 90.9 W/(m2·K) at 5 Hz. Similar trend also existed at 1.1Umf. It should be noted that data point for 1.1Umf at 5 Hz was excluded because complete defluidization occurred during the experiment and it could no longer be considered as fluidized bed.


25


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 59

4.1.3 Heat transfer properties of pine and switchgrass Besides Douglas fir, heat transfer coefficients of pine and switchgrass powders were also investigated. Average superficial gas velocities equivalent to 1.1Umf, 1.3Umf and 1.5Umf of Douglas fir sawdust (with 7% moisture content) were used to fluidize pine sawdust and switchgrass powders so that results could be better compared. The respective gas velocities were 0.26, 0.31 and 0.36 m/s. Produced from the same sawmill, pine sawdust possesses similar size, thermal conductivity and heat capacity to Douglas fir sawdust. Similar trend of heat transfer coefficients is illustrated in Fig. 6(a). Heat transfer coefficients increased with gas flow rate. The highest heat transfer coefficient ̅=0.36 m/s with a pulsation frequency of 1.5 Hz. Just like Douglas fir, at was 103.9 W/(m2·K) at 𝑈 low pulsation frequencies such as 0.5 and 1 Hz, time-averaged heat transfer coefficient was low due to the long and dormant “OFF” periods. Further increasing flow rate to 0.36 m/s, similar h values were found across f=2–4 Hz as the increase in particle convection offset the decline in gas convection, until channeling occurred at 5 Hz that led to a decline in gas-solid contact efficiency ̅=0.26 m/s and 0.31 m/s produced a maximum of h at f=1.5 Hz as well, and heat transfer. Both 𝑈 ̅=0.26 and 0.31 m/s, the bed behaved like a fixed albeit the values were lower. Above 3 Hz for 𝑈 bed during the experiment due to severe channeling followed by defluidization. Therefore, data above this point were excluded. Compared to pine and Douglas fir, switchgrass particles used in this study had a smaller Sauter mean diameter of 0.76 mm and much more fines. The smaller size and higher percentage of fines are possible contributing factors to the higher particle convective heat transfer because of the


26

Page 27 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


shorter residence time of smaller particles. Consequently, in addition to the normal peak of h ̅=0.31 m/s observed around fN in Fig. 6(b), another more prominent peak was seen at 3 Hz for 𝑈 and 0.36 m/s, indicating the increase in particle convection outweighed the decrease in gas convection. For the lower flow rate of 0.26 m/s, only one peak existed around fN, which matched with visual observations made during the experiments where channeling occurred, affecting mixing and heat transfer.

4.1.4 Effect of particle size Molerus and Wirth

34

analyzed experimental results from Wunder

55

where heat transfer

coefficients of glass beads of four different sizes (55, 400, 770 and 2230 µm) had been measured at different superficial gas velocities and concluded that for glass beads of 55 and 2230 µm the predominant heat transfer mechanisms were particle convection and gas convection respectively, whereas for 400 and 770 µm particles heat transfer shifted from particle convection towards gas convection as flow rate increased. It is widely acknowledged that within bubbling regime an asymptotic relationship between excess superficial gas velocity and heat transfer coefficient only occurs when a single heat transfer mechanism is dominating, otherwise heat transfer coefficient will most likely go through a maximum and will eventually decline. In order to test the effect of particle size on heat transfer, Douglas fir sawdust received from the sawmill was separated into different particle size fractions. After being dried sawdust was sieved through a series of screens (7 mm, 4 mm, 3.15 mm and 1 mm) where particles larger than 7 mm were discarded and the following particle fractions were obtained: 0–1 mm, 1–3.15 mm, 3.15–4 mm and 4–7 mm. For particles of 4–7 mm, fluidization was not achieved as the bed was completely defluidized from the beginning. The bed behaved like a fixed bed except a slight piston-like movement where the


27


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 59

bed as a whole was lifted up by a few millimeters due to gas pulsations. Heat transfer suffered as a result, which are demonstrated in Fig. 7. The bed-to-surface heat transfer coefficient increased slightly with increasing superficial gas velocity, from 39.9 W/(m2·K) at 0.25 m/s to 54.3 W/(m2·K) at 0.38 m/s. For particles of 3.15–4 mm at low gas flow rates such as 0.25 and 0.28m/s, a hole formed at the center of the bed, only a small number of particles around it were carried up and down by the gas pulsation, with the rest of the bed remaining stationary. As a higher superficial gas velocity such as 0.38 m/s, particles were able to get partially fluidized accompanied by severe channeling, leading to a gradual increase in heat transfer coefficient. A maximum of 60.0 ̅=0.25 m/s the center of the bed was empty for W/(m2·K) was reached at 0.38 m/s. Similarly, at 𝑈 particles of 1–3.15 mm. At 0.28 m/s the system was partially fluidized. Very good fluidization was restored when superficial gas velocity increased to 0.38 m/s. Therefore, heat transfer coefficient increased by 41%, from 56.3 W/(m2·K) at 0.25 m/s to 79.6 W/(m2·K) at 0.38 m/s. Highest heat transfer coefficient was discovered within 0–1 mm particles. Due to the small size, smooth fluidization was achieved with no channeling and gas bypassing, even at f=5 Hz. The presence of finer particles also resulted in more entrainment when compared to the other particle size fractions. Nevertheless, heat transfer coefficient for 0–1 mm particles almost doubled compared to the 1– ̅ =0.28 m/s. The 0–1 mm also 3.15 mm fraction, with a maximum of 117.1 W/(m2·K) at 𝑈 highlighted the benefit of a fluidized bed.

4.1.5 Effect of fine particles The presence of fine particles in the 0–1 mm particle fraction might have contributed to the smooth fluidization, gas-solid contact as well as heat transfer. Agarwal, et al. 56 studied the effect of fines


28

Page 29 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


on the hydrodynamics of fluidized beds and compared the minimum fluidization velocity for closely sized fractions of iron ore to that with a wide size range, and discovered that the greater the “fines” content, the lower the minimum fluidization velocities. Sun

57

showed that when

superficial gas velocity exceeded 0.1 m/s, the bed with wider particle size distribution usually gave higher particle concentration inside the voids (bubbles or slugs), the smallest bubble size and the greatest dense phase expansion at the same operating conditions leading to a higher reaction rate and improved interfacial heat and mass transfer. In order to further investigate the possible effect of fine particles, or fines on heat transfer, fine Douglas fir sawdust (0–1 mm) was added to the coarse fraction (3.15–4 mm) at certain ratios. Four mixtures were prepared with the weight percentage of fine particles ranging from 33% to 83%. Details of the mixtures are summarized in Table 3. For Group 1 containing 33% of fines, severe channeling occurred at the lowest superficial gas velocity of 0.28 m/s, and the center of the fluidized bed was quickly emptied, followed by complete defluidization. The heating probe was exposed to the rising airflow with only a few particles being carried up and down by the pulsed gas flow. As shown in Fig. 8, the measured heat transfer coefficient was very low (67.5 W/(m2·K)), which was not a good representation of the bed-tosurface heat transfer of the fluidized bed. At higher gas velocities, the fluidized bed still suffered from occasional defluidization, but was able to recover due to the increased intensity of the pulsed ̅=0.38 m/s was measured to be 77.0 W/(m2·K). Gas-solid gas flow. Heat transfer coefficient at 𝑈 ̅=0.26 contact was slightly improved in Group 2, due to the presence of more fines. However, at 𝑈 m/s and 0.28 m/s, the center region of the bed surrounding the heating probe was also emptied, with only a small portion of particles being carried up and down. No obvious bed expansion and gas-solid contact could be observed. The measured h values only reflected the forced convection


29


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 59

of fluidizing air around the probe. Similar to Group 1, at a higher flow rate, fluidization was ̅=0.38 m/s. reached. Shown in Fig. 8, the heat transfer coefficient was 85.1 W/(m2·K) at 𝑈 Much improved mixing was seen with the increased finer particle content in Group 3. No channeling was observed with this group. Even at a low gas velocity of 0.26 m/s, h reached 96.5 W/(m2·K). It appears that further increasing the gas flow rate did not benefit the bed-to-surface heat transfer, as h only increased slightly to 99.9 W/(m2·K) at 0.31 m/s and fell back to 96.8 W/(m2·K) at 0.38 m/s. The highest heat transfer coefficients were found in Group 4 containing 83% of fines. It seems that by adding fine particles into coarse ones both stability and flowability of the latter were improved, in addition to improved gas-solid contact and heat transfer. One possible explanation is the lubrication effect. By examining earlier experimental work, Zenz and Othmer

58

analyzed the

effects of particle size distribution on the pseudo viscosity of fluidized beds and concluded that the fines between the coarse particles could act as a lubricant to reduce the friction between the coarse particles, thereby reducing the bed viscosity and helping to maintain good flowability. Numerical studies by Norouzi, et al. 59 also showed that by adding fines coarse particles were more readily to interact with each other, which resulted in enhanced mobility of larger particles.

4.1.6 Effect of vibration Mechanical vibration has been widely utilized in fluidization of coarse particles,60, 61 drying62, 63 and spray coating64, 65 likely due to its ability to overcome cohesive forces such as the van der Waals force and liquid bridge force. Our previous study demonstrated that mechanical vibration to a pulsed fluidized bed was only beneficial when standalone gas pulsation was insufficient in


30

Page 31 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


breaking up the cohesive forces between particles, especially for particles with high moisture content.21 In this work, the effect of vibration was tested at two different superficial gas velocities ̅=1.05Umf across five pulsation frequencies, as portrayed in Fig. 9(a). The average increase of h at 𝑈 was 4.7% after the introduction of mechanical vibration. As the pulsation frequency increased, gas pulsation can no longer life the entire bed, causing channeling and partial defluidization. By turning on vibration good gas-solid contact was restored and h increased by 6.5% at 5 Hz. ̅=1.3Umf, pulsation was more than enough to maintain good gas-solid With a higher flow rate of 𝑈 contact. Therefore, an average increase of 2.3 % that was less than that of 1.05Umf was observed in Fig. 9(a). The biggest improvement occurred at the lowest pulsation frequency of 0.625 Hz where vibration mobilized the otherwise stationary particles during the “OFF” period, therefore slightly enhanced heat transfer. The high heat transfer coefficient in fluidized beds is attributed to the particle motion, as explained by Botterill and Williams 66. Eccles 67 further explained that mechanical vibration could enhance the contact between the heating surface and the packets, leading to a higher instantaneous heat transfer coefficient. Muchowski

68

attributed the higher bed-to-surface heat transfer in vibrated

beds to the accelerated packet renewal on the heater surface and faster removal of heat from the heater zone. Malhotra and Mujumdar 69 later suggested that the increased heat transfer in vibrated fluidized bed was related to both the change in porosity and the particle mobility near the heating surface, and the enhancement from faster solid renewal outweighed the negative effect of increased bed porosity. To roughly quantify the possible contributions of vibration in gas-solid contact efficiency and heat transfer of biomass, Fast Fourier Transformation (FFT) analysis was performed on the differential


31


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 59

bed pressure signals. In gas-solid fluidized beds, the power spectrum of pressure fluctuations reveals bubble frequency, including bubble formation, splitting and coalescence. In pulsed fluidized beds, on the other hand, it is fairly common that there exists a dominant frequency corresponding to the pulsation frequency. As shown in Fig. 10, peaks are more distinct for pulsed fluidized bed, and several orders of magnitude higher than that of a conventional fluidized bed. The dominant frequency was 1.25 Hz, with overtones at 2.5 Hz, 3.75 Hz and so on that could be attributed to the under-dampened pressure oscillations, and the fact that the pressure fluctuations in pulsed fluidized bed were not perfectly sinusoids. Besides the peaks associated with the dominant frequency of 1.25 Hz and its overtones, a few lower peaks were seen around 60 Hz, which corresponded to the frequency of the vibratory base. The root mean square (RMS) amplitude of the vibration peak (60 Hz) was only one tenth of the pulsation peak (1.25 Hz). By increasing the amplitude or frequency of the vibratory base, the limited contribution of vibration could be further enhanced.

4.1.7 Effect of pulsation duty cycle In previous experiments the “ON” and “OFF” period in a pulsation cycle were kept equal, with a ratio of one. However, the ratio between “ON” and “OFF” periods may alter the pressure buildup, bubble formation and bed behavior. The effect of duty cycle, as defined in Eqn. (27), was then studied,



tON tON  tOFF

(27)


32

Page 33 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


tON (s) and tOFF (s) in Eqn. (27) represent the open and close time intervals of the solenoid valve in a pulsation cycle, respectively. Bed-to-surface heat transfer at a wide range of duty cycles from 25% to 75% was investigated at three different gas flow rates, 1.03𝑈𝑚𝑓 , 1.25𝑈𝑚𝑓 and 1.41𝑈𝑚𝑓 . While the pulsation frequency ( 𝑓 = 1.25 Hz) and average gas flow rate were maintained at constants, the “ON” period within each pulsation cycle was set to 0.25seconds (𝜂 = 25%), 0.40 seconds (𝜂 = 50%) and 0.60 seconds (𝜂 = 75%) for the heat transfer measurements of Douglas fir, with the results being demonstrated in Fig. 9(b). Compared to 50% duty cycle used in previous experiments, the shorter “ON” period, with solenoid valve open for airflow, at 𝜂 = 25% allowed higher pressure build-up during the “OFF” period. The instantaneous gas flow rate when the valve was turned “ON” was higher than that at 𝜂 = 50%, which resulted in a stronger gas pulsation and faster formation of smaller gas bubbles. By increasing the pulsation duty cycle, the same amount of gas entered the fluidized bed over a longer open period, resulting in lower instantaneous gas velocity. Pulsation also became weaker due to the shorter OFF period for pressure to build-up, leading to the formation of larger bubbles. At 1.03 𝑈𝑚𝑓 , pulsation became too weak at 𝜂 > 50% , partial defluidization occurred, which significantly hindered the gas-solid contact and heat transfer. The bed-to-surface heat transfer coefficient decreased from 94.7 W/(m2·K) at 𝜂 = 25% to 90.0 W/(m2·K) 𝜂 = 50%. The bed completely defluidized at 𝜂 = 75%. Increasing flow rates to 1.25𝑈𝑚𝑓 and 1.41𝑈𝑚𝑓 yielded higher heat transfer coefficients. At 𝜂 = 25% pulsation was very intense, but the shorter “ON” period increased the stagnant period of the fluidized bed over the OFF period and limited the gas-solid interaction. Therefore, heat transfer coefficients were lower than that at 𝜂 = 50%. At 𝜂 = 75% mild channeling was observed for both


33


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 34 of 59

U=1.25𝑈𝑚𝑓 and U=1.41𝑈𝑚𝑓 , which slightly reduced the heat transfer rates, the respective heat transfer coefficients were 97.1 W/(m2·K) and 101.9 W/(m2·K). The maximum heat transfer coefficients were obtained at 𝜂 = 50%, which were 101.7 W/(m2·K) and 103.8 W/(m2·K) for 1.25𝑈𝑚𝑓 and 1.41𝑈𝑚𝑓 , respectively. Duty cycle could be varied to better tackle the cohesiveness of biomass particles. With wet or coarse particles, duty cycle could be reduced so that stronger pulsation is generated to achieve smooth fluidization without channeling and partial defluidization, with a trade-off in the heat transfer rate. With dry or finer particles, a duty cycle around 50% could be used to increase the heat and mass transfer rates.

4.2 Comparison with the heat transfer model ̅ = 1.1𝑈𝑚𝑓 from both heat transfer Heat transfer coefficients in the rectangular fluidized bed at 𝑈 models were derived and plotted against the experimental data shown in Fig. 11(a). Both models seemed capable of predicting heat transfer coefficients in their designated frequency ranges. The overestimation of heat transfer coefficients in Model 1 for f > fN was likely attributable to the fact that above fN “OFF” period started to disappear, the bed could no longer be divided into two distinct heat transfer processes. In the second model designated for f > fN good agreement with the experimental data was achieved, and the slight overestimation at higher pulsation frequencies was presumably linked to some mild channeling and bypassing occurred at such frequencies, where gas pulsations were simply not strong enough to penetrate the entire bed. Apply such a model to f < fN will likely produce higher


34

Page 35 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


h2 values because the bed is no longer fluidized continuously, and the increasing “OFF” period should now be taken into consideration. Natural frequency of the pulsed fluidized bed system was calculated based on Eqn. (26). Around fN predicted heat transfer coefficients from Model 1 and Model 2 were very close, so that heat transfer coefficient could be calculated from either model for f=fN. Consequently, combining both models good prediction of heat transfer coefficient in pulsed fluidized bed is obtained,

h h 1 h2

f  fN f  fN

(28)

Above model was also verified at a higher average superficial gas velocity of 1.4Umf as shown in Fig. 11(b).

5. Conclusions Gas pulsation is highly effective in overcoming difficulties associated with the unique natures of biomass particles during fluidization. Biomass particles including Douglas fir, pine and switchgrass were successfully fluidized in a pulsed fluidized bed without the need of inert bed particles. Bed-to-surface heat transfer coefficient was investigated as an essential step towards utilizing pulsed fluidized bed as a potential reactor for biomass drying, torrefaction and pyrolysis. The effect of gas flow rate and pulsation frequency was thoroughly researched. The research shows heat transfer coefficient generally increases with flow rate. In the case of pulsation, an optimum frequency exists, which is close to the natural frequency of the pulsed fluidized bed system. At such a frequency gas pulsations are strong enough to lift the entire bed without too much


35


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 36 of 59

entrainment and enable the pulsed gas flow to overcome channeling and gas bypassing. Meanwhile, the pulsations are frequent enough so that the motionless “OFF” period is eliminated. Together with the resonance effect, highest bed-to-surface heat transfer coefficient was observed. The compensation between particle convective and gas convective heat transfer is also more complex than conventional fluidized beds because instantaneous superficial gas velocity varies not only with the overall gas flow rate, but also with pulsation frequency. A low pulsation frequency will likely generate a very strong instantaneous gas flow at the beginning of “ON” period, which results in a high gas convective heat transfer. On the other hand, at higher pulsation frequencies, the faster bubble generation and solid circulation enhances particle convection. Based on the observation, the heat transfer was modeled as comprised of two different mechanisms, divided at the natural frequency of the system. The combined model showed good agreement with experimental heat transfer data. In addition, the influence of particle size distribution as well as fines content on heat transfer were investigated. Results suggest that the lubrication effect of fines promotes movement of coarse particles and enhances heat transfer, which could explain the higher heat transfer coefficient of fine switchgrass particles. Pine on the other hand, exhibited fairly similar heat transfer properties to Douglas fir. The addition of vibration is only optional, as it offered marginal improvement in gas-solid mixing and heat transfer. Dead zones at the bottom of the column was observed during the experiment, and the elimination of which may further improve solid mixing and heat transfer that should be investigated in the future.


36

Page 37 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Acknowledgement The authors are grateful to the Natural Sciences and Engineering Research Council of Canada (NSERC) for the financial support in the form of a Canada-Japan Joint Strategic Program (Grant No. STPGP430177-12), Tolko Industries Ltd. for the generous donation of biomass samples, and Noram Engineering and Constructors for its in-kind contribution to the project.


37


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 38 of 59

Nomenclature A

Cross-section area of the fluidized bed column, m2

As

Heat transfer area of the heat transfer measurement probe, m2

Ar

Archimedes number

Cpg

Heat capacity of the gas phase, J/(mol·K)

Cp, mf Dense phase heat capacity, J/(mol·K) Cpp

Heat capacity of the solid, J/(kg·K)

db

Bubble diameter, m

db0

Initial bubble diameter at multi-orifice distributor plate, m

dbm

Maximum attainable bubble diameter, m

dp

Particle diameter, m

Ds

Diameter of the heat transfer measurement probe, m

Dt

Hydraulic diameter of the column, m

f

Pulsation frequency, Hz

h

Heat transfer coefficient, W/(m2·K)

h1

Heat transfer coefficient from Model 1, W/(m2·K)

h2

Heat transfer coefficient from Model 2, W/(m2·K)

hc

Conductive-convective heat transfer coefficient, W/(m2·K)

hgc

Gas convective of heat transfer, W/(m2·K)

hi

Instantaneous heat transfer coefficient, W/(m2·K)

hoff

Heat transfer coefficient during the “OFF” period, W/(m2·K)

hon

Heat transfer coefficient during the “ON” period, W/(m2·K)

hp

Average heat transfer coefficient between dense phase and surface, W/(m2·K)


38

Page 39 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


hpc

Particle convective heat transfer component, W/(m2·K)

hs

Integral mean gas convective heat transfer coefficient, W/(m2·K)

hw

Wall heat transfer coefficient in fixed beds, W/(m2·K)

I

Current, A

j

Integer

keff

Effective thermal conductivity of fixed bed, W/(m·K)

kef

Turbulent eddy conduction or lateral mixing component of Keff, W/(m·K)

kes

Effective thermal conductivity of fixed bed with motionless fluid, W/(m·K)

kg

Thermal conductivity of gas, W/(m·K)

kmf

Thermal conductivity of dense phase, W/(m·K)

kp

Thermal conductivity of solid, W/(m·K)

l

Bubble/slug length, m

lt

Turbulent length scale, m

L

Characteristic vertical dimension of heater, m

N

Number of bubbles/slugs

Nor

Number of orifices in distributor

Nup

Particle phase Nusselt number, ℎ𝐿/ 𝑘𝑝

Prg

Particle phase Prandtl number, 𝐶𝑝𝑝 𝜇/ 𝑘𝑔

Ql

Heat loss, W

Rep

Reynolds number of particles, 𝜌𝑔 𝑑𝑝 𝑈 / 𝜇𝑔

t

Time, s

t0

Time interval between each frame of footage, s

tON

Opening time of the solenoid valve, s


39


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

tOFF

Closed time of the solenoid valve, s

T

Temperature, K

Tb

Temperature of the dense phase, K

Ts

Temperature of the probe surface, K

U

Superficial gas velocity, m/s

̅ 𝑈

Average superficial gas velocity, m/s

𝑈𝑏

Bubble rise velocity, m/s

𝑈𝑖𝑛

Initial gas velocity at the beginning of “ON” period, m/s

𝑈𝑚𝑓

Minimum fluidization velocity, m/s

̅𝑜𝑛 𝑈

Average superficial gas velocity in the “ON” period, m/s

V

Voltage, V

z

Height above gas distributor, m

Page 40 of 59

Greek Letters β

Time fraction of the “ON” period

ε

Voidage fraction

εb

Bubble volume fraction

εmf

Bed voidage at minimum fluidization

η

Pulsation duty cycle

λ

distance of a void/bubble travelled, m

μ

Viscosity, kg/(m·s)

μg

Gas viscosity, kg/(m·s)

ρmf

Density of the dense phase, kg/m3


40

Page 41 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ρp

Particle density, kg/m3

ρg

Air density, kg/m3



Time of contact between dense phase and surface, s


41


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 42 of 59

References (1) Cambero, C.; Sowlati, T.; Pavel, M., Economic and life cycle environmental optimization of forestbased biorefinery supply chains for bioenergy and biofuel production. Chem. Eng. Res. Des. 2016, 107, 218-235. (2) Mabee, W.; Saddler, J., Bioethanol from lignocellulosics: Status and perspectives in Canada. Bioresour. Technol. 2010, 101, 4806-4813. (3) Goransson, K.; Soderlind, U.; Engstrand, P.; Zhang, W., An experimental study on catalytic bed materials in a biomass dual fluidised bed gasifier. Renewable Energy 2015, 81, 251-261. (4) Han, J.; Kim, H.; Cho, S.; Shimizu, T., Fluidized bed combustion of some woody biomass fuels. Energy Sources Part a-Recovery Utilization and Environmental Effects 2008, 30, 1820-1829. (5) Jia, D.; Bi, X. T.; Lim, C. J.; Sokhansanj, S.; Tsutsumi, A., Biomass drying in a pulsed fluidized bed without inert bed particles. Fuel 2016, 186, 270-284. (6) Wong, H. W.; Baird, M. H. I., Fluidisation in a pulsed gas flow. Chem. Eng. J. 1971, 2, 104-113. (7) Akhavan, A.; van Ommen, J. R.; Nijenhuis, J.; Wang, X. S.; Coppens, M.-O.; Rhodes, M. J., Improved Drying in a Pulsation-Assisted Fluidized Bed. Ind. Eng. Chem. Res. 2008, 48, 302-309. (8) Nitz, M.; Taranto, O. P., Drying of beans in a pulsed fluid bed dryer: Drying kinetics, fluid-dynamic study and comparisons with conventional fluidization. J. Food Eng. 2007, 80, 249-256. (9) Kudra, T.; Gawrzynski, Z.; Glaser, R.; Stanislawski, J.; Poirier, M., DRYING OF PULP AND PAPER SLUDGE IN A PULSED FLUID BED DRYER. Drying Technol. 2002, 20, 917-933. (10) Reyes, A.; Mahn, A.; Guzmán, C.; Antoniz, D., Analysis of the Drying of Broccoli Florets in a Fluidized Pulsed Bed. Drying Technol. 2012, 30, 1368-1376. (11) Reyes, A.; Campos, C.; Vega, R., Drying of Turnip Seeds with Microwaves in Fixed and Pulsed Fluidized Beds. Drying Technol. 2006, 24, 1469-1480. (12) Massimilla, L.; Volpicelli, G.; Raso, G., A study on pulsating gas fluidization of beds of particles. In Chem. Eng. Prog., Symp. Ser., American Institute of Chemical Engineers: 1966; Vol. 62 p63. (13) Kobayashi, M.; Ramaswami, D.; Brazelton, W. In Pulsed-Bed Approach to Fluidization, Chem. Eng. Prog., Symp. Ser., 1970; Argonne National Laboratory: 1970; pp 47-57. (14) Wang, X. S.; Rhodes, M. J., Pulsed fluidization—a DEM study of a fascinating phenomenon. Powder Technol. 2005, 159, 142-149. (15) Wu, K.; de Martín, L.; Mazzei, L.; Coppens, M.-O., Pattern formation in fluidized beds as a tool for model validation: A two-fluid model based study. Powder Technol. 2016, 295, 35-42. (16) Li, H.; Liu, X.; Legros, R.; Bi, X. T.; Lim, C. J.; Sokhansanj, S., Torrefaction of sawdust in a fluidized bed reactor. Bioresour. Technol. 2012, 103, 453-458. (17) Prins, M. J.; Ptasinski, K. J.; Janssen, F. J. J. G., More efficient biomass gasification via torrefaction. Energy 2006, 31, 3458-3470. (18) Mastellone, M. L.; Arena, U., Fluidized-bed pyrolysis of polyolefins wastes: Predictive defluidization model. AlChE J. 2002, 48, 1439-1447. (19) Bi, X. T.; Liu, X., High density and high solids flux CFB risers for steam gasification of solids fuels. Fuel Process. Technol. 2010, 91, 915-920. (20) Jia, D.; Bi, X.; Lim, C. J.; Sokhansanj, S.; Tsutsumi, A., Biomass drying in a pulsed fluidized bed without inert bed particles. Fuel 2016, 186, 270-284. (21) Jia, D.; Cathary, O.; Peng, J.; Bi, X.; Lim, C. J.; Sokhansanj, S.; Liu, Y.; Wang, R.; Tsutsumi, A., Fluidization and drying of biomass particles in a vibrating fluidized bed with pulsed gas flow. Fuel Process. Technol. 2015, 138, 471-482. (22) Jia, D.; Bi, X.; Lim, C. J.; Sokhansanj, S.; Tsutsumi, A., Gas-solid mixing and mass transfer in a tapered fluidized bed of biomass with pulsed gas flow. Powder Technol. (23) Zabrodskii, S.; Bokun, I., Heat Transfer between a Pulsating Bed of Moist Tetracycline and the Heating Surface. Argonne National Lab.: 1966.


42

Page 43 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(24) Kobayashi, M.; Ramaswami, D.; Brazelton, W. In Heat transfer from an internal surface to a pulsed bed, Chem. Eng. Prog. Symp. Ser, 1970; 1970; pp 58-67. (25) Zhang, D. Hydrodynamics and heat transfer in a pulsed bubbling fluidized bed. M.A.Sc., Dalhousie University (Canada), Ann Arbor, 2005. (26) Nishimura, A.; Deguchi, S.; Matsuda, H.; Hasatani, M.; Mujumdar, A. S., Heat transfer characteristics in a pulsating fluidized bed in relation to bubble characteristics. Heat Transfer—Asian Research 2002, 31, 307-319. (27) Sun, P.; Hui, S. e.; Gao, Z.; Zhou, Q.; Tan, H.; Zhao, Q.; Xu, T., Experimental investigation on the combustion and heat transfer characteristics of wide size biomass co-firing in 0.2 MW circulating fluidized bed. Appl. Therm. Eng. 2013, 52, 284-292. (28) Kalita, P.; Saha, U. K.; Mahanta, P., Effect of biomass blending on hydrodynamics and heat transfer behavior in a pressurized circulating fluidized bed unit. Int. J. Heat Mass Transfer 2013, 60, 531-541. (29) Papadikis, K.; Gu, S.; Bridgwater, A. V., Computational modelling of the impact of particle size to the heat transfer coefficient between biomass particles and a fluidised bed. Fuel Process. Technol. 2010, 91, 68-79. (30) Zhang, L.; Hou, J.; Bi, X. T.; Grace, J. R.; Janke, T.; Arato, C., Fluidization characteristics and charging behavior of fly ash in a vibro-fluidized bed. Powder Technol. 2012, 215–216, 235-241. (31) Tannous, K.; Lam, P. S.; Sokhansanj, S.; Grace, J. R., Physical Properties for Flow Characterization of Ground Biomass from Douglas Fir Wood. Part. Sci. Technol. 2013, 31, 291-300. (32) Riley, N. A., Projection sphericity. J. Sediment. Res. 1941, 11, 94-95. (33) Borodulya, V. A.; Teplitsky, Y. S.; Markevich, I. I.; Hassan, A. F.; Yeryomenko, T. P., Heat transfer between a surface and a fluidized bed: consideration of pressure and temperature effects. Int. J. Heat Mass Transfer 1991, 34, 47-53. (34) Molerus, O.; Wirth, K.-E., Heat Transfer in Fluidized Beds. First Edition ed.; Chapman & Hall: Dordrecht, 1997. (35) Mickley, H. S.; Fairbanks, D. F., Mechanism of heat transfer to fluidized beds. AlChE J. 1955, 1, 374-384. (36) Danckwerts, P. V., Significance of Liquid-Film Coefficients in Gas Absorption. Ind. Eng. Chem. 1951, 43, 1460-1467. (37) Kay, J. M.; Nedderman, R. M., An Introduction to Fluid Mechanics and Heat Transfer with Applications in Chemical and Mechanical Process Engineering. Cambridge University Press: 1974; p 340. (38) Mori, S.; Wen, C., Estimation of bubble diameter in gaseous fluidized beds. AlChE J. 1975, 21, 109-115. (39) Grace, J. R., Fluidized bed hydrodynamics. Handbook of multiphase systems 1982, 5. (40) Yagi, S.; Kunii, D., Studies on heat transfer near wall surface in packed beds. AlChE J. 1960, 6, 97-104. (41) Kunii, D.; Smith, J., Heat transfer characteristics of porous rocks. AlChE J. 1960, 6, 71-78. (42) Deissler, R.; Boegli, J., An investigation of effective thermal conductivities of powders in various gases. ASME Transactions 1958, 8, 1417-1425. (43) Dietz, P. W., Effective thermal conductivity of packed beds. Ind. Eng. Chem. Fundam. 1979, 18, 283-286. (44) Ranz, W., Friction and transfer coefficients for single particles and packed beds. Chem. Eng. Prog. 1952, 48, 247-253. (45) Kimura, M.; Nono, K.; Kaneda, T., Distribution of void in a packed tube. J. Chem. Eng. Jpn. 1955, 14, 397-400. (46) Baskakov, A. P.; Berg, B. V.; Vitt, O. K.; Filippovsky, N. F.; Kirakosyan, V. A.; Goldobin, J. M.; Maskaev, V. K., Heat transfer to objects immersed in fluidized beds. Powder Technol. 1973, 8, 273-282. (47) Gabor, J. D., Heat transfer to particle beds with gas flows less than or equal to that required for incipient fluidization. Chem. Eng. Sci. 1970, 25, 979-984. (48) Catipovic, N. M. Heat transfer to horizontal tubes in fluidized beds: experiment and theory. Oregon State University, Corvallis, OR, 1979.


43


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 44 of 59

(49) Calderbank, P. H.; Pogorski, L., Heat transfer in packed beds. Trans. Inst. Chem. Eng 1957, 35, 195-207. (50) Denloye, A. O. Heat transfer in packed and gas fluidized beds. Ph.D., University of Birmingham (United Kingdom), Ann Arbor, 1976. (51) Baskakov, A., The mechanism of heat transfer between a fluidized bed and a surface. Int. Chem. Eng 1964, 4, 320. (52) Kruggel-Emden, H.; Vollmari, K., Flow-regime transitions in fluidized beds of non-spherical particles. Particuology 2016, 29, 1-15. (53) Hao, B.; Bi, H. T., Forced bed mass oscillations in gas–solid fluidized beds. Powder Technol. 2005, 149, 51-60. (54) Ozkaynak, T. F.; Chen, J. C., Emulsion phase residence time and its use in heat transfer models in fluidized beds. AlChE J. 1980, 26, 544-550. (55) Wunder, R. Wärmeübergang an vertikalen Wärmeaustauscherflächen in Gas-swirbelschichten. Technische Universität München, 1980. (56) Agarwal, J.; Davis, W.; King, D., Fluidized bed coal dryer. Chem. Eng. Prog. 1962, 58, 85-90. (57) Sun, G. Influence of particle size distribution and the performance of fluidized bed reactors. Ph.D., The University of British Columbia (Canada), Ann Arbor, 1991. (58) Zenz, F. A.; Othmer, D. F., Fluidization and fluid-particle systems. Reinhold: 1960. (59) Norouzi, H. R.; Mostoufi, N.; Sotudeh-Gharebagh, R., Effect of fines on segregation of binary mixtures in gas–solid fluidized beds. Powder Technol. 2012, 225, 7-20. (60) Kong, X.; Kang, X.; Li, B.; Song, Z. In Analysis of dynamic behavior of a large-sized vibration fluidized bed, 2nd International Conference on Advances in Product Development and Reliability, PDR'2010, July 28, 2010 - July 30, 2010, Shenyang, China, 2010; Trans Tech Publications: Shenyang, China, 2010; pp 912-915. (61) Jin, H.; Zhang, J.; Zhang, B., The effect of vibration on bed voidage behaviors in fluidized beds with large particles. Braz. J. Chem. Eng. 2007, 24, 389-397. (62) Daleffe, R. V.; Ferreira, M. C.; Freire, J. T., Drying of Pastes in Vibro-Fluidized Beds: Effects of the Amplitude and Frequency of Vibration. Drying Technol. 2005, 23, 1765-1781. (63) Hasatani, M.; Itaya, Y.; Miura, K., Drying of granular materials in an inclined vibrated fluidized bed by combined radiative and onvective heating. Drying Technol. 1991, 9, 349-366. (64) Habibi, M.; Eslamian, M.; Soltani-Kordshuli, F.; Zabihi, F., Controlled wetting/dewetting through substrate vibration-assisted spray coating (SVASC). J. Coat. Technol. Res. 2016, 13, 211-225. (65) Hayes, M. Method of coating a medical appliance utilizing vibration, a system for vibrating a medical appliance, and a medical appliance produced by the method. US20060216403A1, 2006. (66) Botterill, J.; Williams, J., The mechanism of heat transfer to gas-fluidized beds. Trans. Inst. Chem. Eng 1963, 41, 217-230. (67) Eccles, E. R., Flow and heat transfer phenomena in aerated vibrated beds. Drying Technol. 1990, 8, 895-898. (68) Muchowski, E., Heat transfer from heated surfaces to spherical parkings of spheres at atmospheric pressure and under vacuum. Int. Chem. Eng. 1980, 20, 564-576. (69) Malhotra, K.; Mujumdar, A. S., Single Tube Heat Transfer in Aerated Vibrated Beds. In Drying ’85, Toei, R.; Mujumdar, A. S., Eds. Springer Berlin Heidelberg: Berlin, Heidelberg, 1985; pp 186-194.


44

Page 45 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 1. Summary of data acquisition device and acquired signals

Signal Name

Sensor/Actuator

DAQ Model

Sampling Rate

Pressure

Pressure transducer

MC DAS-08 PCI

400 Hz

Temperature

T-type thermocouple

NI 9214

20 Hz

NI USB-6008

10 Hz 100 Hz

Relative humidity Vaisala with temperature probes

humidity

Digital control

Solenoid valve

NI USB-6009

Voltage and current

TTL serial interface

BK Precision E132B

IT-

10 Hz


45


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 46 of 59

Table 2. Properties of the Biomass Samples at 7% moisture content (dry basis)

Biomass species

Bulk density, True density, Sauter mean Sphericity kg/m3 kg/m3 diameter, mm

Minimum fluidization velocity, m/s

Douglas fir

164

1375

1.449

0.42

0.246

Pine

139

1242

1.469

0.43

0.212

Switchgrass

184

1446

0.755

0.34

0.185


46

Page 47 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 3. Details for the mixture of coarse and fine particle fractions of Douglas fir sawdust

Group Number

Weight of Particle Fractions (g)

Percentage of Fine Particles (wt. %)

0–1.0 mm

3.15–4.0 mm

1

66

134

33

2

100

100

50

3

134

66

67

4

166

34

83


47


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 48 of 59

For Table of Contents Only


48

Page 49 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. Experimental setup of the pulsed fluidized bed system Fig. 1 120x81mm (300 x 300 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Assembly of the heat transfer measurement probe Fig. 2 64x48mm (600 x 600 DPI)


Page 50 of 59

Page 51 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. Particle size distribution of the biomass samples Fig. 3 66x52mm (600 x 600 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. Effect of average gas flow rate on heat transfer coefficient at different gas pulsation frequencies in the pulsed fluidized bed of Douglas-fir sawdust Fig. 4 69x57mm (600 x 600 DPI)


Page 52 of 59

Page 53 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 5. Effect of pulsation frequency on heat transfer coefficient at different gas flow rates in the pulsed fluidized bed of Douglas-fir sawdust Fig. 5 65x50mm (600 x 600 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 6. Effect of pulsation frequency on heat transfer coefficients at different gas flow rates in the pulsed fluidized bed of (a) pine and (b) switchgrass particles. Fig. 6 84x39mm (600 x 600 DPI)


Page 54 of 59

Page 55 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 7. Effect of particle size on heat transfer at different superficial gas velocities in the pulsed fluidized bed with Douglas-fir sawdust, pulsation frequency of 1.25 Hz Fig. 7 68x55mm (600 x 600 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 8. Effect of fine particles on heat transfer at different superficial gas velocities in the pulsed fluidized bed with Douglas-fir sawdust, pulsation frequency of 1.25 Hz Fig. 8 65x50mm (600 x 600 DPI)


Page 56 of 59

Page 57 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 9. (a) Effect of mechanical vibration and gas flow rate on heat transfer at different pulsation frequencies in the pulsed fluidized bed with Douglas-fir sawdust (b) Effect of pulsation duty cycle on heat transfer at different gas flow rates in the pulsed fluidized bed with Douglas-fir sawdust, f=1.25 Hz 70x28mm (600 x 600 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 10. Spectrum analysis of the bed pressure drop fluctuations during heat transfer measurement of Douglas-fir sawdust, f=1.25 Hz, U=1.05Umf with 100% vibration Fig. 10 63x47mm (600 x 600 DPI)


Page 58 of 59

Page 59 of 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 11. Experimental and fitted heat transfer coefficients from two models in pulsed fluidized bed, with Douglas-fir sawdust (a) U=1.1Umf (b) U=1.4Umf Fig. 11 142x238mm (600 x 600 DPI)


Heat Transfer in a Pulsed Fluidized Bed of Biomass Particles

Recommend Documents