particle-scale modeling of heat transfer in a double screw ... · model validation: heat transfer...

Particle-scale modeling of heat transfer in a

double screw pyrolyzer

Fenglei Qi and Mark Mba Wright*

Symposium on Thermal and Catalytic Sciences

for Biofuels and Biobased Products

The Friday Center, University of North Carolina-Chapel Hill

November 1-4, 2016

* Presenting author: [email protected]

USDA REE Energy Summit

2

Fast pyrolysis reactors

Fluidized bed reactors are commonly used

• High heat transfer rate

• Relatively mature design experience

• Simple geometry

• Capability for scale-up

• Ease of operation

• High bio-oil yield

Drawback: Large amount of hot inertial fluidized gas is needed

Bubbling fluidized bed

He et al. (2016)

Double screw reactor Horizontal stirred bed reactorVertical stirred

bed reactorLago et al. (2015) Xi et al. (2015)Brown and Brown (2012)


3

Double-screw auger reactor

SandBiomass

Empirical knowledge: Double screws generate better

mixing between particles

Adding heat carrier particles to enhance biomass

particle heating rates

Flexible mass feeding ratio of biomass to heat carrier

Comparable performance to the fluidized bed

reactors in terms of bio-oil yield at low feeding rate

(fb = 1.0 kg/h) reported by Brown and Brown (2012)

Brown et al. (2012)


4

Fundamental understanding of granular flow

and heat transfer physics in the reactor

Experimental work by Kingston and Heindel (2014a, 2014b)

Our goal: Developing an Discrete Element Method (DEM) to resolve particle-scale

physics to gain better insight of granular flow and heat transfer physics

Cold flow

Limited tracking number of

particles (X-ray)

High cost of time and

equipment


5

Discrete element method (DEM)

DEM considers each particle as an entity in a collection of particles and

resolves each particle motion based on Newton’s equations of motion.

Translational

Rotational

d

d

c nc f gii ij ik i i

j k

mt

vF F F F

, ,

d

d

ii t ij r ij

j

It

ωT T

Soft sphere model


Biomass: Blue, Heat carrier: Red

6

Counter rotating down pumping, ω=40 rpm, P/D=1.25

Particle residence time distribution (RTD) and mixing

degree (M) along the axial direction were investigated.

F, Qi, T.J. Heindel, M. M. Wright, Numerical study of particle mixing in a lab-scale screw mixer using the discrete element

method, Powder Technology, 2016. (under review)

Simulation carried out by LIGGGTHS

Simulated granular flow in the reactor with the

Discrete Element Method (DEM)

ω = 40 rpm


7

Particle-scale heat transfer mechanisms

Particle-particle conduction

Particle-fluid-particle conduction

Particle-particle radiation

Particle-fluid convection (forced or natural)

Particle-fluid radiation

The contribution of each heat transfer pathway depends on the particulate flow

parameters such as void fraction, particle thermal properties and gas interactions.

Lumped capacity model


8

Particle-scale heat transfer model

Particle-particle conduction by Zhou et al. (2004)

Particle-fluid-particle conduction by Cheng et al.(1999)

Particle-particle radiation by Cheng and Yu (2013)


9

Interactions between spherical particle and wall

Triangle meshes are employed to render surfaces of reactors

Soft sphere model is applied to resolve the collision between particles and

triangle meshes


10

Sphere-triangle mesh heat transfer

Single cone boundary model

Effective circular surface of triangle mesh


11

Model validation: Heat transfer in pack beds

Effective thermal conductivity is independent of cell size when

Effective thermal conductivity (ETC) is calculated as

kp = 0.83 W/m·K, kf = 0.028 W/m·Kɛp =0.56, dp= 2 mm

8i px d

(K)


12

Comparison of predicted ETC with experimental

measurements

A good agreement is observed between model predictions and experimental

measurements.


13

Thermal properties

(K)

Operation conditions: Counter rotating pumping down, ω=40 rpm, P/D=1.25, dp = 2 mm


14

Distributions of particle and temperature in axial

directions at different feed rates

High feed rate fb=3.5 kg/h

Low feed rate fb=1.0 kg/h

Average temperature at

x positions

Particle temp

distribution in pitches

Temperature distribution aligns with particle

distribution at high feeding rates

Particle mixing has limited effects on heat transfer

at low feed rates


15


directions with different particle sizesParticle size Dp=1.0 mm

Particle size Dp=2.0 mm


x positions

Particle temp


Similar particle & temperature distributions are

observed with different particle sizes


16


directions with different rotation speeds

Rotation speed ω=20 rpm

Rotation speed ω=60 rpm


x positions

Particle temp


In pitch 7, more uniform temperature is achieved at

smaller rotation speed due to increased residence time


17

Average temperature

In current reactor design and operation conditions, it is predicted only one case

(dp=1.0 mm, fb=1.0 kg/h) achieves steady temperature before the reactor outlet.


18

Biomass particle heating rates

Rotation speed Biomass feed rate Particle size

The range of biomass particle heating rate in the reactor is between 20 and 50 K/s

in current operation conditions.

Increasing rotation speed improves biomass heating rate, while increasing biomass

feed rate decreases the heating rate.

Biomass heating rate is not affected by biomass particle size in the range of 1-2 mm


19

Summary

A thermal DEM model considering particle-scale heat transfer was developed and

validated in this research.

Biomass heating rate was predicted within the range of 20-50 K/s in the double screw

reactor at current operation conditions.

The influences of the operating conditions on the temperature profiles in the reactors

were investigated showing that higher rotation speed and longer residence time are

favorable for the heat transfer at constant feed rate while decreasing feed rate could

enhance the heating process of biomass particle.

Biomass heating rate is not affected by biomass particle size in the range of 1-2 mm

when feed rate is greater than 2 kg/h.


Funding: National Science Foundation

EPSCoR program

Acknowledgement

Dr. Robert Brown, Dr. Qi Dang, Tannon

Daugaard and all other colleagues in the

BEI and CoMFRE at Iowa State

University.

20


21

References

Brown, J. N., & Brown, R. C. (2012). Process optimization of an auger pyrolyzer with heat carrier using response surface methodology. Bioresource

technology, 103(1), 405-414.

Cheng, G. J., Yu, A. B., & Zulli, P. (1999). Evaluation of effective thermal conductivity from the structure of a packed bed. Chemical Engineering Science,

54(19), 4199-4209.

Cheng, G. J., & Yu, A. B. (2013). Particle scale evaluation of the effective thermal conductivity from the structure of a packed bed: Radiation heat transfer.

Industrial & engineering chemistry research, 52(34), 12202-12211.

He, Y., Yan, S., Wang, T., Jiang, B., & Huang, Y. (2016). Hydrodynamic characteristics of gas-irregular particle two-phase flow in a bubbling fluidized bed:

An experimental and numerical study. Powder Technology, 287, 264-276.

Kingston, T. A., & Heindel, T. J. (2014). Granular mixing optimization and the influence of operating conditions in a double screw mixer. Powder

Technology, 266, 144-155.

Kingston, T. A., Morgan, T. B., Geick, T. A., Robinson, T. R., & Heindel, T. J. (2014). A cone-beam compensated back-projection algorithm for X-ray

particle tracking velocimetry. Flow Measurement and Instrumentation, 39, 64-75.

Lago, V., Greenhalf, C., Briens, C., & Berruti, F. (2015). Mixing and operability characteristics of mechanically fluidized reactors for the pyrolysis of

biomass. Powder Technology, 274, 205-212.

Xi, Y., Chen, Q., & You, C. (2015). Flow characteristics of biomass particles in a horizontal stirred bed reactor: Part I. Experimental measurements of

residence time distribution. Powder Technology, 269, 577-584.

Zhou, J. H., Yu, A. B., & Horio, M. (2008). Finite element modeling of the transient heat conduction between colliding particles. Chemical Engineering

Journal, 139(3), 510-516.


22

Thank you

Question?


23

Backup slides


24

Question: what physics and scales should be considered in the mathematical model

development for fast pyrolysis modeling?

Motivation: biomass fast pyrolysis modeling

Scale

Co

mp

lexity

Kinetic modeling

Single particle modeling

Eulerian method

Temperature,

Reaction time

Intra-particle heat

and mass transfer

Particle-particle

and particle-gas

interactions

Particle flow modeling

DEM method


Top view

x

y

z

Side view

Operation conditions: Counter rotating pumping down, ω=40 rpm, P/D=1.25

Experimental photo from Kingston and Heindel (2014)

25

Model validation

Biomass: Blue, Heat carrier: Red


26

Discrete element method (DEM)

Soft sphere model (spring-dashpot model)

, ,

c c c

ij n ij t ij F F F

• Overlap in normal and tangential directions are allowed. Force-displacement

models and damping models are adopted

, ,ˆ ˆ ˆ( )c

n ij n n n rel Ck C F n v n n

, , , ,contact t ij contact n ijF F

, ,ˆ ˆ( )c

t ij t t t rel Ck C F δ v n n

Hertz-Mindlin model is employed in this research.


27

Particle-scale heat transfer model

Double cones boundary model (Cheng et al., 1999)

Boundary parameters are related

to local void fraction in the bed

Voronoi tessellation of a 2D packed bed

Cheng et al. (1999)

ij


28

View factor

Definition of view factor

Monte Carlo Method (Mont3D)

/ 2h C R


29

Correlations of view factor

Very loose random packing (ɛs=0.56)


30

Distributions of conduction pathways

Heat conduction by particle-fluid-particle is the dominant contribution to the

total conductive heat transfer when kp/kf<100.


31

Time scales

Collision time

In order to resolve collision, computational time step is usually < 1/20 of

Biomass residence time

Computational cost

~ 107-108

• Smaller Young’s modulus is used to reduce computational cost

• MPI technique is employed to simulate large particle systems


32


directions at different feeding rates

High feeding rate (fb=3.5 kg/h) Low feeding rate (fb=1.0 kg/h)

At high feeding rate, particle temperature distribution in transverse sections

aligns with particle distribution

At low feeding rate, particle mixing has limited impact on heat transfer

particle-scale modeling of heat transfer in a double screw ... · model validation: heat transfer...

Documents