3.4 - Jarvinen - Status of BL Combustion Modeling

HELSINKI UNIVERSITY OF TECHNOLOGY

Current Status of Black Liquor Droplet Combustion Modelling Modelling and CFD SubModel Development Mika Järvinen Helsinki University of Technology

Christi Christian an Muelle Muellerr and and Mikko Mikko Hupa Hupa Åbo Akademi University

Contents 1. 2. 3. 4. 5. 6. 7. 8.

Introduction Black liliquor co combustion m moodeling si since 1980’s Deta Detaililed ed drop drople lett mode model,l, Järv Järvin inen en et al. al. (200 (2002) 2) Important observations from latest dr droplet mo models and their implications Descr escriipt ptio ionn of the the new new sim simplifi lifieed comp compre rehe hens nsiv ivee dr drople oplett model Towards a sub-model for CFD applications First ap application of th the su sub-model Summary

Introduction • Many single droplet combustion models have been developed for black liquor since 1980’s • Some models have been used to study combustion behavior under constant reactor conditions, allowing a very detailed description of the processes • Models that are used in connection to CFD software should be relatively simple but effective, development work very challenging • This presentation will concentrate on the current status of black liquor combustion modeling mainly in Finland, but also gives a short history review of single droplet model development

Black liquor combustion modeling since 1980’s • Modeling of black liquor combustion was initiated by Merriam (1980) introducing a 1D equilibrium furnace model  – Droplet model was generally based on coal work, no BL specific data available  – Modeling approved to be very useful tool for RB analysis • 1989 Walsh and Grace presented a first FLUENT based recovery boiler CFD model  – Droplet combustion and spray properties were based on experiments  – Isothermal droplet, sequential stages  – Included drying, devolatilization, char oxidation and smelt oxidation

• Frederick (1990) presented a single droplet model based on large set of experimental data and conservation laws for mass and energy  – Experimental determination of droplet swelling during drying and devolatilization and size reduction during char burnout  – Non-isothermal droplet during drying and devolatilization, isothermal assumption gave too high drying rates  – Char conversion by external mass transfer of O2, CO2 and H2O, intra-particle mass transfer and reaction included later by Frederick et al. (1993)  – This model has been widely used later by other researches • Wåg et al. (1995) presented a detailed isothermal char conversion model  – Char could react with O2, CO2 and H2O, and also with sulfates and carbonates  – Model considered boundary layer gas mass transfer with pore diffusion and heterogeneous reactions in pores  – Good correlation with experimental data

• Models of Verrill and Wessel (1996) and Saastamoinen (1996) numerically solved intra-particle temperature distribution during drying and devolatilization  – Particles found to be highly non-isothermal  – Simultaneous drying and devolatilization  – Role of intra-particle thermal radiation found important, Verrill and Wessel • Järvinen et al. extended (2002) the work of Verrill and Wessel and Saastamoinen by developing a general heat and mass transfer model for porous particles  – Solving numerically the boundary layer and intra-particle heat- and mass transfer and reactions during the gas phase  – Overlapping of all combustion stages, important role of char conversion during drying and devolatilization

Detailed droplet model • Published by Järvinen et al. 2002 (Doctoral thesis) • Tekes/CODE programme 1999 2002 • Academy of Finland 2003-2005 • Tekes ChemCom 2005-2007

Detailed droplet model - Internal profiles 300 Dry

240

Pyro

1200

   )   s

   3

  m180

900

120

600

   /    l   o   m    (    i

      R

60

300

0

   )    C    °    (   e   r   u    t   a   r   e   p   m   e    T

Temperature and reaction profiles during pyrolysis

50   s

   3

40

  m    /    l   o 30   m  ,    i

0.2 0.4 0.6 0.8 Radial coordinate (r/r s)

20

0 0

1250

150

1000 H2O + C CO2 + C O2 + C CO + H2O

   )   s

   3

  m100    /    l   o   m    ( 50    i

750 500

      R

0

250

-50

0 0

0.2 0.4 0.6 0.8 Radial coordinate (r/r s)

1

r 

10

1

200

H2O + C O2 + C CO2 + C

      R

0 0

Ri

60

1500

   )    C    °    (   e   r   u    t   a   r   e   p   m   e    T

Temperature and reaction profiles during char combustion

0.2 0.4 0.6 0.8 Radial coordinate, r/r s

1

50 40 O2 + Na2S M2CO3 + 2 C M2SO4 + 2 C

   )   s    3

  m30    /    l   o   m    (    i

20

      R

10 0 0

0.2 0.4 0.6 0.8 Radial coordinate (r/r s)

1

Important observations and their implications • Overlapping drying and devolatilization stages in thin cores with characteristic temperatures: Tb ~ 150°C, Tp ~ 250-300°C → Shrinking core approach applicable • Intraparticle thermal radiation is important aR ~ 850 1/m → should be considered • During drying and devolatilization, temperature profile approaches quasi-steady state profile → analytical solution for T(r) possible, sensible heat term dT/dt negligible (however, used to obtain a stable solution)

• Char conversion occurs simultaneously with drying and devolatilization at the particle surface → Overlapping effects to be considered • During ”pure” char combustion stage particle is almost isothermal ,t) → Possible to model chemical reactions at T(t), not T(r  • No single dominating char reaction → ”All” reactions to be included

The simplified comprehensive droplet model Tg

Ts(t)

Tp= const Tb= const H2O(l)

C(s) + DS N(s) MCl(s) DS M2S(s) M2SO4(s) M2CO3(s)

- only one temperature Ts(t) + 8 tracked species mi(t) to be solved - const. Tb, Tp - Na + K => M

Reactions in the simplified model H2O(l) → H2O DS (dry solids)

→

volatiles, C(s), N(s), M2S(s), M2SO4(s), M2CO3(s), MCl

Pyrolysis and drying modeled as a heat transfer limited processes, taking place at a single temperature determined by the detailed model, resulting a shrinking core approximation Volatile combustion around droplet / in far field ?

Parallel oxidation reactions C(s) + 0.5O2 → CO

Smith 1982

M2S(s) + 2 O2 → M2SO4(s)

-”-

Heterogeneous oxidation reactions are calculated from analytical expressions, including mass transfer and continuous reaction in porous isothermal char layer.

Char gasification C(s) + H2O → CO + H2

Li et al. 1991

C(s) + CO2 → 2 CO

Li et al. 1990

Heterogeneous reactions of char are calculated from analytical expressions, including mass transfer and continuous reaction in porous isothermal char layer.

Inorganic reactions M2CO3(s) + 2 C(s) → 2 M + 3 CO

Li et al. 1991

M2SO4(s) + 2 C(s) → Ma2S(s) + 2 CO2

Wåg et al., 1995

Inorganic reactions of char are calculated from kinetic expressions, no mass transfer effects, isothermal char layer 

Towards a sub-model for CFD • Developed C-source code of the stand-alone droplet model delivered to ÅA by HUT • Conversion of source code to Fluent UDF and implementation into Fluent at ÅA • Fluent UDF in use at ÅA and HUT, common test case • Sensitivity analysis, comparison and application of all single droplet/particle models developed at HUT and ÅA at well defined conditions  – Detailed and comprehensive black liquor droplet model  – Current droplet/particle models in the ÅA Furnace Model

First application of the sub model • Comprehensive single droplet model programmed in form of  one DPM law • Iterative determination of droplet conditions per timestep x on the droplet trajectory

x

x x

x

x

x

Computationally demanding

x x

• Common procedure: 10 fluid flow iterations per DPM iteration

First application of the sub model Testcase – 625 tDS/day boiler 

28.5 m A3, 11m (22%)

Domsjö

- 2450 t/day air  - 625 tDS/day black liquor  - rotational firing mode of tertiary air 

BL, 7 m A2, 3m (45%) A1, 1m (33%) 6.2 m

6.7 m

First application of the sub model Droplet Diameter [m]

First application of the sub model Droplet Temperature [K]

First application of the sub model Droplet Drying [kg H2O/s]

0

First application of the sub model Droplet Diameter [m] – 500 Droplet Trajectories

First application of the sub model Computational Demand, CPU-time [s] • 10 droplets from one liquor gun, 50 stochastic tries 500 droplet trajectories • ÅA Furnace Model: 9 s • Comprehensive SDM: 109 s • Increase in computational time per DPM iteration: ~ factor 12

Summary • A new comprehensive single black liquor droplet combustion sub-model has been implemented into FLUENT software and is currently in use at HUT and ÅA • Simplified model is based on the experimentally validated detailed model, using the most essential physical and chemical mechanisms observed by simulations and experiments • Close co-operation between Helsinki University of Technology and Åbo Akademi University has been essential for effectively combining the knowledge on detailed model development, combustion experiments and CFD model implementation • The project has been definitely a “Win- Win deal” for both parties