direct quadrature method of moments (dqmom)

DIRECT QUADRATURE METHOD OF MOMENTS (DQMOM) APPROACH FOR VERTICAL GAS-LIQUID BUBBLY FLOWS OF LARGE PIPE 1,2

3

3

G.H. Yeoh , L. Deju , S.C.P. Cheung , and J.Y. Tu

3

1

Australian Nuclear Science Technology Organisation (ANSTO) Locked Bag 2001, Kirrawee DC, NSW 2232, Australia [email protected] 2 School of Mechanical and Manufacturing Engineering University of NSW, NSW 2052, Australia [email protected] 3 School of Aerospace, Mechanical and Manufacturing Engineering RMIT University, Victoria 3083, Australia [email protected], [email protected]

ABSTRACT Gas-liquid bubbly flows with wide range of bubble sizes are commonly encountered in many industrial gas-liquid flow systems. The performance of direct quadrature method of moments (DQMOM) has been assessed against the homogeneous MUlti-SIze-Group (MUSIG) model and Average Bubble Number Density (ABND) approach in tracking the changes of bubble size distribution and gas volume fraction under complex flow conditions. Numerical studies have been performed to validate predictions from the different population balance approaches against experimental measurements of vertical bubbly flows in a large diameter pipe. In general, predictions of DQMOM were in good agreement with experimental data. The encouraging results demonstrated the capability of DQMOM in capturing the dynamical changes of bubbles size due to bubble interactions and the transition from wall peak to core peak gas volume fraction profiles caused by the presence of small and large bubbles. Predictions of the DQMOM appeared to offer substantial reduction of computational times in reaching a converged solution when compared to MUSIG for the computation of vertical bubbly flows in large diameter pipe. INTRODUCTION Two-phase gas-liquid flows exist in numerous industrial applications such as in areas of chemical, civil, nuclear, mineral, energy, food, pharmaceutical and metallurgy. Because of the rather complex two-phase flow structures that are found within these technological systems and since such flow structures can evolve dynamically and transit to different flow regimes, the phenomenological understanding of bubble size and its dispersion behaviour is of paramount importance. Particularly for the bubbly flow regime, the spectrum of bubble size usually shows a single modal or single-peaked distribution due to moderate bubble coalescence and break-up. With increasing gas volume fraction and bubble number density, the bubble size distribution gradually becomes bimodal or double-peaked, which is caused by significant bubble interactions. Such changes of the bubble size spectrum are further complicated by the presence of

G.H. Yeoh, S.C.P. Cheung, J.Y. Tu lateral forces acting on the bubbles (Tomiyama, 2004, Bothe et al., 2006). The lateral lift force which acts in the perpendicular direction to the main flow in a vertical pipe, affects dramatically the bubble size distribution. Experiments have clearly shown that small bubbles, driven by positive lift forces, generally flow along the pipe wall are separated from large bubbles which migrate instead towards the pipe centre turning into cap or even larger Taylor bubbles via additional coalescence. Majority of experimental data which form the basis of verification and validation of two-phase physical models for upward flow in vertical pipe has thus far concentrated on so-called small diameter pipe – a hydraulic diameter of around 50 mm or below (Hibiki et al., 2001). Nevertheless, flow conditions deployed for validation generally exhibited a rather narrow bubble size range due to the considerably moderate bubble interaction rates occurring within the two-phase flow system (Cheung et al., 2007a,b). Theoretically speaking, the muted bubble interaction behaviour that had been experienced in experiments reduced the non-linearity of the problem thereby producing a less challenging prospect in fully assessing the model’s capability. The exercise allows the model to be validated to some degree but may conceal the possible potential of predicting complex gas-liquid flows where rigorous bubble interactions would be significant. In this paper, the application of DQMOM to predict the two-phase characteristics within vertical bubbly flows of large diameter pipe is considered. The main advantage of this population balance approach is essentially the computational economy of which the two-phase flow problem can be condensed by only tracking the evolution of a small number of moments (normally 4-6). This becomes particularly important in modelling complex bubbly flows when the bubble dynamics is strongly coupled with already timeconsuming calculations of turbulence multiphase flows especially in large diameter pipe bubbly flows. DQMOM can offer a powerful approach in describing bubbly flows undergoing coalescence and break-up processes in the context of computational fluid dynamics simulations. Specific attention is directed towards evaluating the performance of the two models in capturing the transition from wall peak to core peak radial void fraction distribution, corresponding to the prevalence of lift forces acting on the smalland large-sized bubbles. Model predictions via DQMOM are verified against homogeneous MUSIG model (Lo, 1996), and ABND approach (Duan et al., 2011) and validated against gas-liquid flow experiments in vertical pipes of large size measured in the Forschungszentrum Dresden-Rossendorf FZD facility (Prasser et al., 2007). MATHEMATICAL MODELS Two-Fluid Model For isothermal vertical bubbly flows, the equations for the ensemble-averaged of continuity and momentum governing each phase are solved simultaneously. By denoting the liquid as the continuous phase (αl) and the gas (i.e. bubbles) as disperse phase (αg), these equations can be written as: Continuity equation of liquid phase ∂ l l ρ α ) + ∇ ⋅ ( ρ l α l ul ) = 0 ( ∂t

(1)

Continuity equation of gas phase ∂ g g ρ α f i ) + ∇ ⋅ ( ρ g α g u g f i ) = Si ( ∂t

(2)

G.H. Yeoh, S.C.P. Cheung, J.Y. Tu Momentum equation of liquid phase

)

(

T ∂ l l l ρ α u ) + ∇ ⋅ ( ρ l α l ul ul ) = −α l ∇P + α l ρ l g + ∇ ⋅ α l µel ∇u l + ( ∇ul )  + F lg (    ∂t

(3)

Momentum equation of gas phase

(

)

T ∂ g g g ρ α u ) + ∇ ⋅ ( ρ g α g u g u g ) = −α g ∇P + α g ρ g g + ∇ ⋅ α g µeg ∇u g + ( ∇u g )  + F gl (   ∂t

(4)

In equation (2), Si represents the additional source terms due to coalescence and breakup effects. The total interfacial force Flg appearing in equation (3) is formulated according to consideration of different sub-forces affecting the interface between each phase. For the liquid phase, the total interfacial force as demonstrated by Frank et al. (2004) is given by the drag, lift, wall lubrication and turbulent dispersion, viz., lg lg lg F lg = Fdrag + Fliftlg + Flubrication + Fdispersion

(5)

Note that the total interfacial force in equation (4) is given by Fgl = − Flg. The interphase momentum transfer between gas and liquid due to drag force is given by 1 lg Fdrag = CD aif ρ l u g − u l ( u g − ul ) 8

(6)

Lift force in terms of the slip velocity and the curl of the liquid phase velocity can be described as Fliftlg = α g ρ l CL ( u g − u l ) × ( ∇ × ul )

(7)

Wall lubrication force, which is in the normal direction away from the heated wall and decays with distance from the wall, is expressed by lg Flubrication −

(

)

α g ρ l ( u g − u l ) − ( u g − u l ) ⋅ n w n w     Ds

2

Ds   Cw1 + Cw 2  nw yw  

(8)

Turbulence induced dispersion based on the consistency of Favre-averaging developed by Burns et al. [13] is applied: lg dispersion

F

µtg  ∇α g ∇α l  = −CTD CD g − l   ρ Scb  α g α 

(9)

The drag coefficient CD in equation (6) is determined according to Ishii and Zuber [14] correlation for several distinct Reynolds number regions of different shape bubbles. The constant CL has been correlated according to Tomiyama et al. (2004) – a relationship being expressed as a function of the Eotvos number (Eo) allowing positive and negative lift coefficients depending on the bubble size and also accounts for the effects of bubble deformation and asymmetric wake of the bubble. The wall lubrication constants Cw1 and Cw2 are taken to have values of –0.0064 and 0.016 according to by Krepper et al. [6]. The coefficient CTD is normally set to a value of unity and Scb is the turbulent bubble Schmidt number with an adopted value of 0.9. The Shear Stress Transport (SST) model developed by Menter (1994), which applies the two-equation k-ω model near the wall and the two-equation k-ε model in the bulk flow, has been shown to provide more realistic prediction of gas volume fraction or void fraction close to the wall of the flow domain. Effective viscosity ( µel ) for the

G.H. Yeoh, S.C.P. Cheung, J.Y. Tu l continuous phase comprises of the laminar ( µlam ) , shear-induced turbulence ( µtsl ) and

Sato’s bubble-induced turbulent viscosities ( µtdl ) (Sato et al., 1981): l µel = µlam + µtsl + µtdl

(10)

where µtsl = Cµ ρ l ( k l ) ε l and µtdl = Cµ p ρ lα g Ds u g − u l . For the disperse gas phase, 2

the dispersed phase zero equation model is utilized and the turbulent viscosity of gas phase can be obtained as

µtg =

ρ g µtl ρl σ g

(11)

where σ g is the turbulent Prandtl number of the gas phase which has a value of unity.

DQMOM for Bubbly Flow Based on Marchisio and Fox (2005), the quadrature abscissas and weights can be formulated as transport equations. In order to retain consistency with the variables employed in the two-fluid model, the weights and abscissas can be related to the size fraction of the dispersed phase (fi) and an additional variable defined as ψi = fi / Mi where M represents the mass. Considering a homogeneous system in which the bubbles are assumed to travel with a common gas velocity, the transport equations become ∂ g g ( ρ α fi ) + ∇ ⋅ ( ρ g u g α g fi ) = bi ∂t

(12)

∂ g g ( ρ α ψ i ) + ∇ ⋅ ( ρ g u g α gψ i ) = ai ∂t

(13)

where the terms ai and bi are related to the birth and death rate of population which forms 2n linear equations where the unknowns can be evaluated via matrix inversion according to AΦ = B

(14)

The 2n × 2n coefficient matrix A = [ A1 A2 ] in the above linear equation takes the form:

   A1 =         A2 =    

1 0

L L

− M 12 M 2(1 − n) M 12 n −1

L O L

0 1

L L

2M 1 M (2n − 1) M 12 n − 2

L O L

    − M n2  M  2 n −1  2(1 − n) xn 

(15)

    2M n  M  (2n − 1) M n2 n − 2 

(16)

1 0

0 1

where the 2n vector of unknowns Φ comprises essentially the terms ai and bi in equations (12) and (13):

G.H. Yeoh, S.C.P. Cheung, J.Y. Tu

a  T Φ = [ a1 K an b1 K bn ] =   b 

(17)

In equation (14), the source or sink term is defined by B = [ S0 K S 2 n −1 ]

T

(18)

The moment transform of the coalescence and break-up of the term S k can be expressed as S k = BkC − DkC + BkB − DkB

(19)

where the birth ( Bk ) and death ( Dk ) rates can be written in terms of mass:

BkC =

k 1 Ni N j ( M i + M j ) a ( M i , M j ) ∑∑ 2 i j

(20)

DkC = ∑∑ M ik a ( M i , M j ) N i N j

(21)

BkB = ∑∑ M ik r ( M j , M i ) N j

(22)

DkB = ∑∑ M ik r ( M i , M j ) N i

(23)

i

i

i

j

j

j

where superscripts C and B denote coalescence and break-up respectively. The coalescence and break-up kernels a ( M i , M j ) and r ( M i , M j ) are described according to the turbulent collision taken from the proposal of Prince and Blanch (1990) are bubble binary break-up under isotropic turbulence situation suggested by Luo and Svendsen (1996). They are (Duan et al., 2011):

a (Mi , M j ) =

 t  0.5 2  d i + d j  ( uti2 + utj2 ) exp  − ij   τ  4  ij 

π

1/3

 ε  r ( M i , M j ) = 0.923 (1 − α g )  2     dj 

1

∫

ξ min

(1 + ξ ) ξ 11/3

2

(24)

  12c f σ   dξ × exp −  2 ρ l ( ε l )2/3 d 5/3ξ 11/3    (25)

EXPERIMENTAL DETAILS In this experimental facility, a large size vertical cylindrical pipe with height 9000 mm and inner diameter of 195.3 mm inner diameter was adopted as the test section. Water was circulated from the bottom to the top with a constant temperature of 30ºC, maintained by a heat exchanged installed in the water reservoir. Different gas injection device was employed in this TOPFLOW experiment. A variable gas injection system was constructed by equipping with gas injection units at 18 different axial positions from Z/D = 1.1-9.9. Three levels of air chambers were installed at each injection unit. The upper and the lower chambers have 72 annular distributed orifices of 1 mm diameter for small bubble injection while the central chamber has 32 annularly distributed orifices of 4 mm diameter for large bubble injection. A fixed wire-mesh

G.H. Yeoh, S.C.P. Cheung, J.Y. Tu sensor was implemented at the top of the pipe where instantaneous information of gas volume fraction as well as bubble size distribution was measured.

NUMERICAL DETAILS AND RESULTS Numerical calculations were achieved through the use of the generic computational fluid dynamics code ANSYS-CFX11. For DQMOM, transport equations governing weights and abscissas were solved to predict the bubble size distribution of which the evaluation of the source terms ai in equation (13) and bi in equation (12) has been achieved through the use of a user subroutine incorporated within the CFD computer code. Radial symmetry was assumed in both experimental conditions thereby allowing the computational geometry to be simplified through consideration of a 60o radial sector of the pipe with symmetry boundary conditions being imposed at both vertical sides of the computational domain. Two sets of experiment data under two different flow conditions – hereafter denoted as case T107 and T118 – were selected from the TOPFLOW experiment.

[ jl

Z / D =0

[ jl

Z / D =0

[α g [ DS

Z / D =0

Case T107

Case T118

]

(m/s)

1.017

1.017

]

(m/s)

0.140

0.2194

(%)

[12.1]

[17.72]

(mm)

[20.18]

[23.28]

]

z / D = 0.0

] Tab.1: Inlet boundary conditions of test cases for numerical simulations In order to represent the wall injection method in TOPFLOW, 12 equally spaced point sources of the gas phase were placed at the circumference of the 60o radial sector. The gas injection rate at each point source was assumed to be identical. Details of the boundary conditions are summarized in Tab. 1. Based on grid sensitivity test performed for the TOPFLOW experimental cases, grid independent solutions have revealed that a computational mesh which consisted of 48,000 elements did not appreciably change even though finer computational meshes were tested. Comparing with the finer mesh, the predicted cross-sectional averaged volume fractions were found only within differences of 2%. For all flow conditions, reliable convergence criterion based on the RMS (root mean square) residual of 1.0× 10-4 was adopted for the termination of numerical calculations.

Sensitivity Study of Moments for DQMOM Comparison of bubble size distribution and average bubble Sauter mean diameter for case T107 at Z/D = 39.9 between measurement and predictions made by DQMOM as well as MUSIG with 20 bubble classes is depicted Fig. 1 and Tab. 2 respectively. Moment independence investigation for DQMOM was assessed for 4, 6 and 8 moments. It could be seen from Fig. 1 that DQMOM was, in general, able to predict a dominant peak closed to the peak experimental bubble diameter. Nevertheless, the peak predicted from MUSIG was found to be skewed towards the small size bubble class. As shown in Tab. 2, the maximum difference of the predicted average bubble Sauter mean diameter by DQMOM with 6 and 8 moments was found to differ less than 1%. It can therefore be concluded that DQMOM with 6 numbers of moments were sufficient in obtaining moment independent solutions.

G.H. Yeoh, S.C.P. Cheung, J.Y. Tu 3

3 DQMOM (6MOMENTS)

Bubble Size Distribution, [%/mm]


DQMOM (4MOMENTS) 2.5

Experiment MUSIG

2

(a) T107 Case Case T107 Z/D = 39.9 Z/D=39.9

1.5

1

0.5

0

0

20

40

2.5

Experiment MUSIG

2

(b) T107 Case Case T107 Z/D = 39.9 Z/D=39.9

1.5

1

0.5

0

60

0

20

Bubble Diameter, [mm]

40

60



3 DQMOM (8MOMENTS) Experiment

2.5

MUSIG 2

(c) T107 Case Case T107 Z/D = 39.9 Z/D=39.9

1.5

1

0.5

0

0

20

40

60


Fig. 1: Local bubble size distribution profiles at Z/D = 39.9 for case T107 Location

Z/D=39.9

Sauter Mean Diameter Experiment

(mm)

[7.49]

DQMOM (4Moments)

(mm)

[11.23]

DQMOM (6 Moments)

(mm)

[10.15]

DQMOM (8 Moments)

(mm)

[10.21]

Tab.2: Average bubble Sauter mean diameter at Z/D = 39.9 for case T107

Assessment of DQMOM with MUSIG and ABND Fig. 3 shows the comparison of predicted cross-sectional bubble size distribution from DQMOM with 6 number of moments along with those of MUSIG of 20 bubble classes against TOPFLOW data at different axial location of Z/D = 1.7, 22.6 and 39.9. It could be seen that the prediction from DQMOM agreed rather well with the measurement. Interestingly for case T118 at Z/D = 1.7 especially close to the vicinity of injection unit, both DQMOM ad MUSIG were able to capture the bimodal bubble size distributions; DQMOM prediction was nonetheless gave better agreement with measurement. Downstream, the bubble size distribution recovered to a single modal profile of which both DQMOM and MUSIG were also found to be in satisfactory agreement with the measurement. Despite the reduced number of scalars in DQMOM, the encouraging results clearly demonstrated that the evolution of bubble sizes was well captured by

G.H. Yeoh, S.C.P. Cheung, J.Y. Tu DQMOM in comparison to MUSIG where extra equations were required to be solved for each bubble class. 1

1.2 DQMOM (6MOMENTS)



DQMOM (6MOMENTS) Experiment

0.8

MUSIG

0.6

(a)

Case T107 Case T107 Z/D=1.7 Z/D = 1.7 0.4

0.2

0

0

20

40

60

1

MUSIG 0.8

(d) T118 Case Case T118 Z/D = 1.7 Z/D=1.7

0.6

0.4

0.2

0

80

Experiment

0

20


3

Experiment MUSIG

2

(b)

Case T107 Case T107 Z/D=22.6 Z/D = 22.6

1.5

1

0.5

0

20

40

60

2.5

MUSIG 2

(d)

Case T118 Case T118 Z/D=22.6 Z/D = 22.6

1.5

1

0.5

0

80

Experiment

0

20


40

60

80


3

3

DQMOM (6MOMENTS) Experiment MUSIG

2.5

DQMOM (6MOMENTS)



80

DQMOM (6MOMENTS)



DQMOM (6MOMENTS)

2

(c)

Case T107 Case T107 Z/D=39.9 Z/D = 39.9

1.5

1

0.5

0

60

3

2.5

0

40


0

20

40


60

80

2.5

Experiment MUSIG

2

(f) Case T118 Case T118 Z/D Z/D=39.9 = 39.9

1.5

1

0.5

0

0

20

40

60

80


Fig. 2: Local bubble size distribution profiles at Z/D = 1.7, 22.6 and 39.9 for cases T107 and T118 Tab. 3 summarizes the computational time taken for MUSIG, ABND and DQMOM with 4, 6 and 8 number of moments. MUSIG required 82 hours of computational time to obtain a converged solution when compared to other two population balance approaches. ABND which solved only a single transport equation for the average bubble number density required the least computational time to reach convergence. Nevertheless, the computational times fo DQMOM with 4 and 6 moments were found to be around 50%-80% lower than MUSIG. Even for DQMOM with 8 moments, the computational time was still significantly lower comparing to MUSIG.

G.H. Yeoh, S.C.P. Cheung, J.Y. Tu Fig. 3 shows the void fraction distribution at the position Z/D = 1.7 and 39.9. The transition process from wall peak to core peak of the gas volume fraction profile was successfully captured by the different population balance approaches. In TOPFLOW, the bubbles were injected from the gas injection orifices located on the circumference of the pipe of which highly concentrated bubbles were formed within the wall proximity that resulted in a wall peak void fraction distribution close to inlet of the pipe. In general, predictions from the DQMOM for cases T107 and T118 were in reasonable agreement with the measurement. Case T118

Computational Time

MUSIG

82 hours

ABND

12.96 hours

DQMOM (4 Moments)

14 hours

DQMOM (6 Moments)

42 hours

DQMOM (8 Moments)

68 hours

Tab.3: Computational times of different population balance approaches for case T118 40

35


35


30

MUSIG

30

MUSIG

ABND

Void Fraction, [%]

Void Fraction, [%]

40

25 20 15

Case T107 (a) Z/D = 1.7 Case T107

10

ABND 25 20 15

Case T118 (c) Z/D = 1.7 Case T118

10

Z/D=1.7

Z/D=1.7

5 0

5

0

0.2

0.4

0.6

0.8

0

1

0

0.2

Radial position, [-]

35

0.6

0.8

1

0.8

1

50 45


30


40

MUSIG 25

MUSIG

ABND

Void Fraction, [%]

Void Fraction, [%]

0.4


20

15

10

ABND

35 30 25 20 15

(b)

Case Case T107 T107 Z/D=39.9 Z/D = 39.9

5

(d) T118 Case Case T118 Z/D = 39.9 Z/D=39.9

10 5

0

0

0.2

0.4

0.6


0.8

1

0

0

0.2

0.4

0.6


Fig. 3: Local void fraction profiles at Z/D = 1.7 and 39.9 for cases T107 and T118

CONCLUSION In this study, DQMOM has been assessed and recognized as an effective approach to represent the evolution of the bubble size distribution due to coalescence and breakage. The predicted results for the local radial distributions of the bubble size and void

G.H. Yeoh, S.C.P. Cheung, J.Y. Tu fraction were compared against MUSIG and ABND as well as the TOPFLOW measurement data. Sensitivity study was carried out for DQMOM model with 4, 6 and 8 moments. Using 6 or 8 moments generated similar results, which are different from those found with 4 moments. Considering the increase in computational time with an increasing number of moments, 6 moments appeared to be a suitable choice. Comparing with MUSIG, DQMOM seemed to produce marginally better results in terms of bubble size distribution for some flow conditions. The transition of wall peak to core peak was successfully captured by DQMOM as exemplified by the local radial gas volume fraction profiles.

REFERENCES Burns, A.D., Hamill, I., Frank, T., & Shi, J., The Favre averaged drag model for turbulent dispersion in Eulerian multiphase flow, Proc. Fifth Int. Conf. Multiphase Flow, Yokohama, Japan (2004). Cheung, S.C.P., Yeoh, G.H. & Tu, J.Y., On the modeling of population balance in isothermal vertical bubbly flows – average bubble number density approach, Chem. Eng. Proc., (2007a) 46 742-756. Cheung, S.C.P., Yeoh, G.H. & Tu, J.Y., On the numerical study of isothermal bubbly flow using two population balance approaches, Chem. Eng. Sci., (2007b) 31 1641072. Duan, X.Y., Cheung, S.C.P., Yeoh, G.H., Tu, J.Y., Krepper, E., Lucas, D., Gas-liquid flows in medium and large vertical pipes, Chem. Eng. Sci., (2011) 66 872-883. Frank, T., Shi, J. & Burns, A.D., Validation of Eulerian multiphase flow models for nuclear safety application, Proc. Third Symp. Two-Phase Modeling and Experimentation, Pisa, Italy (2004). Hibiki, T., Ishii, M. & Xiao, Z., Axial interfacial area transport of vertical bubbly flows, Int. J. of Heat and Mass Transfer, (2001) 44 1869-1888. Ishii, M. & Zuber, N., Drag coefficient and relative velocity in bubbly, droplet or particulate flows, AIChE J., (1979)25: 843-855. Krepper, E., Lucas, D. & Prasser, H., On the modeling of bubbly flow in vertical pipes, Nuc. Eng. Des., (2005) 235 597-611. Prince, M. J. & Blanch, H. W., Bubble coalescence and break-up in air-sparged bubble column, AIChE J., (1990) 36 1485-1499. Lo, S. M., Application of population balance to CFD modelling of bubbly flow via the MUSIG model, AEA Technology, AEAT-1096 (1996). Luo, H. and Svendsen, H., Theoretical model for drop and bubble break-up in turbulent dispersions. AIChE J., (1996) 42 1225-1233. Marchisio, D.L. & Fox, R.O., Solution of population balance equations using the direct quadrature method of moments, J. Aerosol Sci. (2005) 36 43-73. Menter, F.R., Zonal two equation k-ω turbulence models for aerodynamics flows, AIAA paper 93-2906 (1993). Prasser, H.M., Beyer, M., Carl, H., Gregor, S., Lucas, D., Pietruske, H., Schutz, P., Weiss, F.P., Evolution of the structure of a gas-liquid two-phase flow in a large vertical pipe, Nuc. Eng. & Des., (2007) 237 1848-1861. Sato, Y., Sadatomi, M. & Sekoguchi, K., Momentum and heat transfer in two-phase bubbly flow – I, Int. J. Multiphase Flow (1981) 7 167-178. Tomiyama, A., Struggle with computational bubble dynamics, Proc Third Int. Conf. Multiphase Flow, Lyon, France (1998).

G.H. Yeoh, S.C.P. Cheung, J.Y. Tu

BRIEF BIOGRAPHY OF PRESENTER Dr Guan Heng Yeoh graduated with a BEng degree (First Class Honours) in Mechanical Engineering from University of New South Wales (UNSW), Australia, and obtained his PhD from the same university later on. He has over 15 years of research experience in industry, given many prominent seminar presentations at academic institutions/organisations world wide, and is currently a Senior Research Scientist at ANSTO and Associate Professor at UNSW. His areas of expertise are: Computational Fluid Dynamics (CFD), Multiphase Flows, Nuclear Reactor Safety, Fire Engineering and Mechanical Ventilation. He has written three books, attained the prestigious Brennan Medal from IChemE and a recognition award from NASA. Over 100 international refereed publications in areas of his expertise have been published. He is currently the Editor-in-Chief for Journal of Computational Multiphase Flows published by Multi-Science (UK).