• No results found

Mathematical modelling of alumina feeding

N/A
N/A
Protected

Academic year: 2022

Share "Mathematical modelling of alumina feeding"

Copied!
20
0
0

Laster.... (Se fulltekst nå)

Fulltekst

(1)

Mathematical Institute Universit y of Oxfo rd

Mathematical modelling of alumina feeding

Attila Kovacs

Chris Breward, James Oliver and Andreas Münch (Oxford) Svenn Halvorsen and Ellen Nordgård-Hansen (Norce)

Eirik Manger (Norsk Hydro)

March 14, 2019

(2)

Mathematical Institute Universit y of Oxfo rd Hall-Héroult cell

Alumina (300K)

Electrolysis in the hot cryolite (1200K)

Cathode Molten aluminium

Anode Anode

Tapping Crust

Current (200 kA)

Insulation

(3)

Mathematical Institute Universit y of Oxfo rd Alumina feeding processes

A

time

C

E D

B

How does the molten cryolite

infiltrate and dissolve a porous

alumina structure?

(4)

Mathematical Institute Universit y of Oxfo rd Previous Analysis: Solid alumina particle

Molten Cryolite Stefan Condition Thermal Contact r

t Alumina particle

Frozen Cryolite

Dissolution

t

D

t

M

t

F

R

f

Freezing time Melting time Dissolution time R

0

t

F

a2πc2AkρAA

t

M

a2c3kAρA(Tm−TA)

c(Tc−Tm)

t

D

a22D(Cρa(1−Csc)

s−C0)

(5)

Mathematical Institute Universit y of Oxfo rd Alumina feeding: raft problem

Fluid Saturated

Uninfiltrated

Dissolution boundary Infiltration boundary

Front movement Mush?

A

time

C

E D B

(6)

Mathematical Institute Universit y of Oxfo rd Alumina feeding: raft problem

x y

Uninfiltrated

Saturated

Fluid

Mush

(7)

Mathematical Institute Universit y of Oxfo rd 1D Infiltration problem

T

c

T

m

y Front movement

φ

0

(top)

Solid Alumina Liquid Cryolite

Solid Cryolite

Air 1

Volume fraction Temperature

y = H y = c(t )

y = b(t)

φ

0

+ ψ

(8)

Mathematical Institute Universit y of Oxfo rd 1D Infiltration problem: uninfiltrated region

c(t)

b(t) H

y 3 For c (t) < y < H :

∂T

3

∂t = k

ρ

A

c

A

φ

2

T

3

∂y

2

At y = c (t ) : T

3

= T

m

, ρ

c

ψ c ˙ = − k

3

∂T

3

∂y At y = H : k

3

∂T

3

∂y = h(T − T

e

)

(9)

Mathematical Institute Universit y of Oxfo rd 1D Infiltration problem: mushy region

c(t)

b(t) H

y 2

For b(t) < y < c (t) :

∂ψ

∂t = 0

∂y ((1 − φ − ψ)u

2

) = 0

−K (ψ + φ) µ( 1 − φ − ψ)

∂p

2

∂y + ρ

c

g

= u

2

At y = c (t) : p

2

= p

a

− σ

c

r

pore

, u

2

= ˙ c

1 + ψρ

1 − φ − ψ

At y = b(t) : [p

i

] = 0 , u

1

− 1 − φ − ψ

1 − φ u

2

= ψ b ˙ 1 − ρ

1 − φ

(10)

Mathematical Institute Universit y of Oxfo rd 1D Infiltration problem: infiltrated region

c(t)

b(t) H

y For 0 < y < b(t): 1

∂T

1

∂t + ∂

∂y

1 − φ

a u

1

T

1

= k

1

A

c

A

2

T

∂y

2

∂y ((1 − φ)u

1

) = 0

− K (φ) µ(1 − ψ)

∂p

1

∂y + ρ

c

g

= u

1

At y = b(t) : T

1

= T

m

, ρ

c

ψ b ˙ = − k

1

∂T

∂y , [p

i

] = 0, u

1

− 1 − φ − ψ

1 − φ u

2

= ψ b ˙ 1 − ρ

1 − φ ,

At y = 0 : T

1

= T

h

, p

1

= ρgH

(11)

Mathematical Institute Universit y of Oxfo rd Full dimensionless problem

y

t H

b(t)

∂T3

a(t)

∂t

=

γPe1

T2

2T3

∂y2

∂T1

∂t

+

∂y

1−φ

α

u

1

T

1

=

αγPe1

T

2T1

∂y2

∂y

(Φu

2

) = 0

∂y

((1 − φ)u

1

) = 0 u

1

=

−K(φ)1−φ

∂p

1

∂y

+ β

∂ψ

∂t

= 0,

u

2

=

−K(ψ+φ)1−φ−ψ

∂p

2

∂y

+ β T

2

= 1

1 2 3

(top)

0

T

1

= 1 + ε T

i

= 1 T

i

= 1

∂T∂y3

= Nu(T

3

− θ) p

1

= 0 [p

i

]

+

= 0 p

2

= −1

γψ c ˙ = −

∂T∂y2

γκψ b ˙ = −

∂T∂y1

u

2

= ˙ c

1 +

ψρΦ

u

1

1−φΦ

u

2

= ψ b ˙

1−ρ1−φ

y = 0 y = b(t) y = c(t) y = 1

(12)

Mathematical Institute Universit y of Oxfo rd Dimensionless parameters

Meaning Symbol Equation Value

Infiltrating force γ KP τ /(µ

c

H

2

) 40–400

Newton cooling Nu hH /k

3

Condictivity ratio κ k

/k

m

0.2

Thermal diffusivity Pe

T

H

2

ρ

c

c

c

/(k

m

τ ) 1.3 Thermal diffusivity Pe

T2

H

2

ρ

A

c

A

φ

τ k

A

(k

air

/k

A

)

1−φ

1.77

Particle coldness θ T

A

/T

m

0.246

Overheat ε

T

T

c

/T

m

− 1 0.015

Density ratio ρ ρ

f

c

1.00966

Importance of gravity β ρ

c

gH/P 0.01

Energy stored in phases α φ

ρρAcA

ccc

+ ( 1 − φ) 0.81

(13)

Mathematical Institute Universit y of Oxfo rd Small time asymptotics

• Expect b, c = O ( √

t) as t → 0

+

, so need small t asymptotics to initiate numerical solution.

• Seek similarity solution given T

i

∼ f

i

(η), u

i

∼ g √

i

(η)

t , p

i

∼ h

i

(η), ψ ∼ ψ

c

, b ∼ 2λ

b

t, c ∼ 2λ

c

√ t with η = y/2 √

t = O ( 1 ) as t → 0

+

and λ

b

, λ

c

, ψ

c

to be determined.

t

y O( √

t)

c(t) b(t)

1

(14)

Mathematical Institute Universit y of Oxfo rd Transcendential system for λ b , λ c , ψ c

K

2

c

) = −2 (λ

b

− λ

c

)g

2

( 1 − φ − ψ

c

) + λ

b

g

1

( 1 − φ) K

2

c

) K

1

γ 2κψ

c

λ

b

= 2 √

Pe

T1

e

−PeT1(ag1b)2

√ π erf √

Pe

T1

(ag

1

+ λ

b

)

− erf ag

1

√ Pe

T1

2γψ

c

λ

c

= − 2(θ − 1) √

Pe

T2

e

−PeT2λ2c

√ πerfc λ

c

√ Pe

T2

where g

1

and g

2

constants given by

g

1

= λ

c

+ (λ

c

− λ

b

)(ρ − 1 )ψ

c

1 − φ g

2

= λ

c

1 + ψ

c

ρ 1 − φ − ψ

c

(15)

Mathematical Institute Universit y of Oxfo rd Constraint surfaces

Two physical solutions with γ = 200 and Pe

T2

= 1/7.

(16)

Mathematical Institute Universit y of Oxfo rd Similarity solution for γ = 200, Pe T 2 = 0.14

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0

0.5 1

T

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0

0.1 0.2 0.3 0.4

η

ψ

(17)

Mathematical Institute Universit y of Oxfo rd Sanity check: sublimits

φ

0

Solid Alumina

Liquid Cryolite Air

1

No cooling and heating (ε, θ = 1) φ

0

Solid Alumina Liquid Cryolite Solid Cryolite

Air 1

φ

0

+ ψ

No heating (ε = 0, λ

b

= 0)

y

y

Volume fraction

(18)

Mathematical Institute Universit y of Oxfo rd Comparison with numerical results

(19)

Mathematical Institute Universit y of Oxfo rd Nonexistence of Solutions

No physical solutions with γ = 75 and Pe

T2

= 0.14 due to no intersection in the physically admissible region.

No physical solutions with γ = 75 and Pe

T2

= 1.7 due to the Stefan condition (blue).

ψ > 1 − φ, meaning freezes more

than the available fluid.

(20)

Mathematical Institute Universit y of Oxfo rd Conclusions and Future work

Conclusions:

• Developed a simple model for the infiltration of molten cryolite into a cold porous alumina structure (with phase change).

• Similarity solution at small times yields nonexistence and nonuniqueness.

• Preliminary numerical simulation suggest one solution is admissible while the other may blow up.

Future work:

• Characterise regions for similarity solution

• Investigate small time behaviour numerically: Alternatives to similarity solution?

• Are we missing physics e.g.

• capillary pressure dependence on porosity

• dropping local thermodynamical equilibrium

• including composition effects

Referanser

RELATERTE DOKUMENTER

It is found how the anaerobic portion of total energy depends on time, and a sprint cross-country skiing scenario is used to study how racing time and recovery depends on

In general, fewer large agglomerates are observed after horn sonication, compared to bath sonication, indicating that the higher energy input used in the horn sonication is

Mathematical modelling and computer simulation of evolutionary processes (genetic drift, selection, immigration) for use in genetic resource management. At Norwegian

This work is supported by the EPSRC Centre For Doctoral Training in Industrially Focused Mathematical Modelling (EP/L015803/1) in collaboration with NORCE and Hydro

In this limit the inltrated region does not exist, so only mushy region and uninltrated is important.. MathematicalInstituteUniversityofOxford Small overheat ( ε

• Der Derian, James (1987), On Diplomacy: A Genealogy of the Mediation of Estrangement, Oxford: Basil Blackwell.. • Der Derian, James (1992), Antidiplomacy: Spies, Terror, Speed

Oliver Berget Skjølsvik (Skanska), Eirik Haram (Skanska) og Sverre Smeplass (Skanska) Beregning av rissrisiko CrackTeSt COIN.. FA:

helsesektoren - en studie av bedrifter med OFU-kontrakt. Thomas Halvorsen og