1. Chemical & Physical Processes
in Glass Melting
Quality of glass
melting process
Buchmayer
Ruud Beerkens
TNO Glass Group
Eindhoven, The Netherlands
Glass Service
ICG – EFONGA Spring School Montpellier 4-5 May 2009
1
2. Contents
• Overview of processes essential for glass melting
• Tools of analysis of industrial glass melting
• Melting-in of Raw materials
• Kinetics of Sand Dissolution
• Removal of Gas bubbles & Dissolved gases
• Evaporation processes
• Homogenisation
ICG – EFONGA Spring School Montpellier 4-5 May 2009
2
3. ICG – EFONGA Spring School Montpellier 4-5 May 2009
3
4. refractory corrosion
evaporation:
NaOH, KOH,
NOx and
HBO2,
heat transfer
water infiltration PbO, NaCl,
HF, SO2 etc.
flue gas
chemistry
melting kinetics foaming
fining and redox
refractory corrosion
deposition and
dust formation
1. Overview
Chemistry & Physics
of Glass
Melting Processes
emissions: Na2SO4, Na2B4O7 and PbO dust
HCl, HF, SO2, SO3 , SeO2, HBO2, H3BO3 etc.
ICG – EFONGA Spring School Montpellier 4-5 May 2009
4
5. Side-effects
•
•
•
•
Evaporation from melt
Refractory corrosion
Combustion & heating
Emissions from evaporation & combustion
Furnace
Raw materials mixed
Or
Homogeneous molten glass
Crucible
Melting-in of batch – chemical reactions/endothermic effects
Sand* grain dissolution
Removal of bubbles/gases
Dissolution of seed (fine bubbles) residue
Homogenisation
- Diffusion (slow)
- Velocity gradients – stretching of inhomogeneities
ICG – EFONGA Spring School Montpellier 4-5 May 2009
5
6. Parameters for processes in melting
Important parameters:
• Heat transfer
Temperature
viscosity, surface tension,
chemical activity, reaction kinetics, gas evolution..
• Flow characteristics – convection in melt, stirring
• Residence time: time-temperature history
• Exposure of melt to (reactive) atmosphere and refractory
lining
How to assess:
Temperatures and flows in glass melt ?
ICG – EFONGA Spring School Montpellier 4-5 May 2009
6
7. 2. Tools: CFD Simulation Models Glass Furnaces
• Discretize total volume of furnace in small volume cells (> 1 million)
cells LxBxH: (1-10) x (1-10) x (1-10) cm
– GRID / mesh
• Input data process: pull, batch composition, fuel distribution, air
number
• Input furnace:
– Design
– Wall construction, including insulation
• Input glass: viscosity, heat conductivity, density, thermal
expansion, electric conductivity, solubility sand, solubility gases,…
• For each volume cell in tank & combustion chamber
– Energy conservation
– Momentum conservation
– Mass conservation (continuity) for melt and each chemical element
– Respect electro-neutrality
ICG – EFONGA Spring School Montpellier 4-5 May 2009
7
8. CFD Simulation Models
example: local conservation of energy
∂ (ρ c pT
∂t
) = − div (ρ c T v ) + div (λ
r
increase sensible
heat
p
convection
of heat
grad T ) + q
heat conduction
local
boosting or
cooling
Energy equation, conservation law for
energy in each volume element
ICG – EFONGA Spring School Montpellier 4-5 May 2009
8
9. Results of CFD simulation models
• Temperatures at all possible positions
– Combustion space
– Glass melt
– Refractory
• Glass melt and Combustion gas velocities
• Trajectories (particle tracing) in tank
– Thousands of different paths can be identified from charging end to throat or
spout
• Redox and dissolved gases
– Redox state of melt at each position (pO2 or Fe2+/Fe3+)
• Residence time distribution
– Minimum residence time is of importance for melting process
• Glass melt quality indices per trajectory
– Trajectory with minimum melting or fining index is decisive for glass
ICG – EFONGA Spring School Montpellier 4-5 May 2009
9
10. Application of CFD models
• For furnace design (lowest energy, highest glass quality)
–
–
–
–
Optimum depth of tank
Position bubblers or dam or burners
Size and design of throat
Design combustion chamber (LowNOx, less evaporation)
• For optimum process settings
–
–
–
–
Optimum fuel-boosting ratio
Temperature profile (energy distribution)
Bubbling rate
Creation of distinct spring zone to avoid short cut
• Time-transient (time dependent) for colour or pull change
– Optimize colour change process: reduce transition time
• Time-transient for process control (rMPC)
– Sensors give model continuous new information: model tracking
– Model continuously gives recommendation for input parameter changes to
follow optimum process path (low energy, high glass quality, constant T)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
10
11. Geometry & Grid
for computational fluid dynamics (CFD)
analysis of glass furnace
Port Necks
crown
Burner port
tank
Deep Refiner
Batch Boosting electrodes
ICG – EFONGA Spring School Montpellier 4-5 May 2009
11
12. Example result CFD computation
Temperature contours
ICG – EFONGA Spring School Montpellier 4-5 May 2009
12
13. NOx
End-port fired furnace horizontal cross section at level of burners
Base case
4 inch higher crown
ICG – EFONGA Spring School Montpellier 4-5 May 2009
13
14. NOx
End-port fired furnace vertical cross section at 25 % from furnace length from port
NOx scaling in mole fraction
Base case
4 inch
higher crown
Burner port
Exit port (flue gas)
Lower NOx-concentration in exit
ICG – EFONGA Spring School Montpellier 4-5 May 2009
14
15. Glass melt path with lowest ‘temperature‘ index
canal
Temperatur indexpath i
e
T
= ∫
dt
η
doghouse
Temperature course of glass (melt) in typical float glass furnace with
minium temperature index
1800
1600
Temperature in oC
1400
1200
1000
800
600
400
200
0
0
2
4
6
8
10
12
14
Time in hours
ICG – EFONGA Spring School Montpellier 4-5 May 2009
15
16. Sensors (short overview)
• Combustion gases
• gas composition:
• laser optics
• electrochemical sensors (oxygen)
• Glass melt
• chemical composition
• LIBS = laser induced breakdown spectroscopy
emf
• redox / colour parameters
t/c
• Potentiometry
• Voltammetry
Type B
(mV)
(mV)
Alumina rod
Pt / Ni-NiO // ZrO2 // pO2 (glass) / Pt
EMF =
RT
pO 2 (glass)
⋅ ln
nF
pO 2 (ref.Ni/NiO)
Pt measuring
electrode
ICG – EFONGA Spring School Montpellier 4-5 May 2009
Zirconia cell
Ni/NiOreference mix
16
17. 3. Melting-in of batch
• In glass furnaces (industrial)
– Kinetics determined by heat transfer through
batch blanket
• In small crucibles:
– Kinetics determined by contact between
different batch constituents and temperature
ICG – EFONGA Spring School Montpellier 4-5 May 2009
17
18. Return flow
for batch heating
Generation blisters
from refractory
Day
hopper
Refining
Bubble absorption
Hot spot &
evaporation
Conditioning of melt
Thermal homogeneity
Zone for sand
grain dissolution
Batch melting
• 40-60 minutes
• 80-90 % of net heat flux
Spring zone
& primary fining
Return flow
from working end
ICG – EFONGA Spring School Montpellier 4-5 May 2009
18
19. Scheme of melting process of batch blanket, charging velocity vg
(m/s)
combustion space
heat transfer
gas release
thickness
temperature
profile
reaction zone
figure 1b
glass melt layer
normal batch
Zipfel
glass
level
reaction zone
glassmelt
flow
heat
transferred
figure 1c
ICG – EFONGA Spring School Montpellier 4-5 May 2009
19
20. Detailed re-presentation of the batch melting process
in glass furnace
1500
o
C
Layer
glassmelt
melts
Dissolution
sand grains
loose batch
sand
grains
gas
melting
reactions
batch
melting
reactions
carbonates
(soda/lime)
gas
dissolution sand
grains
sand
grains
b. top of batch blanket
glass melt 1400
c.
o
C
bottom side of batch
blanket
ICG – EFONGA Spring School Montpellier 4-5 May 2009
20
21. Example:
Melting reactions of soda lime (dolomite) silica batch
• De-hydratation (100 oC physical bonded water & > 100 oC hydrates)
– Important for energy consumption: water evaporation is energy intensive
• Solid state reactions, formation of silicates, e.g.:
Carbonate route < 900 oC at fast heating rate
(melts at ±820 oC)
High amount of
heat required
(550-850 oC)
Na2CO3 + CaCO3
Na2Ca(CO3)2
Na2Ca(CO3)2 +2SiO2
Na2SiO3/CaSiO3 + 2CO2↑ reaction enhances > 820 oC
Na2CO3 + 2SiO2
Na2SiO3 + CO2↑
(790-850 oC)
• Formation of primary melt phases (alkali rich carbonates), e.g.:
Tm Na2CO3
Tm Na2Ca(CO3)2
Tm K2CO3
= 850 oC
= 820 oC
= 890 oC
ICG – EFONGA Spring School Montpellier 4-5 May 2009
21
22. Melting reactions of soda lime silica batch
limited kinetics may shift some reactions to higher temperatures
• Decomposition reactions of (Ca- and Mg-) carbonates:
heat required
CaCO3 + heat
CaO + CO2↑
MgCO3 + heat
MgO + CO2↑
MgCO3·CaCO3 + heat
MgO + CaCO3 + CO2↑
(910 oC at pressure 1 bar)
(540 oC at pressure of 1 bar)
(650 oC, 1 bar)
MgO still present up to 1150 oC.
• Dissolution reactions of SiO2, e.g. (coarse limestone)
Na2O·2SiO2 + CO2↑ T > 790 oC
forms with SiO2 an eutectic melt
Or at further heating
fast Na2O·SiO2 formation (850 oC) –
limestone decomposes and:
2CaO + (SiO2 + Na2O·2SiO2 )eutectic melt
Na2O·2CaO·3SiO2 (> 900 oC)
Reactive calcination: Na2CO3 + 2SiO2
Silicate route: Silicate melt + SiO2
silica enriched melt T > 1000-1100 oC
Eutectic melt phases are formed above ±800-840 oC
ICG – EFONGA Spring School Montpellier 4-5 May 2009
22
23. Phase diagram for the system Na2O – SiO2 showing
eutectic sodium silicate melt phases
100 % SiO2
ICG – EFONGA Spring School Montpellier 4-5 May 2009
23
24. Scheme of melting reactions of soda lime glass batch
1080 oC: T s Na2SiO 3
910 oC : CaCO 3
CaO + CO 2 (gas)
o
850 C : T s Na2CO 3
820 oC : T s Na2Ca(CO 3)2
790 oC : T eut Na 2O.2SiO 2 + SiO 2
740 oC : T eut Na 2Ca(CO 3)2 + Na2CO 3
650 oC : MgCO 3.CaCO 3
MgO+CaCO 3+CO 2 (gas)
540 oC : MgCO 3 -> MgO + CO 2 (gas)
Dissolution of SiO 2, CaO,
MgO, Al2O 3 e.d. in melt phases
primary melts
decomposition
carbonates
solid state reactions
volatilisation of water
0
200
400
600
800
1000
temperature in oC
1200
ICG – EFONGA Spring School Montpellier 4-5 May 2009
1400
24
25. Overall chemical energy demand
-1
CaCO3(s) -> CaO(s) + CO2(g)
Chemical energy consumption rate [kJ·kgbatch ·K ]
MgCO3·CaCO3(s) -> MgO(s) + CO2(g) + CaCO3(s)
-1
10
8
Na2CO3(s) + SiO2(q) -> Na2O·SiO2(s) + CO2(g)
Na2CO3(s) -> Na2CO3(l)
6
Na2CO3(l) + SiO2(q) -> Na2O·SiO2(s) + CO2(g)
Na2O·SiO2(s) + SiO2(q) -> NS(l)
CaO(s) + melt
4
2
0
600
650
700
750
800
850
900
950
1000
-2
Temperature [°C]
Chemical enthalpy of batch reactions for float glass
from soda-sand-dolomite and limestone (positive: endothermic effects)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
25
26. 4. Dissolution of ‘refractory’ type raw material
in silicate melt
example: sand grains
ICG – EFONGA Spring School Montpellier 4-5 May 2009
26
27. Sand grain
Glass melt
CSiO2
Ce(T) = saturation level SiO2 in melt
100
Cb
Ce (T)
Cb
= bulk SiO2 level in melt
(depends on amount
dissolved sand)
Moving boundary
Diffusion of SiO2 in melt
ICG – EFONGA Spring School Montpellier 4-5 May 2009
27
28. Dissolution front
One-dimensional dissolution in multi-component liquid
Mass fraction SiO2 in saturated melt: we
Multi-component liquid
Dissolving material
w is mass fraction SiO2 in melt
a
x
dissolution
we
∂w
∂x a
w
ρ SiO2
∂w
ρe ⋅
da
∂x a
⋅
= − D⋅
dt
(1 − VA ⋅ ρ e ⋅ w e )
ICG – EFONGA Spring School Montpellier 4-5 May 2009
28
29. Mathematical description
(Ready & Cooper 1966)
-
Spherical symmetry – diffusion in 3 dimensions
Assuming constant diffusion coefficient in melt
Ideal solution, partial molar volume of SiO2 in melt is constant
Convection (term u) due to change of partial molar volume of SiO2 in sand versus in melt
Moving boundary: dissolving sand is partly staying in volume it came from
Mass flux (j) of dissolved SiO2
J
D
r
R
t
ρ
C
u
= mass flux of SiO2
= diffusion coefficient of SiO2 in silicate melt (m2/s)
= radial co-ordinate (distance from sand grain centre) (m)
= radius sand grain (m)
= time (s)
= density of melt (kg/m3)
= local SiO2 mass concentration (kg/m3)
= mass average velocity radial direction due to expansion by dissolution
(change in molar volume) (m/s)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
29
30. Solution dissolution sand grain
without forced convection
Effect of moving boundary
a
=
actual grain size radius (m)
VA
=
partial specific volume of SiO2 in molten glass (m3/kg)
Cs
=
density of sand grain (kg/m3)
Ca
=
mass concentration SiO2 in saturated melt (kg/m3)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
30
31. Sand grain dissolution with convection by
glass melt velocity gradients & density
differences
ρ SiO2
h
we
ws
VA
a
t
ρs
ρSiO2
D
da
⋅
= − h ⋅ (w e ρ e − w sρ s ) /(1 − VA ⋅ w e ⋅ ρ e )
dt
= mass transfer coefficient SiO2 into melt (m/s)
= mass fraction SiO2 in saturated melt (depends on T, and glass) (kg/m3)
= mass fraction SiO2 in bulk melt (depends on dissolved sand)) (kg/m3)
= partial specific volume of SiO2 in molten glass (m3/kg)
= actual radius sand grain (m)
= time (s)
= density of melt (kg/m3)
= density sand grain (kg/m3)
= diffusion coefficient of SiO2 in silicate melt (m2/s)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
31
32. Mass transfer coefficient
Sh
1
1
h=
⋅D ⋅ +
R
2
Sh
⋅D ⋅ t
π⋅
2
h
=
D =
mass transfer coefficient (m/s)
diffusion coefficient of SiO2 in the molten glass, based on concentration
profiles given in mass fraction (D in m2/s)
R =
grain radius (m),
t
time (s),
=
Sh =
Sherwood number for mass transfer from spherical grain,
≈
2 + 0.89 {Re · Sc + (Gr · Sc)3/4}1/3 *
=
2 (no convection)
=
f (R2/3 , D-1/3, (grad v)1/3) (convection flow of the glass melt)
=
f (R3/4 , h-1/4 , D-1/4)
(free convection of surrounding melt relative to the sand grain: v = flow velocity of the
melt relative to the sand grain (m/s), η = viscosity (Pa.s)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
32
33. 30000
no convection
Dissoltion time in s
25000
20000
v-gradient
15000
-1
0.001 s
10000
v-gradient
5000
0.025 s
0
1350
-1
1450
1550
1650
1750
1850
T in K
Dissolution time required for complete dissolution of sand grains in
almost static and stirred soda-lime silica glass melts (forced convection
with velocity gradient grad v) at different temperatures. Initial size Ao=100 mm.
ICG – EFONGA Spring School Montpellier 4-5 May 2009
33
34. Dissolution of sand and alumina grains in static and convective sodalime-silica glass melt at 1700 K, moving boundary effect taken into
account (not on concentration profiles)
1.20E-04
1.00E-04
sand,
grad v=0
sand,
grad v= 0 steady state
sand,
-1
grad v=0.001 s
radius in m
8.00E-05
sand,
-1
grad v=0.01 s
6.00E-05
alumina,
-1
grad v=0.01 s
4.00E-05
alumina,
-1
grad v=0.001 s
sand,
steady state
-1
grad v=0.001 s
2.00E-05
0.00E+00
0
5000
10000
15000
20000
alumina,
grad v = 0
25000
30000
35000
time [s]
ICG – EFONGA Spring School Montpellier 4-5 May 2009
34
39. Bubbles & Seeds just after melting
•
Many small seeds and bubbles (Blisters) in glass melt
combustion space
(Mulfinger 1976 GTB)
heat transfer
gas release
– More than 100.000 per kg glass melt
thickness
reaction zone
figure 1b
glass melt layer
temperature
profile
glass
level
normal batch
– Most bubble diameters: 0.05 -0.4 mm
reaction zone
glassmelt
flow
•
heat
transferred
figure 1c
In most glass melts (using carbonates):
– bubbles in batch melting area: contain often mainly CO2
•
Large concentrations dissolved CO2 in melt
•
During sand grain dissolution in melt: generation
of fine CO2 seeds (Gispen)
from Glass Service
ICG – EFONGA Spring School Montpellier 4-5 May 2009
39
40. Fining
Objective of Fining:
Removal of bubbles and dissolved gases from the glass melt
Rising velocity of bubble:
v ascension
ρ =
Density of the glass melt [kg/m3]
η =
Viscosity of the melt [Pa·s]
R =
Bubble radius [m]
g =
Acceleration of gravity [m/s2]
c =
c ⋅ ρ ⋅ g ⋅R
=
η
2
Factor (e.g. Stokes c = 2/9)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
40
41. Time to reach glass level at 1 meter [h]
Fining
Time to reach glass surface (1 meter)
250
1400 OC
200
1450 OC
1350 OC
150
100
1500 OC
50
0
0
100
200
300
400
500
Bubble diameter [µm]
ICG – EFONGA Spring School Montpellier 4-5 May 2009
41
42. III. Fining gases and
other dissolved
gases diffuse
strongly into bubble
II. start of fining:
gases diffuse into
bubble
I. static bubble
Reaction in melt: release of fining gases
Pgases melt > pt (pt is pressure in bubble)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
42
43. Two fining steps
•
First step: primary fining
– High temperatures
– Bubble agglomeration and bubble size growth
– Dissolved gases diffuse from melt in to bubbles (like
bubbles in soda drinks)
– Ascension to glass melt surface
• Second step: Secondary fining/Refining (secondary fining)
– Dissolution of (small) remaining bubbles
• Only effective if bubble contains gases (CO2, O2, SO2+O2)
that dissolve in cooling melts
• Glass melt should be lean in dissolved gases
ICG – EFONGA Spring School Montpellier 4-5 May 2009
43
44. Mechanism sulfate primary fining
Fining process in glass melt
pSO2 · pO21/2 ·aNa2O
K=
pSO2· pO21/2
K’ =
aNa2SO4
[SO3]
Increasing temperatures lead to increasing K-values →
extra oxygen gas & SO2 gas release:
- oxygen & SO2 molecules diffuse into growing bubbles
- bubble ascension increases (vascension~R2)
- sulfate retention decreases
ICG – EFONGA Spring School Montpellier 4-5 May 2009
44
45. Fining (primary)
– Fining agents added to the batch to enhance the
rising velocity of bubbles
– Often used fining agent: Sodium sulphate
Fining reaction: T > TFining onset
Na2SO4 ⇔ Na2O + SO2 (gas) +1/2 O2 (gas)
Cm CO2
pSO2 ⋅ pO2
K =
[SO3 ]
'
CO2
Stripping of CO2
and N2 from melt
Cm N2
SO2
O2
N2
Dilution of N2 & CO2 in bubble by fining gases
ICG – EFONGA Spring School Montpellier 4-5 May 2009
45
46. Multicomponent diffusion of gases in bubbles
d
Shi Di
1
1
⋅ ⋅ Rg ⋅ T ⋅ (Csi − Cii) ⋅ +
(4πR3 ⋅ pt ) /(3Rg ⋅ T) = 4πR2 ⋅ Σi
R
dt
2 pt
Shi
⋅ Di ⋅ t
π⋅
2
[
]
Shi = 1+ (1+ 2·v·R/Di )1/3
ICG – EFONGA Spring School Montpellier 4-5 May 2009
46
47. Fining/Refining: degassing & removal of bubbles
• Mostly applied fining agents in glass industry: Na2SO4 & Sb2O5
– Na2SO4 (m) → SO2 (g) + 0.5 O2 (g) + Na2O (m)
– Sb2O5 (m) → Sb2O3 (m) + O2 (g)
• Na2SO4 added in concentrations 0.1 – 1 wt. % to batches of:
– Soda lime glass for container, float and tableware
– E-borosilicate glass for fibres
• Na2SO4 partly decomposes during batch melting & releasing SO2 in
early melting stages
• Dissociation temperature of Na2SO4 in melt:
– Between 1350 – 1480 ºC, depending on redox state
– Between 1100-1350 oC (reduced batches) Na2SO4+Na2S
reactions forming SO2 and or S2 gas.
ICG – EFONGA Spring School Montpellier 4-5 May 2009
47
48. Fe2+/Fetotal
Sulfur retention (wt.%SO3)
0.6
0.5
80
70
60
Sulfur only in Sulfur in
form of S2form of SO42-, S2-
40
25
15
%
Sulfur only in
form of SO42-
(probably also SO32-?)
0.4
0.3
1400 oC
0.2
0.1
1500 oC
0.0
-8
-7
-30
-6
-5
-4
-3
Log pO2 in the melt at 1400°C (bar)
-20
-10
0
+10
-2
-1
+20
redox number
ICG – EFONGA Spring School Montpellier 4-5 May 2009
48
50. High temperature test facility
ICG – EFONGA Spring School Montpellier 4-5 May 2009
50
51. Fining/Refining: degassing & removal of bubbles
1. Primary fining
– Removal of bubbles by rising of bubbles to melt
surface
– Bubble growth under influence of fining agents
– Stripping of dissolved gases by growing of gas bubbles
(dilution)
Fining
ICG – EFONGA Spring School Montpellier 4-5 May 2009
51
52. Enhanced Sulfate Fining by
Dissolved Water in Melt
O2
N2
CO2
O2
H2O
H2O
Oxy-case
SO2
SO2
O2
N2
CO2
O2
SO2
H2O
Air case
SO2
In oxygen-fired glass furnace:
peH2O = 0.25-0.40 bar
Fining only if:
peSO2 + peO2 > 0.70 - 0.75 bar
In air-fired furnace:
peH2O = 0.10-0.15 bar
Fining only if :
peSO2 + peO2 > 0.9 bar
ICG – EFONGA Spring School Montpellier 4-5 May 2009
52
53. Evolution of fining gas: water lean & rich melts
50
water vapor pressure
40
0 bar
30
0.20 bar
20
0.60 bar
3
m /batch
Volume of gas in
Gas evolution during sulfate fining of soda lime glass
melt - effect of water vapor level 60
10
0
1300
1400
1500
1600
o
Temperature in C
ICG – EFONGA Spring School Montpellier 4-5 May 2009
53
54. Stripping of dissolved gases from melt
ICG – EFONGA Spring School Montpellier 4-5 May 2009
54
55. Pressure in melt before and during fining & cooling
Total internal pressure in melt
bar
[S]initial= 0.3 mass% SO3 300 mgr water/kg [Fe2+]/ [Fetotal] = 20 %
1.2
1.0
0.8
0.6
0.4
0.2
0.0
1200
1300
1400
1500
1600
o
Temperature in C
ICG – EFONGA Spring School Montpellier 4-5 May 2009
55
56. Partial pressure in float glass melt during heating
Partial pressure in float glass melt
in bar
[S]initial = 0.3 mass% SO3 300 mgr water/kg [Fe2+]/ [Fetotal] = 20 %
1.E+00
SO2
1.E-01
H2 O
1.E-02
O2
1.E-03
CO2
1.E-04
N2
1.E-05
1250
1350
1450
1550
o
Temperature in C
ICG – EFONGA Spring School Montpellier 4-5 May 2009
56
58. Fining/Refining: degassing & removal of bubbles
2. Secondary fining (refining)
– Re-absorption of residual gases during controlled cooling
• Chemical solubility SO2 and O2 increases with
decreasing temperature: gases will be re-absorbed
during cooling.
• Physical solubility of dissolved gases increases slightly
with decreasing temperature: these gases will also be reabsorbed during cooling
Refining
ICG – EFONGA Spring School Montpellier 4-5 May 2009
58
59. Fining at low pressure
- Same amount of gas needs large volume
- Low partial pressures in bubble will stimulate gas diffusion
from melt into bubble
ICG – EFONGA Spring School Montpellier 4-5 May 2009
59
61. Multi-component glasses
• Different oxides
• Oxides in glass with high chemical activity or vapour
pressure:
– React at glass melt surface with combustion gases
– Evaporate from glass melt surface
– Show depletion at surface layer
INCONGRUENT EVAPORATION
ICG – EFONGA Spring School Montpellier 4-5 May 2009
61
62. Evaporation mechanisms
•
Direct evaporation of volatile glass components from the surface of the melt,
e.g. volatilization of PbO from lead crystal melts;
•
Evaporation of components by reactions in the melt itself, forming volatile
compounds; generally such volatile compounds exhibit high activity coefficients
and weak bonding with other glass melt species.
–
•
An example: formation of alkali borates in alkali borosilicate melts,
subsequently evaporation of alkali meta-borates/tetra-borates takes place,
Na2O(melt) + B2O3 (melt) ⇔ 2NaBO2 (melt) ⇒ 2 NaBO2 (vapor)
Evaporation by reactions of certain glass melt components with gas
species at the surface of the melt.
The evaporation rate & vapor pressure depends on the composition of
the gas atmosphere above the melt.
B2O3 (glass melt) + H2O
⇒ 2HBO2 (vapor)
Na2O (glass melt) + H2O (gas)
⇒ 2NaOH (vapor)
Na2O(glass melt) + CO (gas)
⇒ 2Na (vapor) + CO2
ICG – EFONGA Spring School Montpellier 4-5 May 2009
62
63. Kinetics of evaporation
The volatilization rate is often determined by:
• mass transport of the relevant species from the
melt interior (bulk) to the surface;
• the vapor pressures of the volatile components at the
surface of the melt, dependent on the glass composition,
temperature and gas atmosphere;
• the mass transfer of evaporated species from the surface
of the melt into the main gas stream above the melt.
ICG – EFONGA Spring School Montpellier 4-5 May 2009
63
64. Reactive evaporation
For reactive evaporation of a component j in the melt reacting with
a gas k and forming gaseous species i with saturation pressure
pi*, according to reaction:
n·j (melt) + m·k (gas)
Reaction equilibrium: pi*q
q·i (gas)
= K · ajn · pkm
The values of K (equilibrium constant, assuming chemical
equilibrium at the glass melt surface) and aj (activity of component j
in the molten glass at the surface) can be determined
experimentally or by thermodynamic modeling
ICG – EFONGA Spring School Montpellier 4-5 May 2009
64
65. Static melt and static atmosphere above the
melt (interface x = 0)
Diffusion in melt of reacting glass melt component j:
dCj/dt = Dm,j·δCj2/δx2
Time dependency: Cjsurface(t).
Dm,j is the inter-diffusion coefficient of the volatile component j in the melt.
at t = 0
at t > 0
at t > 0
-∞ < x < 0
x
−∞
x =0
Cj = Cjbulk
Cj = Cjbulk
Cj = Cjsurface(t)
For the vapor i in a static gas phase with partial vapor pressure pi,
the diffusion process in the gas phase can be described in a similar way:
δ(pi/RgT)/δt = Dg,i ·δ2(pi/RgT) /δx2
Dg,i is the diffusion coefficient of the vapor I in the gas phase.
at t = 0
at t > 0
at t > 0
0<x<∞
x
∞
x =0
pi = pi,gasbulk
pi = pigasbulk
pi = pi*(t)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
65
66. -3
Na2 O concentration [mol.m ]
4970
4968
4966
4964
4962
5 sec.
4960
50 sec.
250 sec.
4958
4956
DNa2O=3.3 10-11 m 2.s -1
4954
0
0.01
0.02
0.03
0.04
0.05
0.06
distance from surface [mm]
Calculated time dependent- Na2O concentration profiles in static melt
Situation: static conditions in semi-infinite gas phase with 0.55 bar vapor pressure
and semi-infinite soda-lime-silica melt
(13 wt% Na2O, 10 wt% CaO, 5 wt% MgO, 72 wt% SiO2).
Dg,NaOH = 2.7 10-4 m2·s-1 , Dm,Na2O = 3.3 10-11 m2·s-1
ICG – EFONGA Spring School Montpellier 4-5 May 2009
66
67. Evaporation in gas flow above molten glass
pb,i
main gas stream
velocity, vg
diffusion of gas i in
gas boundary layer, Dgi
surface reaction:
n·j (melt)+ m·k (gas) ⇒ q·i (gas) p*i
glass melt surface
C j-profile
Example:
Na2O (m) + H2O(g) ⇔ 2NaOH (g)
Transport of component
j in the melt, Dm,j
ICG – EFONGA Spring School Montpellier 4-5 May 2009
67
68. Evaporation in gas flow
Average evaporation rate (over length Lg of gas flow above
melt ) of component i (formed by reaction of glass
compound j) into (turbulent) gas phase:
Qm,j =(ni/qi)·A·vg0.8·ρg0.47·µgas-0.47·Dg0.667·Lg-0.2· Rg-1·T-1·B·Cj,x=0(t)
The proportionality parameter B depends on the furnace
atmosphere composition and the chemical activity of the
volatile component in the melt.
For NaOH-evaporation, the B value depends on the water
vapor pressure in the furnace atmosphere, B ∼ pH2O0.5
ICG – EFONGA Spring School Montpellier 4-5 May 2009
68
69. Mass transfer equations
Average evaporation rate (rate of loss of glass component j)
over length Lg from leading edge:
Qm,j = -Dm,j·(δCj/δx)x=0 = α·Cj,x=0(t)
α = (ni/qi) · A· vg0.8·ρg 0.47·µgas-0.47·Dg0.667·Lg-0.2· Rg-1·T-1· B
Turbulent flow of gas
v = velocity, g refers to gas phase, Rg is universal gas constant, T in K,
B ratio between vapour pressure i and surface concentration component j
A = between 0.03 and 0.04 for turbulent gas flow (Re > 300000 or for disturbed flows)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
69
70. Solution in flowing gas and static melt
For kd defined as α/DmNa2O the solution for a single component j
Evaporating from a static melt in flowing gas phase
Assuming complete depletion at surface for t
∞
MQm.j is the total evaporation mass loss per unit surface area between
time 0 and τ
MQm.j = (Cj,0/kd)·{exp(kd2·DmNa2O·τ)·erfc[kd·(DmNa2O·τ)0.5] -1
+ 2kd·(DmNa2O·τ/π)0.5}
Cj,x=0(t) = Cj,0· exp(kd2·DmNa2O·τ)·erfc[kd·(DmNa2O·τ)0.5]
Cj,0 = bulk concentration compound j at t=0
ICG – EFONGA Spring School Montpellier 4-5 May 2009
70
71. 5000
-3
melt [mol.m ]
Na2O concentration in
5500
4500
stagnant gas
4000
Lg= 2 m, v= 2 m.s-1
Lg= 2 m, v = 5 m.s-1
3500
Lg= 0.5 m, v= 2 m.s-1
3000
0
0.5
1
1.5
2
2.5
distance from surface [mm]
Local concentration profile in soda-lime silica melt after 7200 seconds exposure time,
calculated for NaOH-evaporation from static melt in static or flowing gas phases,
(Lg= downstream distance from leading edge ).
Temperature = 1500 oC, pH2O = 0.55 bar. Dm,Na2O = 3.3 10-11 m2·s-1
Glass composition (mass %): SiO2 =72, Na2O =13, MgO = 5, CaO = 10
ICG – EFONGA Spring School Montpellier 4-5 May 2009
71
72. parameter is temperature:
5000
1723K
1773 K
1823 K
4000
-3
[mol.m ]
Na2O-surface concentration
6000
1873 K
3000
2000
1000
0
0
5000
10000
15000 20000
time [s]
25000
30000
Change in Na2O-surface concentration soda-lime-silica melt at different
temperatures in flowing gas (5 ms-1), 1 meters downstream.
pH2O in gas = 0.55 bar & Na2O in glass = 13 mass%.
Dm,Na2O= 8 10-10 exp(-5655/T) in upper graph
ICG – EFONGA Spring School Montpellier 4-5 May 2009
72
73. Experimental – Set up to study
(reactive) evaporation from molten glass
Thermocouples
Platinum funnel
Porous
plate
Gases IN:
Platinu
m gas
samplin
g probe
Platinum
coating (30 cm)
Platinum boat
N2, H2O,
O2
melt
ICG – EFONGA Spring School Montpellier 4-5 May 2009
73
74. Mass transfer in gas phase during
transpiration – evaporation test
Shtranspiration
0 .5
2
1 .5
= C1 + C 2 ⋅ Re d ⋅ Sc +
⋅ C3 ⋅ (Re d ⋅ Sc )
1 + 22 ⋅ Sc
h g ,i =
Q g ,i =
Vessel with
liquid of melt
1
3
Shtranspiration ⋅ D g ,i
d
hg ,i
R ⋅T
(
⋅ pi (t ) * − pi bulk
)
2.5E-03
x=0
pi*(t) (e.g. p*NaOH or p*NaBO2)
can be derived from evaporation
(transpiration experiments)
From measured Qg,i and Sherwood
relations derived with model liquids
Water evaporation rate
QH2O (moles s -1 m-2)
2.0E-03
1.5E-03
Measurerments
1.0E-03
CFD model
Empirical equation (2.19)
5.0E-04
0.0E+00
0
100
200
Reynolds number
Re (-)
ICG – EFONGA Spring School Montpellier 4-5 May 2009
300
400
74
75. 1.E-06
Modeled aNa2O (-)
SiO2:Na2O:CaO = 74:16:10 mol
1.E-07
Na2O.2SiO2
y = 0.9823x
R2 = 0.9283
1.E-08
1.E-09
1.E-09
1.E-08
1.E-07
1.E-06
Measured aNa2O (-)
Na2O activity at glass melt surface determined by transpiration test
measuring p*NaOH: Na2O + H2O
2 NaOH
ICG – EFONGA Spring School Montpellier 4-5 May 2009
75
76. Derivation chemical activity of volatile glass
component at surface of melt
K = exp(-∆G/RT) = p*NaOH2/aNa2O.pH2O
From thermodynamic tables:
∆G = GfNa2O+GfH2O- 2GfNaOH(g)
p*NaOH is measured from QNaOH and pH2O is
controlled
aNa2O (surface) can be determined
K is calculated by standard Gibbs free energy values of
products & reactants of reaction
ICG – EFONGA Spring School Montpellier 4-5 May 2009
76
77. Non static melt & non static gas phase
free convection by density gradients
Gas flow
Low Na2O
Mid Na2O
High Na2O
Float glass melt with Na2O concentration differences
ICG – EFONGA Spring School Montpellier 4-5 May 2009
77
78. 6. Homogeneity of glass
Micro-mixing: transfer from high to low chemical activity by diffusion
Cm
- π 2 . D.t
= A exp
Co
Lo 2
Macro-mixing: elongation of in-homogeneities exposed to velocity gradient in melt
C(x,t)
dC
d 2C
= D⋅ 2
dt
dx
t=0
Cm (t)
Slow diffusion processes
t = t1
t = t2
Co
Lo
ICG – EFONGA Spring School Montpellier 4-5 May 2009
78
79. Macro-mixing
Reduction of diffusion distance, Lo
velocity v + dv
velocity v
y
In the case, t dv/dy >> 1:
L=
L0
dv
⋅
t
dy
For macro-mixing in combination with diffusion (by approximation):
Cm
π 2 . D . t 3 .( dv / dy ) 2
= A .exp −
Co
L2
0
A
= proportionality factor dependent on the shape of the cord
ICG – EFONGA Spring School Montpellier 4-5 May 2009
79
80. Macro versus only Micro mixing
• Small velocity gradients (> 0.01 m/s per m) enhance
homogenisation process with factor 20 to 100
• Velocity gradients by:
– Stirring
– Bubbling
– Temperature gradients
free convection
ICG – EFONGA Spring School Montpellier 4-5 May 2009
80
81. Thank you for your attention
What does a number tell us without the proper unit?
ICG – EFONGA Spring School Montpellier 4-5 May 2009
81