Journal of Non-Crystalline Solids 358 (2012) 35593562

Contents lists available at SciVerse ScienceDirect

Journal of Non-Crystalline Solids

journal homepage: www.elsevier.com/ locate/ jnoncrysol

Nuclear waste vitrication efciency: Cold cap reactions

P. Hrma a, b,, A.A. Kruger c, R. Pokorny d
a

Division of Advanced Nuclear Engineering, Pohang University of Science and Technology, Pohang, Republic of Korea
Pacic Northwest National Laboratory, Richland, WA 99352, USA
c
U.S. Department of Energy Hanford Tank Waste Treatment and Immobilization Plant Federal Project Ofce, Engineering Division, Richland, WA 99352, USA
d
Institute of Chemical Technology, Prague, Czech Republic
b

a r t i c l e

i n f o

Article history:
Received 26 July 2011
Received in revised form 5 January 2012
Available online 7 March 2012
Keywords:
Glass melting;
Glass foaming;
Waste vitrication;
Cold cap

a b s t r a c t
Batch melting takes place within the cold cap, i.e., a batch layer oating on the surface of molten glass in a
glass-melting furnace. The conversion of batch to glass consists of various chemical reactions, phase transitions, and diffusion-controlled processes. This study introduces a one-dimensional (1D) mathematical
model of the cold cap that describes the batch-to-glass conversion within the cold cap as it progresses in a
vertical direction. With constitutive equations and key parameters based on measured data, and simplied
boundary conditions on the cold-cap interfaces with the glass melt and the plenum space of the melter,
the model provides sensitivity analysis of the response of the cold cap to the batch makeup and melter conditions. The model demonstrates that batch foaming has a decisive inuence on the rate of melting. Understanding the dynamics of the foam layer at the bottom of the cold cap and the heat transfer through it
appears crucial for a reliable prediction of the rate of melting as a function of the melter-feed makeup and
melter operation parameters. Although the study is focused on a batch for waste vitrication, the authors expect that the outcome will also be relevant for commercial glass melting.
2012 Elsevier B.V. All rights reserved.

1. Introduction
The cold cap, or batch blanket, is the layer of glass batch oating
on molten glass in an electrical glass-melting furnace (a melter). In
melters producing commercial glasses, the batch is typically spread
in a layer of uniform thickness on the whole top surface area of the
melt. In melters for nuclear waste glass, the melter feed, typically a
slurry containing 4060% water, is charged through one or more nozzles. The cold cap covers 9095% of the melt surface. Mathematical
models of melters have been well developed [15] in all aspects except for the batch melting, which has rarely been addressed in other
than a simplied manner [612].
This work presents an initial step toward the modeling of a cold
cap in a melter for high-level-waste glass while taking advantage of
the availability of data for the key properties and the reaction kinetics
for a high-alumina melter feed [1316] considered for the Waste
Treatment and Immobilization Plant, currently under construction
at the Hanford Site in Washington State, USA. The 1D model is
based on the ideas by Hrma [9] and Schill [10,11].
Though the 1D modeling greatly simplies mathematical treatment, the main challenge lies in the complexity of the conversion
process, which consists of a host of phenomena: water evaporation,
gas evolution, melting of salts, borate melt formation, reactions of
Corresponding author at: Pacic Northwest National Laboratory, Richland, WA
99352, USA.
E-mail address: pavel.hrma@pnl.gov (P. Hrma).
0022-3093/$ see front matter 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.jnoncrysol.2012.01.051

borate melt with molten salts and amorphous solids, precipitation

of intermediate crystalline phases, formation of a continuous glassforming melt, growth and collapse of primary foam, dissolution of residual solids, and the formation of secondary foam at the bottom of
the cold cap. In the present model, this complex situation is simplied
by assuming that the feed consists of two major phases: the condensed phase and the gas phase.
The authors expect that the 1D two-phase model presented in this
paper will be extended in the future to a two-dimensional (2D)
multi-phase model to be incorporated in a complete 3D model of
the waste glass melter.
2. Theory
The 1D model views the cold cap as a blanket of uniform thickness
that receives steady uniform heat uxes from both the molten glass
below and the plenum space above (Fig. 1). The mass balances of
the condensed phase and the gas phase are
db db vb
rb

dx
dt


dg d g vg

rg
dt
dx

where t is the time, the spatial density, v the velocity, r the mass
source associated with reactions, and the subscripts b and g denote

3560

P. Hrma et al. / Journal of Non-Crystalline Solids 358 (2012) 35593562

Table 1
Melter feed composition (in g) to make 1 kg of glass.

Fig. 1. Cold-cap structure.

the condensed phase and the gas phase, respectively. The mass uxes,
jb and jg, respectively, are jb = bvb (the minus sign in is used
because the condensed phase moves in the negative direction) and
jg = gvg. The energy balance for the condensed phase is:
b c b

dT b
dT
dq
b vb cb b b H s
dt
dx
dx

Compound

g/kg

Al(OH)3
SiO2
B(OH)3
NaOH
Li2CO3
Fe(OH)3
CaO
NaF
Bi(OH)3
Fe(H2PO2)3
Na2CrO4
NiCO3
Pb(NO3)2
Zr(OH)40.654H2O
NaNO3
Na2SO4
NaNO2
KNO3
Zn(NO3)24H2O
Na2C2O43H2O
Mg(OH)2
Total

367.49
305.05
269.83
97.14
88.30
73.82
60.79
14.78
12.80
12.42
11.13
6.36
6.08
5.11
4.93
3.55
3.37
3.04
2.67
1.76
1.69
1352.11

where T is the temperature, c is the heat capacity, q is the heat ux, H

is the internal heat source, and s is the heat transfer between phases.
The heat ux is subjected to Fourier's law, qb Eff dT
, where the efdx
fective value of heat conductivity, Eff, involves both conductive and
radiative modes of heat transfer in the feed.
The boundary conditions are given in the term of both uxes and
temperatures, i.e., qb(0) = QB, qb(h) = QT, T(0) = TB, and T(h) = TT,
where h is the cold cap thickness, TB and TT are the bottom and top
cold-cap temperatures, respectively, and QB and QT are the heat uxes
to cold-cap bottom and from cold-cap top, respectively.
To solve the energy balance equation, we used, for the sake of simplicity and comprehensibility, the nite difference method. For numerical simulations, experimental data were used together with
constitutive equations that were taken from the literature and modied for our problem as described in Section 3. All algorithms were
coded in Mathworks MATLAB 7.
3. Material properties
The melter feed composition is shown in Table 1.
Fig. 2 shows the degree of conversion, g, calculated as g(T)
= (m mF)/(mM mF), where m is the temperature-dependent feed
mass, mF is the melter-feed mass, and mM is the glass mass; the subscript g indicates that the degree of conversion is based on the release
of gas from the feed. Because the g(T) function changed little with
the rate of heating, we used the degree of conversion obtained by
thermogravimetric analysis at the heating rate 15 K min 1.
Following Schill [11], the heat capacity of the gas phase (cg) was
approximated by that of carbon dioxide: cg = 1003 + 0.21T
1.93 10 7T 2 (T 373 K), where cg is in J kg 1 K 1.
The melter feed density decreased as the temperature increased as
a result of mass loss and the nearly constant volume of the sample at
T b ~ 700 C, i.e., b = b0m/mF. Once the sample was shrinking between ~ 700800 C, the density increased to a maximum. Then,
above 800 C, the density decreased as the bubbly melt turned to
foam, reaching a minimum. The density was estimated from the expansion data reported in [13,14].
For the heat conductivity, we used the relationship reported by
Schill [17] for a high-level-waste melter feed: Eff = 0.06571 +
0.002114T (373 K b T b 1000 K), where is in W m 1 K 1. For
T > 1000 K, we used a linear interpolation between Eff at 1000 K

and Eff(1100 C) = 4.56 W m 1 K 1, the value for molten glass

(Schill [17]). To the best of our knowledge, the heat conductivity of
the foam layer at the cold cap bottom has not been determined experimentally. The heat conductivity of the foam layer was estimated to
be close to half of the heat conductivity of bubble-free melt.

4. Results
As Fig. 1 illustrates, a layer of boiling slurry rests on the top of the
cold cap, which consists of two layers: the main layer in which batch
reactions occur and batch gases are escaping through open pores, and
the bottom foam layer.
For the calculations, the cold-cap bottom temperature, TB = 1100 C,
was estimated as the temperature at which the motion of the condensed phase can no longer be considered one-dimensional and the circular convection of the melt takes over. The temperature at the interface
between the cold cap and the slurry layer was TT = 100 C, the temperature of the boiling slurry pool, ignoring the elevation of the boiling
point by dissolved salts.
Based on melter experiments reported by Matlack et al. [18], we chose
as the baseline case a slurry feed containing 52.2 mass% water and the
glass production rate (the rate of melting) jM =0.0141 kg m 2 s 1,
(1220 kg m 2 day 1).

Fig. 2. Degree of conversion with respect to gas phase versus temperature and heating
rate.

P. Hrma et al. / Journal of Non-Crystalline Solids 358 (2012) 35593562

Fig. 3. Cold-cap temperature prole for three heat uxes from plenum space.

The major part of the total heat ux to convert the slurry to molten glass at 1100 C comes from the cold-cap bottom (QB), of which
a part enters the boiling slurry (QT), while the remaining heat ows
to boiling slurry from the plenum space (QU). We have computed
the temperature distributions within the cold cap as a function of
two parameters: QF, the total heat ux to convert the slurry to
1100 C melt, and the fractional heat ux from above, QU/QS, where
QS is the heat ux to turn the slurry into dry feed. Fig. 3 compares
the temperature distributions within the cold cap for three values of
QU, while the total heat ux supplied to the slurry layer is constant.
The proles exhibit three intervals with distinct temperature gradients. Between 250 and 350 C, the temperature gradient is affected
by endothermic reactions. As Fig. 4 shows, most of the heat for melting is consumed in the upper part of the cold cap, 0.5 to 4 cm below
the top surface. Within 350 and 800 C, the presence of molten
phase results in an increased heat conductivity and a relatively mild
temperature gradient as compared to the temperature interval of
800 to 1100 C, where a steeper temperature gradient is caused by
the low Eff of the foam layer.
Fig. 5 shows the condensed phase velocity versus the vertical position within the cold cap, x, for QU/QS = 0.8. The higher velocity of
the condensed phase in the foam layer results from the high porosity,
while the lowest velocity occurs where the porosity is minimum; velocity changes little with x in the upper part of the cold cap.
Fig. 6 displays the thickness of the cold cap as a function of QU/QS
for three values of QF, i.e., different rates of melting, showing that a
higher melting rate makes the cold cap thinner.
5. Discussion

3561

Fig. 5. Velocity prole within the cold cap.

makeup and melter operation parameters. Foam continuously arises

from batch reactions and redox reactions within molten glass. Because foam has low heat conductivity, the foam layer hinders the
heat transfer to the cold cap and leads to a lower melting rate [15].
However, because the rate of melting affects the cold-cap thickness,
thus affecting the temperature gradient (i.e., the heating rate that
the feed experiences), the resulting shifts in the reaction peaks affect
the rate of foaming [19]. Hence, a self-regulating mechanism exists
that balances foaming, cold-cap thickness, and the heat uxes from
the top and bottom of the cold cap. Accordingly, batch foaming has
a decisive inuence on the rate of melting.
In the temperature interval of 700800 C at which the density is
minimum, a continuous layer of high-viscosity melt presents a barrier
through which neither bubbles can ascend nor low-viscosity molten
salts can drain. The foam continuously collapses vial bubble coalescence into cavities that are presumed to release the gas through the
free melt surface in the vent holes and around the cold-cap edges.
This process is clearly at least 2-dimensional.
An advanced cold-cap model, to be incorporated into the overall
melter model, should describe at least two 2D phenomena, the horizontal movement of bubbles and cavities toward cold cap edges and
a horizontal ow of migrating molten salts. Horizontal migration of
the salt melt into the vent holes and around the cold-cap edges can
disrupt the steady state, leading to cold-cap freezing because of depletion of uxes. Restoring the process requires melter idling, causing
a loss of efciency. The 2D model will determine the conditions for an
uninterrupted steady state.
Apart from foam dynamics and molten salt migration, the future
model will also address the fate of quartz particles (their dissolution

Understanding the dynamics of the foam layer at the bottom of

the cold cap and the heat transfer through it appears crucial for a reliable prediction of the rate of melting as a function of the melter-feed

Fig. 4. Effective heat capacity distributions within the cold cap.

Fig. 6. Cold-cap thickness versus upper and total heat ux; numbers represent QF in
kW m 2.

3562

P. Hrma et al. / Journal of Non-Crystalline Solids 358 (2012) 35593562

and clustering [16]), the fractions and sizes of solid particles (spinel
and quartz residues) leaving the cold cap into the melt pool, and
the thermal analysis-based reaction kinetics for gas evolution and enthalpy changes.
In a commercial electric glass melter, the top surface of the cold
cap is dry, and the top-surface boundary condition is not limited to
100 C as in a slurry-fed waste glass melter. Because the melting processes in both types of cold caps are similar in other aspects, the authors expect that the cold-cap model presented in this paper can be
adapted for commercial glass melting.
6. Conclusions
1. At a constant total heat ux to the cold cap, the cold-cap thickness
increases as the fraction of heat ux from above increases.
2. At a constant fractional heat ux from above, the cold-cap thickness decreases as the total heat ux to the cold cap increases.
3. As the cold-cap thickness changes, the reaction zone shifts, affecting the rate of foaming at the cold-cap bottom.
4. An advanced 2D cold-cap model should include behaviors of the
main liquid and solid phases within the cold cap, and the foam
layer under the cold cap. The model will be incorporated into the
complete model of the melter.
Acknowledgements
The authors are grateful to the U.S. Department of Energy Federal
Project Ofce Engineering Division for the Hanford Tank Waste Treatment and Immobilization Plant for nancial support. Richard Pokorny
is also pleased to acknowledge support from Czech Grant Agency
(GACR No. P106/11/1069). The authors would also like to thank
their colleagues at the Pacic Northwest National Laboratory DongSang Kim and Jaehun Chun for insightful discussions and David Pierce
for providing thermogravimetric analysis and differential scanning

calorimetry data. Pacic Northwest National Laboratory is operated

for the U.S. Department of Energy by Battelle under Contract DEAC05-76RL01830. In its nal stages, this research was supported by
WCU (World Class University) program through the National Research
Foundation of Korea funded by the Ministry of Education, Science and
Technology (R31-30005).

References
[1] W.S. Kuhn, in: D. Krause, H. Loch (Eds.), Mathematical Simulation in Glass Technology,
Springer, 2002.
[2] C. Moukarzel, W.S. Kuhn, Glas. Sci. Technol. 76 (2003) 8190.
[3] Z. Feng, D. Li, G. Qin, S. Liu, J. Am. Ceram. Soc. 91 (2008) 32293234.
[4] A. Abbassi, Kh. Khoshmanesh, Appl. Therm. Eng. 28 (2008) 450459.
[5] C.C. Yen, W.S. Hwang, Mater. Trans. 49 (2008) 766773.
[6] H. Mase, K. Oda, J. Non-Cryst. Solids 3839 (1980) 807812.
[7] R. Viskanta, X. Wu, J. Am. Ceram. Soc. 67 (1984) 376380.
[8] A. Ungan, R. Viskanta, AIChE Symposium Series, No. 236, Vol. 80, 1984,
pp. 446451.
[9] P. Hrma, Glastech. Ber. 55 (1982) 138150.
[10] P. Schill, Ceram. Silik. 26 (1982) 155163.
[11] P. Schill, Ceram. Silik. 26 (1982) 209222.
[12] P. Schill, J. Chmelar, J. Non-Cryst. Solids 345 & 346 (2004) 771776.
[13] P. Hrma, M.J. Schweiger, C.J. Humrickhouse, J.A. Moody, R.M. Tate, T.T. Rainsdon,
N.E. TeGrotenhuis, B.M. Arrigoni, J. Marcial, C.P. Rodriguez, B.H. Tincher, Ceram.
Silik. 54 (2010) 193211.
[14] S.H. Henager, P. Hrma, K.J. Swearingen, M.J. Schweiger, J. Marcial, N.E.
TeGrotenhuis, J. Non-Cryst. Solids 357 (2011) 829835.
[15] D.A. Pierce, P. Hrma, J. Marcial, B.J. Riley, M.J. Schweiger, Int. J. Appl. Glass Sci. (in
press).
[16] M.J. Schweiger, P. Hrma, C.J. Humrickhouse, J. Marcial, B.J. Riley, N.E.
TeGrotenhuis, J. Non-Cryst. Solids 356 (2010) 13591367.
[17] P. Schill, in: W. Lutze (Ed.), Modeling the behavior of noble metals during HLW
vitrication in the DM1200 melter, VSL-05R5740-1, Vitreous State Laboratory,
The Catholic University of America, Washington DC, 2005.
[18] K.S. Matlack, H. Gan, M. Chaudhuri, W. Kot, W. Gong, T. Bardakci, I.L. Pegg, J.
Innocent, DM100 and DM1200 melter testing with high waste loading glass formulations for Hanford high-aluminum HLW streams, VSL-10R1690-1, Vitreous
State Laboratory, The Catholic University of America, Washington, DC, 2010.
[19] F. Raether, M. Krauss, Glass Sci. Technol. 77 (2004) 118123.