J.

of Supercritical Fluids 70 (2012) 112–118

Contents lists available at SciVerse ScienceDirect

The Journal of Supercritical Fluids

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

Heat transfer of supercritical mixtures of water, ethanol and nitrogen in a bluff

body annular flow
P. Stathopoulos, K. Ninck, Ph. Rudolf von Rohr ∗
Institute of Process Engineering, Swiss Federal Institute of Technology (ETH), Zurich, Switzerland

a r t i c l e i n f o a b s t r a c t

Article history: Hydrothermal spallation drilling is a promising drilling technique that could prove economically advanta-
Received 19 April 2012 geous over rotary techniques for deep wells, where hydrothermal flames can provide the required heat to
Received in revised form 12 June 2012 spall the rock. Assisted ignition of hydrothermal flames must be understood prior to the field implemen-
Accepted 14 June 2012
tation of the technology. The convective heat transfer in a burner setup has been therefore investigated,
where the flow conditions are similar to those of a bluff body wake flow in annular geometry. Various
Keywords:
ternary mixtures of water, ethanol and nitrogen were used as model working fluids to simulate the com-
Bluff body
bustion conditions of water–ethanol mixtures with oxygen. Water ethanol mixtures were pre-heated
Supercritical heat transfer
Ternary mixture
between 350 ◦ C and 420 ◦ C, and nitrogen was pre-heated up to 400 ◦ C, while the working pressure was
Water–ethanol–nitrogen mixture set at 260 bar. The convective heat transfer coefficient from an electrically heated surface to the mixtures
Annular flow is presented.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction
The hot surface ignition of a hydrothermal flame is a topic rele-
vant to previous research projects [1], but has not been investigated
so far. The mechanisms of hot surface ignition [2] indicate that
ignition takes place locally at the points where the minimum heat
transfer coefficient of the flow occurs. In order to implement exist-
ing models for the hot surface ignition, at least the values of the
mean heat transfer coefficient from the heated surface to the flow
have to be known, estimated or measured. The heating power of
the igniter, its materials and the electrical connections are directly
connected to the heat transfer characteristics of the flow in the
given geometry and conditions. This motivated us to carry out the
measurements of the heat transfer coefficient for a model mixture
(water–ethanol–nitrogen), which will be the initial input for the
ignition experiments and the respective model of a combustible
mixture used in our lab (water–ethanol–oxygen).
The convective heat transfer coefficient of supercritical fluids
has been the topic of numerous experimental investigations, a
review of which can be found in the works of Pioro et al. [3,4].
In most cases experiments in simple tubular and annular geome-
tries were performed with the focus on pure fluids mainly water
and carbon dioxide. These investigations may be of use for nuclear
engineering, but offer little help, once a combustible mixture flows
over a heated surface in a diffusion flame combustion chamber.
The geometry in this case leads to a flow similar to the one of a
∗ Corresponding author. Tel.: +41 446322488; fax: +41 446321325.
E-mail address: pstathop@ethz.ch (Ph.R. von Rohr).
bluff body, to promote mixing of the reactants and thus it is not
possible to use the correlations acquired from flows in tubular
geometries.
Few publications can be found in the open literature on the heat
transfer from mixtures, the work of Rogak [5], being the most rele-
vant for the measurements of the present work. His measurements
with mixtures of supercritical water (SCW) and oxygen have shown
some interesting trends of the heat transfer coefficient and its cor-
responding maximum values. The shift of the bulk temperatures,
where these maximum values were observed and the lower abso-
lute values of the coefficients, attributed mainly to the lower cp
values of the resulting mixtures, were some of the most important
findings of this work.
According to the works of Bazargan and Mohseni [6] and Hiroaki
et al. [7], the position of the pseudo-critical point of a mixture
in the thermal boundary layer of its flow is responsible for the
behavior of its heat transfer coefficient. Although the pressure and
temperature values of the pseudo-critical points are known for
all the constituents of the ternary mixtures investigated, they are
not available for the mixtures themselves. Only estimations of the
expected values can be made based on the data of the critical points
for water–ethanol and water–nitrogen mixtures presented form
Abdurashidova et al. [8] and Japas and Franck [9] respectively. Pur-
pose of the current work is firstly to investigate the heat transfer
coefficient of a ternary mixture relevant for technical application in
supercritical water, and secondly to give the necessary data for the
design optimization of an ignition setup for hydrothermal flames.
The aimed heat transfer coefficient measurements are crucial for
the dimensioning of a hot wire igniter, its electrical power and its
geometry.

0896-8446/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.supflu.2012.06.006
 P. Stathopoulos et al. / J. of Supercritical Fluids 70 (2012) 112–118
Nomenclature
Latin letters
A Surface (m2 )
b0 andb1 Regression parameters for the resistance-
temperature function
d Diameter (m)
V Voltage (V)
mf Fuel mass flow (kg/h)
mN2 Nitrogen mass flow (Nl/min)
q̇ Heat flux (W/m2 )
R Electrical resistance ()
T∞ Fluid bulk temperature (◦ C)
Tw Igniter surface temperature (◦ C)
V̇ Volumetric flow rate (m3 /s)
Subscripts and abbreviations
AWG American wire gauge
SCW Supercritical water
2. Materials and methods
2.1. Measurement concept
The measurement of the convective heat transfer coefficient of
a flow, is based on the relation shown in Eq. (1), defining the con-
vective heat transfer coefficient. The exchanged heat flux between a
heated surface and a fluid (q̇), the bulk temperature (T∞ ) of the fluid
and the surface temperature (Tw ), have to be known or measured.
q̇ = h(Tw − T∞ )
Due to the difficulty of the measurements in a high pressure–high
temperature environment and the lack of space in the high pres-
sure setup in our lab, a simple experimental setup had to provide as
much data as possible, with as less measuring parameters as pos-
sible. A K-type thermocouple with a 3 mm diameter was used for
the bulk temperature measurement, while the heat flux was cal-
culated from Eq. (2) based on the voltage and current values fed to
the heated surface, and the surface value itself.
VI
q̇ =
A
The challenging direct measurement of the surface temperature
was resolved with the implementation of a ceramic igniter made
of silicon nitride, which had a linear and favorable temperature
dependence of its resistance. By calculating the resistance through
measurement of the voltage and current values fed to the igniter,
the average temperature of its heated length could be measured.
As a result, the available data for the heat transfer conditions
could be acquired through the voltage and current measurements,
supplemented from the measurements of a K-type thermocouple.
2.2. Heated surface characteristics – Igniter
The igniter used as a heated surface is a cylindrical body con-
sisting primarily of hot pressed silicon nitride (HPSN). The electrical
resistance is implemented in the body by a sintering procedure and
it consists of approximately 80 vol.%. silicon nitride while the rest
is additives, MoSi2 and TaN. The outer surface of the igniter is elec-
trically insulating, its total length is 90 mm, while its heated zone is
40 mm long and has a 4 mm diameter. The dependence of its resis-
tance from its temperature was calibrated in a high temperature
oven up to temperatures of 520 ◦ C. The calibration of the igniter at
higher temperatures was not possible because the material used for
soldering its electrical contacts melted above 550 ◦ C. Nevertheless,
113
80
78
76
Resistance [Ohm]
74
72
70
68
66
64
62
60
300 320 340 360 380 400 420 440 460 480 500
Temperature [°C]
Fig. 1. Igniter calibration curve (R = b0 + b1 Tw ). Confidence interval ±0.5%.
calibration data provided from the construction company [10] for
a reference igniter show a linear dependency up to temperatures
of 1000 ◦ C. A calibration line is presented in Fig. 1, together with
its confidence intervals, computed according to the the orthogonal
regression method [11].
The temperature in the calibration experiments was measured
with two K-type thermocouples positioned near the igniter and
its resistance was measured with a digital multi-meter. The accu-
racy of the thermocouple measurement was ±1.5 ◦ C, and of the
multi-meter 0.8% of the measured value. The error of the surface
temperature measurement through the resistance measurement
(inverse regression) was calculated from Eq. (A.2).
The resistance-temperature correlation defined from the cali-
bration experiments gives the values of the temperature in the core
(1) of the igniter, which has a diameter approximately 3 mm. This value
will not be the same with the surface temperature of the igniter,
when it operates in a strongly convective environment. The ther-
mal resistance of the electrically insulating layer of HSPN must be
also taken into account in this case. In our experiments the thermal
conductivity of this layer is estimated from the values reported in
the work of Watari et al. [12] at the same temperature as the core.
Accordingly, the value of the surface temperature of the igniter is
corrected from simple 1-D conduction in a cylinder.
2.3. Experimental setup
(2)
The experiments were conducted in the new hydrothermal spal-
lation drilling plant of our lab. The plant is using water–ethanol
mixtures of various compositions as a fuel and pure oxygen as an
oxidation medium and is capable of reaching fuel power of 120 kW.
Fuel and oxygen can be preheated at temperatures 420 ◦ C and
400 ◦ C respectively prior to their injection in the pressure vessel
of the plant. Two high pressure plunger pumps provide the cool-
ing water, a membrane pump feeds the fuel and an air driven gas
booster is used to compress the gas. Core of the plant is its high pres-
sure vessel, where all experiments are carried out. It has a volume
of approximately 5.8 l, and it is designed to withstand pressures of
600 bar at wall temperatures of 500 ◦ C. Its inner diameter is 14 cm
and the length of its inner space is 40 cm. Optical access to the ves-
sel is provided by two small sapphire windows on its head. Probe
access and positioning, while under pressure, is achieved with two
individual positioning devices similar to the ones presented by
Prikopsky [1], the positioning accuracy of which is 0.2 mm.
A drawing of the upper part of the vessel is presented in Fig. 2.
The inner space of the pressure vessel is divided by a stainless
steel tube in the outer cooling mantle and the inner space where
the hydrothermal flame will be ignited. Pressure balance between
the two volumes is provided form 12 holes (diameter 1 mm) posi-
tioned at the bottom of the dividing steel tube. The diameter of
114 P. Stathopoulos et al. / J. of Supercritical Fluids 70 (2012) 112–118
Fig. 2. Technical drawing of the upper part of the used pressure vessel. (For inter-
pretation of the references to color in the text, the reader is referred to the web
version of the article.)
the inner space is 10 cm, and its cooling water is fed through two
holes opposite to one another to ensure homogeneous cooling. The
pre-heated fuel mixture is fed axially through a specially designed
burner nozzle, a drawing of which is shown in Fig. 3. Oxygen is
fed from a radial hole (not shown in Fig. 2), and then injected into
the burning chamber parallel to the axis of the vessel through the
annulus between the burner nozzle and the reactor body.
The heat transfer measurements in the present work concen-
trate on the heat transfer conditions in the burning camber of the
vessel (highlighted red in Fig. 2). A detailed drawing of the burning
chamber where the control volume is located is presented in Fig. 5.
In Fig. 4 a graphical representation of the control volume with the
operating conditions of the experiments is presented.
The mixing of the fuel and the gas streams takes place in the vol-
ume where we are measuring the heat transfer coefficient. The fuel
mixture is injected through the burner nozzle in the parallel flow
of oxidant with an angle of 30◦ to its direction. This leads to a flow
similar to a bluff-body flow with very intense recirculation in the
whole length of the heated surface. Computational fluid dynamics
modeling showed that due to the angle between the two flows, very
intense mixing results and a homogeneous mixture is produced in
the cases where one phase flow is present.
The electrical connection wires are fed to the reactor through
a positioning device similar to the one presented in the work of
Prikopsky [1]. The length of the feeding line is divided in a high tem-
perature 250–420 ◦ C and a low temperature region (40–250 ◦ C).
For the low temperature region kapton-insulated copper wires
(AWG20) are used. Their end is welded to wires made of Alumel
(95% Ni, 2% Mg, 2% Al, 1% Si) (diameter 0.8 mm), which are directed
Fig. 3. Technical drawing of the burner tube used throughout the experiments.
Fig. 4. Burning chamber with operational conditions.
through a ceramic capillary made of aluminium oxide. The sealing
of the wires is realized with special pressure glands, while a very
thin-walled aluminum oxide tube (0.24 mm) was used for the elec-
trical insulation of the contacts of the igniter from the walls of the
burner nozzle.
The resistance of the feeding line was 0.95  at 20 ◦ C and 1.4  at
300 ◦ C, when it operated in the system. This change was taken into
consideration in the interpretation of the data, although it was very
small relatively to the absolute value of the igniter resistance. The
resulting voltage feeding system allows for the use of a maximum
current value of 9 A and a maximum voltage value of 230 V.
2.4. Measuring procedure
Two measurement campaigns have been performed, one only
with the fuel stream (ethanol–water mixtures) and one with both
the fuel and nitrogen streams. The first has been carried out as test
campaign, while the aim of the second was to investigate the effect
of ethanol and nitrogen addition on the heat transfer coefficient.
All the measurements were performed at a constant pressure of
260 ± 2 bar and the fluids were preheated at temperatures between
350 ◦ C and 420 ◦ C. Four fuel-stream compositions were used having
0 wt.%, 10 wt.%, 20 wt.% and 30 wt.% ethanol, and various mass-flow
values for nitrogen. The operational conditions of the measure-
ments are presented on Table 1.
The cooling water for the main volume of the vessel was not pre-
heated and its mass flow was constant throughout the experiments
and equal to 150 kg/h. The igniter projected 45 mm in the control
volume (see Fig. 5) and 40 mm of them were heated electrically.
 P. Stathopoulos et al. / J. of Supercritical Fluids 70 (2012) 112–118 115

Table 1
Operational conditions for the heat transfer experiments with both fluid streams.

Fuel stream Fuel mass flow (mf ) Nitrogen mass flow

composition (wt.% (kg/h) (mN2 ) (Nl/min)
EtOH)

10
20
0 130
30
40
10
20
10 130
30
40
10
130
20
20
30 180
35 220
10
130
15
30
20 180
25 220 Fig. 6. Voltage ramp example for a measurement only with the fuel stream.

1. Once the predefined fluid temperatures were reached, voltage

Furthermore all the measurements have been conducted for two
was fed to the igniter.
heat flux values: 0.5 and 0.7 (±0.05) MW/m2 , to investigate its effect
2. A defined voltage ramp was followed for each measurement
on the heat transfer coefficient.
leading to the first heat flux value (see Fig. 6).
The procedure for each measurement consisted of the following
3. The respective voltage value was kept constant for 60 s and it
steps:
was simultaneously recorded with the current value and the bulk
temperature.
4. In the next step, the voltage was adjusted to reach the second
heat flux value (see Fig. 6).
5. After the 60 s of the second heat flux measurement a ramp was
followed for the reduction of the voltage on the igniter.
6. Once the voltage was switched off, the bulk temperature was
recorded for 60 s.

The reported bulk temperature is the mean value of the temper-

ature measured during the last 60 s of the procedure and the value
measured during the heating with each respective heat flux.
Radiation was taken into account in all the measurements,
because in many cases the surface temperature of the igniter
reached values up to 900 ◦ C. The correction was performed based
on the case of the radiation heat exchange of two concentric cylin-
ders, the inner being at the igniter surface temperature and the
outer at the cooling water temperature.

3. Results and discussion

3.1. Igniter oxidation and measurements correction

After the first experiments at temperatures 360 ◦ C only with

the fuel stream, a change has been observed on the igniter sur-
face, which was identified as a result of silicon nitride oxidation
from desalinated water at elevated temperatures and pressures.
The reaction reported in various publications [13–15], for the oxi-
dation phenomenon follows the equation below.

Si3 N4 + 6H2 O  3SiO2 + 4NH3

Fig. 5. Burning chamber with dimensions given in mm.
It was shown that the reaction rate follows Arrhenius kinetics with
different values of the kinetic constants depending on the type and
humidity of the environment.
During the experiments only with the fuel stream, the silicon
oxide layer produced from the oxidation of the igniter, affected
the heat transfer coefficient measurement, by introducing an addi-
tional thermal resistance layer between the surface and the fluid.
In contrast to that, the growth of the layer during the ternary
mixture experiments was slower and a correction of the heat
transfer coefficient measurements based on the observation of
116 P. Stathopoulos et al. / J. of Supercritical Fluids 70 (2012) 112–118

Fig. 7. Reference measurements for the oxidation of the igniter from water. Condi-
tions: mf : 40 kg/h, mN2 : 130 kg/h, T∞ : 326 ◦ C, q̇: 0.5 MW/m2 .
Fig. 8. Influence of the used heat flux for a water–nitrogen mixture (mf = 40 kg/h,
mN2 = 130 Nl/min).

its thickness development was possible. This layer development

water–nitrogen case a temperature difference of approximately
was studied by a series of reference experiments, at pre-heating
40 ◦ C is enough for the values to coincide). This is in accordance
temperature 385 ◦ C. During these experiments the heat transfer
with the data presented from Yamagata et al. [16].
coefficient of a fresh igniter was measured at the beginning and
Furthermore, the measurement error is considerably higher
the same measurement was performed every fifteen minutes for
for the measurements with the lower heat flux. According to Eq.
6 h. The difference of the measured heat transfer coefficient from
(A.3) (presented in the appendix), the parts of the right hand side
the initial value is attributed to the additional thermal resistance
are inversely proportional either to the temperature difference
introduced from the growth of the SiO2 layer. Consequently, the
between the surface and the bulk or to is squared value. The use
thickness of the layer could be calculated from a simple one-
of lower heat flux values leads to a lower temperature difference,
dimensional heat transfer calculation in cylindrical coordinates and
which in turn leads to higher measurement errors.
its growth was observed and presented in Fig. 7. The change is small
Another important observation is that the absolute values of the
(maximum 9%) but nevertheless important for the measurements
heat transfer coefficient reported here are 30–70% lower in com-
and a function of the layer thickness development was fitted to the
parison with the values reported from Yamagata et al. [16] and
measured values. Based on the operational time of each igniter a
Rogak and Faraji [5]. Concerning the work of Yamagata, the use
respective layer thickness was calculated and used for the correc-
of water mixtures with ethanol and nitrogen, must be responsible
tion of the measurements. Each igniter was used for one day (5–6
for the difference of the reported values, since the implemented
measuring hours) and then replaced. In addition to the aforemen-
heat flux values are of the same order of magnitude. For the case
tioned measurements, reference measurements were performed
of the measurements of Rogak, the heat flux values used here are
during each campaign, to observe the layer development.
two to five times higher from the ones implemented in his work,
which could lead to the observed differences. Moreover, we must
3.2. Heat transfer coefficient results stress that the values presented here are average values over the
whole heated surface. Additionally the heated surface is situated in
The aim of the experiments is to investigate the influence of the wake flow of a bluff-body, which is highly turbulent, non uni-
the ethanol and nitrogen addition to the heat transfer characteris- form and recirculating. These facts hinder the analysis of the local
tics both in a qualitative and a quantitative way. Nevertheless, the heat transfer coefficient and could partially cancel out high (or low)
used system, provided a limited number of distinct values for the temperature regions of the heated surface.
selected bulk temperature. The lowest difference between these
values (5 ◦ C) achieved within reasonable experimental time, was
implemented for the most interesting regions of the heat transfer
coefficient.

3.2.1. Influence of heat flux on the convective heat transfer

coefficient
It has been reported in various research works [3,4], that the
heat transfer coefficient of fluids near their critical point depends
on the heat flux used for the measurements. We performed all
the measurements for two heat flux values 0.5 and 0.7 MW/m2 ,
to investigate this dependency for the mixtures in consideration.
The results for a water–nitrogen and a 20% water–ethanol mixture
with nitrogen are presented in Figs. 8 and 9 respectively.
For both cases a heat flux increase leads to a reduction of the
convective heat transfer coefficient around the point where its
maximum value occurs. The more the bulk temperature diverges
from this point, the more the values of the heat transfer coeffi-
cient for the two heat fluxes approach until they become practically Fig. 9. Influence of the used heat flux for a water–ethanol (30%)–nitrogen mixture
the same for high and low values of the bulk temperature (for the (mf = 25 kg/h, mN2 = 220 Nl/min).
 P. Stathopoulos et al. / J. of Supercritical Fluids 70 (2012) 112–118 117

Fig. 10. Convective heat transfer coefficient values for water–ethanol–nitrogen Fig. 11. Convective heat transfer coefficient values for water–nitrogen mixtures
mixtures with different fuel stream compositions (0%, 10%, 20%). Conditions: mf : with different mass flow ratios. Conditions: mN2 : 130 Nl/min, mf : 20, 30, 40 kg/h, q̇:
20 kg/h, mN2 : 130 Nl/min, q̇: 0.7 MW/m2 . 0.7 MW/m2 .

3.2.2. Influence of ethanol addition on the convective heat

transfer coefficient
To observe the influence of ethanol addition on the system, the
flow rates of the fuel stream and the nitrogen stream were kept con-
stant, and the convective heat transfer coefficient was measured
for three different mass concentrations of the fuel stream (0%, 10%,
20%). The results of these measurements are presented in Fig. 10, for
a mass flow of the fuel stream and nitrogen 20 kg/h and 130 Nl/min
respectively.
The heat transfer coefficient exhibits the same trend for all the
compositions. With increasing temperature its value increases until
it reaches a maximum and then it decreases again to values similar
to the ones at lower temperatures. The addition of ethanol to the
fuel stream introduces a small change to the maximum values of
the heat transfer coefficient, which lies within the error tolerance
of our measurements; it is nevertheless observable.
More interesting is the influence of ethanol addition on the bulk
Fig. 12. Convective heat transfer coefficient values for water–ethanol
fluid temperature for which this maximum value of the heat trans- (10 wt.%)–nitrogen mixtures with different mass flow ratios. Conditions:
fer coefficient occurs. It is obvious that this temperature drops the mN2 : 130 Nl/min, mf : 20, 30, 40 kg/h, q̇: 0.7 MW/m2 .
more ethanol is added to the system, and this occurs in a non-linear
manner. which the maximum heat transfer coefficient occurs. It is obvi-
This temperature value is the pseudo-critical temperature of the ous that this temperature is shifted to lower values for a higher
ternary mixture under investigation, for the pressure used [4,5]. concentration of nitrogen in the ternary system.
The change of the pseudo-critical temperature in respect with the The results are in agreement with the results of Rogak and Faraji
addition of ethanol to the mixture is qualitative and quantitative [5] for the water–oxygen mixtures. A very important difference is
similar to the one observed from Abduraslidova et al. [8] for the that the temperatures where the heat transfer coefficient peaks
critical temperature of simple water–ethanol mixtures. occur at lower bulk temperature values. Some reasons for this effect
are the increased nitrogen concentration in the present system, and
3.2.3. Influence of nitrogen addition on the convective heat the substitution of oxygen with nitrogen in our measurements.
transfer coefficient
The nitrogen influence to the heat transfer conditions of water 4. Conclusions
and a water–ethanol mixture (10 wt.%) could be investigated in the
current setup by keeping the mass flow of nitrogen constant at The convective heat transfer coefficient of ternary mixtures of
130 Nl/min and choosing three values for the mass flow rate of the water–ethanol and nitrogen was measured in a hydrothermal diffu-
fuel stream (20 kg/h, 30 kg/h and 40 kg/h). sion flame combustion chamber. The mixture was used as a model
This way the mass flux of the ternary mixture was not kept for future ignition experiments.
constant and only the influence of nitrogen addition on the pseudo- The influence of the heat flux and the ethanol and nitrogen con-
critical temperature of the mixture could be investigated. Its centrations on the heat transfer has been clearly demonstrated. An
influence on the absolute values of the heat transfer coefficient increase in the used heat flux led to lower values of the heat transfer
can not be observed, because its values are a function of the mass coefficient around the pseudo-critical point of the ternary mix-
flux. The results of the measurements with water and nitrogen are ture, as it was expected from the existing literature. Furthermore
presented in Fig. 11, and the results with the ternary mixture are an increase of the ethanol concentration in the ternary mixture
presented in Fig. 12. In both cases the same influence with that introduced a reduction of the pseudo-critical temperature of the
of ethanol addition is observed on the bulk fluid temperature for mixture. Additionally a decrease of the absolute values of the heat
118 P. Stathopoulos et al. / J. of Supercritical Fluids 70 (2012) 112–118

transfer coefficient was observed. For the case of nitrogen addition Appendix B. Supplementary Data
to the mixture, only its influence on the pseudo-critical tempera-
ture could be observed. An increase of this concentration led to a Supplementary data associated with this article can be
decrease of the aforementioned temperature. found, in the online version, at http://dx.doi.org/10.1016/j.
A significant problem during the measurements was the oxi- supflu.2012.06.006.
dation of the ceramic heated surface used and the additional heat
transfer resistance introduced from a silicon oxide layer produced
from this phenomenon. The phenomenon was not so intense during References
the measurements with the ternary mixture and could be neglected
[1] K. Prikopsky, Characterization of continuous diffusion flames in supercritical
by replacing the heated surface after 8–10 h of operation. In future water, Ph.D. thesis, ETH Zurich, 2007.
experiments this ignition source will be replaced with Ni/Cr 60/15 [2] D.A. Frank-Kamenetskii, Diffusion and Heat Transfer in Chemical Kinetics, N.Y.
heating coils and the whole setup of the system will be changed to Plenum Press, 1969.
[3] I.L. Pioro, R.B. Duffey, Experimental heat transfer in supercritical water flow-
work with DC power. ing inside channels (survey), Nuclear Engineering and Design 235 (2005)
2407–2430.
Appendix A. Error calculations [4] I.L. Pioro, H.F. Khartabil, R.B. Duffey, Heat transfer to supercritical fluids
flowing in channels empirical correlations (survey), Nuclear Engineer-
ing and Design 230 (2004) 69–91, 11th International Conference on
The uncertainty of the measurements was calculated by using Nuclear Energy.
simple Gaussian error propagation for each product variable. The [5] S.N. Rogak, D. Faraji, Heat transfer to water–oxygen mixtures at supercritical
pressure, Journal of Heat Transfer 126 (2004) 419–424.
directly measured variables were the voltage and the current on the [6] M. Bazargan, M. Mohseni, The significance of the buffer zone of boundary layer
igniter, and the bulk temperature. The accuracy of the voltage and on convective heat transfer to a vertical turbulent flow of a supercritical fluid,
current measurements was 0.75 V and 50 mA respectively, while The Journal of Supercritical Fluids 51 (2009) 221–229.
[7] T. Hiroaki, N. Niichi, H. Masaru, T. Ayao, Forced convection heat transfer to
the digitalization was done with a 16-bit card, and the respective fluid near critical point flowing in circular tube, International Journal of Heat
cut-off error has been neglected. Hence the resistance measure- and Mass Transfer 14 (1971) 739–750.
ment error was calculated from Eq. (A.1). [8] A.A. Abduraslidova, A.R. Bazaev, E.A. Bazaev, I.M. Abdulagatov, The thermal
 properties of water–ethanol system in the near-critical and supercritical states,
 V 2  VI 2 High Temperature 45 (2007) 178–186, Times Cited: 0.
R = + − (A.1) [9] M. Japas, E. Franck, High pressure phase equilibria and pvt-data of the
I I2 water–nitrogen system to 673 K and 250 mpa, Berichte der Bundesgesellschaft
fuer Physikalische Chemie 89 (1985) 793–800.
The error of the regression variables (b0 and b1 ) was calcu- [10] J.P. Rue, B.R. Bach, Personal communication, 2011 (Data on the used
lated from the error-in-variables orthogonal regression procedure igniters).
[11] J. Gillard, T. Iles, Methods of fitting straight lines where both variables
presented in the work of Gillard and Iles [11]. The error of the
are subject to measurement error, Current Clinical Pharmacology 4 (2009)
temperature measurements was calculated from Eq. (A.2). 164–171.
   2 [12] K. Watari, M.E. Brito, M. Toriyama, K. Ishizaki, S. Cao, K. Mori, Thermal conduc-
 R 2  b 2 b0 − R
tivity of y2o3-doped si3n4 ceramic at 4 to 1000 K, Journal of Materials Science
0
Tw = + − + b1 (A.2) Letters 18 (1999) 865–867 http://dx.doi.org/10.1023/A:1006696126661.
b1 b1 b21 [13] Shigeyuki, Somiya, Hydrothermal corrosion of nitride and carbide of silicon,
Materials Chemistry and Physics 67 (2001) 157–164.
The value of the heat transfer coefficient was calculated from [14] E. Proverbio, F. Carassiti, Low-temperature oxidation of silicon nitride by water
in supercritical condition, Journal of the European Ceramic Society 16 (1996)
Eq. (A.3). 1121–1126.
 2  2  2 [15] S. Singhal, Effect of water vapor on the oxidation of hot-pressed silicon
I V VI
h2 = V + I + − A nitride and silicon carbide, Journal of the American Ceramic Society 59 (1976)
A(Tw − T∞ ) A(Tw − T∞ ) A2 (Tw− T∞ ) 81–82.
 2  2 (A.3) [16] K. Yamagata, K. Nishikawa, S. Hasegawa, T. Fujii, S. Yoshida, Forced convective
VI VI heat transfer to supercritical water flowing in tubes, International Journal of
+ − Tw + T∞
A(Tw − T∞ )
2
A(Tw − T∞ )
2 Heat and Mass Transfer 15 (1972) 2575–2593.