You are on page 1of 13

Available online at www.sciencedirect.

com

Solar Energy 86 (2012) 2293–2305


www.elsevier.com/locate/solener

Liquid sodium versus Hitec as a heat transfer fluid in solar


thermal central receiver systems
Nicholas Boerema a,⇑, Graham Morrison a, Robert Taylor a, Gary Rosengarten a,b
a
School of Mechanical and Manufacturing Engineering, University of New South Wales, Sydney, NSW 2052, Australia
b
School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University, Melbourne, Vic. 3001, Australia

Received 15 July 2011; received in revised form 26 March 2012; accepted 1 May 2012
Available online 28 May 2012

Communicated by: Associate Editor Yogi Goswami

Abstract

Selection of an appropriate HTF is important for minimising the cost of the solar receiver, thermal storage and heat exchangers, and
for achieving high receiver and cycle efficiencies. Current molten salt HTFs have high melting points (142–240 °C) and degrade above
600 °C. Sodium’s low melting point (97.7 °C) and high boiling point (873 °C) allow for a much larger range of operational temperatures.
Most importantly, the high temperatures of sodium allow the use of advanced cycles (e.g. combined Brayton/Rankine cycles). In this
study, a comparison between the thermophysical properties of two heat transfer fluids (HTFs), Hitec (a ternary molten salt 53%
KNO3 + 40% NaNO2 + 7% NaNO3) and liquid sodium (Na), has been carried out to determine their suitability for use in high-temper-
ature concentrated solar thermal central-receiver systems for power generation. To do this, a simple receiver model was developed to
determine the influences of the fluids’ characteristics on receiver design and efficiency. While liquid sodium shows potential for solar ther-
mal power systems due to its wide range of operation temperatures, it also has two other important differences – a high heat transfer
coefficient (an order of magnitude greater than Hitec) and a low heat capacity (30–50% lower than Hitec salt). These issues are studied
in depth in this model. Overall, we found that liquid sodium is potentially a very attractive alternative to molten salts in next generation
solar thermal power generation if its limitations can be overcome.
Ó 2012 Elsevier Ltd. All rights reserved.

Keywords: Sodium; Hitec; Central receiver systems; Solar thermal; Heat transfer fluid

1. Introduction energy is then transferred from the receiver through tube


walls to heat a heat transfer fluid (HTF). The HTF is then
To date the majority of concentrated solar power passed directly through a heat exchanger to generate steam
deployment has used parabolic trough technology. How- for a Rankin cycle. Thermal storage can also be incorpo-
ever, central receiver systems (CRSs) have the potential rated into the system, either combined into the solar heat-
to achieve the lowest cost in the long term because of their ing loop or through the use of a third loop and the use of
ability to reach higher temperatures and utilise more effi- an additional heat exchanger (Ortega et al., 2008).
cient thermodynamic cycles (Hinkley et al., 2011). The maximum temperature of the working cycle is the
In solar thermal central receiver systems, sunlight is predominant indicator for the theoretical and actual power
concentrated through the use of a field of heliostats onto that can be extracted from the thermal energy, with higher
a central receiver. For a conventional system the incident temperatures leading to higher efficiencies. Whilst most
CRS deployments to date have used steam temperatures
of approximately 380 °C, the Gemasolar (previously known
⇑ Corresponding author. Tel.: +61 428828015. as Solar Tres) central receiver project recently began opera-
E-mail address: nicholas.boerema@student.unsw.edu.au (N. Boerema). tion and is designed to utilise inlet and outlet solar loop

0038-092X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.solener.2012.05.001
2294 N. Boerema et al. / Solar Energy 86 (2012) 2293–2305

Nomenclature

Symbols CR concentration ratio


V_ volumetric flow rate (m3/s) k thermal conductivity
m_ mass flow rate (kg/s) h heat transfer coefficient
Nu Nusselt number Q power (W)
Cp specific heat capacity at constant pressure (J/ A area
kg °C)
Re Reynold’s number Greek symbols
Pr Prandtl number Wp heat capacity (J/m3°C)
Prs Prandtl number at the pipe surface e pipe roughness height
f friction factor q density (kg/m3)
D pipe diameter l viscosity
V average fluid velocity r Stefan-Boltzmann constant
DPL pressure drop per metre (Pa/m)
T temperature (K) Subscripts
Ts pipe surface temperature (K) S sodium
T 11 fluid temperature away from the wall H Hitec
T1 temperature of the surrounds f fluid
Ta ambient temperature rad radiation
r1 pipe inside radius conv convection
r2 pipe outside radius abs absorbed
Ipipe pipe surface radiation intensity (W/m2) seg segment
Iap radiation intensity over the receiver area i section i

temperatures of about 285 °C and 565 °C respectively, using  Low viscosity to reduce pumping losses.
a molten salt HTF. To increase turbine efficiencies and to  Stability at high temperatures and a high boiling point
reduce storage volumes the aim is to further increase the to allow for no or low overpressure requirements and
maximum operating temperature. Whilst the current mol- to prohibit HTF degradation.
ten salts will begin to break down as temperatures are  Low melting point to minimize the use of freeze pro-
increased, sodium, with its high vaporisation temperature tection equipment.
of 873 °C at ambient pressure, will remain a suitable choice.  Low cost per kilogram.
Whilst costs can vary dramatically from project to pro-
ject, estimations have been made to quantify the main However there is no ideal fluid and so compromises
expenses associated with a CRS plant. Hinkley et al. must be made. Furthermore other issues must be consid-
(2011) show that the main costs for CRSs are due to financ- ered, such as:
ing and insurance (71%) and the main constituent of these
costs is the solar field at about 39%. Thus, this is an area  Security of supply (availability).
for cost reduction that could result in large cost savings  Fluid and plant life expectancy.
for the deployment of solar thermal electricity. Increasing  Environmental issues.
the efficiency of the working cycle will reduce the size of  Operation and maintenance issues (safety issues).
the heliostat field required to receive the same energy. This  Additional component requirements.
can be achieved through increasing the working fluid  Compatibility with other materials.
temperature, as detailed, and through increasing receiver  Corrosion or coking characteristics
efficiency. As the working temperature and the receiver
efficiency are both very dependent on the HTF used in The main HTF used today in CRSs is Hitec (53%
the solar loop, finding the optimum HTF is important KNO3 + 40% NaNO2 + 7% NaNO3). The issue with Hitec
for achieving this goal (Bignon, 1980). is that it has a melting point of 142 °C and is only usable up
The properties to be required for a HTF are: to about 535 °C. Another HTF used is the binary salt 60%
NaNO3 and 40% KNO3 commonly known as solar salt.
 Good heat exchange for cost efficient receivers and This salt has similar properties to Hitec but is compara-
heat exchangers. tively cheaper and is usable up to about 585 °C, however
 High heat capacity to reduce the storage volume it has a much higher melting point of 220 °C (Kearney et
required. al., 2003). Both of these nitrate salt mixtures become
N. Boerema et al. / Solar Energy 86 (2012) 2293–2305 2295

thermally and chemically unstable at temperatures above 1.1. The solar receiver
600 °C (Bradshaw and Siegel, 2008; CoastalChem, 2011).
In nitrate salts, the nitrogen and oxygen decompose and, The aim of the receiver is to accept the heat flux deliv-
depending on conditions, form the following gases: O2, ered by the heliostats and transmit it with minimal losses
N2, NO, NO2, N2O2, N2O3, N2O4, N2O5 (Stern, 1972). to the HTF. The intensity and flux map incident on the
Thus, using these HTFs the design difficulty is to achieve receiver changes with the elevation of the sun in a daily
a cost effective and high efficiency receiver design whilst and yearly cycle, and also more rapidly with cloud tran-
insuring that the HTF film temperature limits will not be sients, thus the flow rate of the HTF through the receiver
exceeded, leading to HTF degradation. needs to be controlled such that the desired outlet temper-
Sodium has a lower melting point of 97.7 °C and a high ature is achieved. This outlet temperature must be around
boiling point of 873 °C, giving it a large operating temper- 30 °C higher than the required turbine temperature in
ature range. The disadvantage of sodium is its high combus- order to allow for parasitic heat loss through the piping
tibility when in contact with water. Even if no air is present, and to maintain a suitable temperature difference across
a rapid and violent reaction will occur between sodium and the heat exchanger.
water, with both hydrogen and steam evolving. If air is pres- The main energy losses from the receiver are from reflec-
ent, an explosive mixture will be generated as the hydrogen tion, re-radiation and convection. Thus, to minimise heat
mixes with the air. However, this risk can be controlled. The losses the receiver area should be minimised and efficient
main experience on the safe use of sodium as a HTF has heat transfer to the HTF should be achieved in order to
come from the nuclear industry, where sodium has been minimise the receiver surface temperature. Minimising the
used extensively since the 1950s. One particular area where receiver area will create additional financial advantages
expertise has been developed is the sodium–water heat through reducing the wind loading of the receiver, thus fur-
exchanger. The heat exchanger presents a greater risk due ther reducing the material requirements of the receiver and
to the close proximity of sodium and water. The most likely support tower.
leak to occur would be due to a crack induced by thermal
strain cycling. This type of leak would be small initially 1.2. Heat transfer
and can be discovered at an early stage by using a sensor
to detect hydrogen in the sodium system (Fraas, 1989). The receiver efficiency is strongly dependent on the HTF
Sodium has also been used at the International Energy used and its ability for effective heat transfer from the recei-
Agency’s Small Solar Power Systems Project (IEA-SSPS), ver surface to the HTF.
a 0.5 MW central receiver project run during the 1980s. Two options for increasing the effectiveness of the heat
High intensity tests where undertaken, during which the transfer process are through using a HTF with better heat
system was operated with an intensity of over 2.5 MW/ transfer properties (such as one with better thermal con-
m2 whilst maintaining efficiency slightly above its design ductivity) and by increasing the heat transfer coefficient.
value of 90%. These values give example to what is achiev- Increasing the fluid mixing will increase the heat transfer
able by taking advantage of sodium’s beneficial heat trans- coefficient. Increased heat transfer can, in general, also be
fer characteristics (Schiel and Geyer, 1988). However, this achieved through using pipes with roughened or finned
plant also resulted in a sodium spray fire from a valve that inner surfaces. This, however, will also increase the pres-
was being repaired, which resulted in 14 tons of sodium sure loss across the receiver, which is often an important
being released over 30 min. Due to the nature of sodium design constraint and thus must also be considered.
fires (short flames, strong aerosol production) the propaga- Increasing the effectiveness of the heat transfer also reduces
tion of the fire was limited to natural convection so its the difference between the wall temperature and the fluid
spread was limited (Luster and Freudenstein, 1996). mean temperature which allows for lower wall/fluid skin
The use of liquid sodium with its added fire risks must temperatures. This is particularly important for molten
be justified. Of particular importance is its high reactivity salts as they are less stable at higher temperatures.
with water and high aerosol release when burnt. These
aerosols increase the difficulty of extinguishing the fire
where ventilation is restricted. However the low heat 2. Comparison of properties
release from a liquid sodium fire means that the risk of fire
propagation is less severe than other common combustible To analyse the benefit of using sodium as the HTF in
fuels. For example, the heat released from liquid sodium central receiver system, we must first obtain the thermo-
burning (per m2) is about 10 times less than that of a ker- physical properties of Hitec and sodium across their oper-
osene fire (Rodriquez and Sawada, 2006). Considerable ating temperature range. These properties have then been
work has been completed on understanding and modelling used to calculate volumetric specific heats, the heat transfer
the combustion characteristics of liquid sodium and meth- coefficient, the pressure drop per metre for various mass
ods for sodium fire extinguishing have been developed as a flow rates and the expected film temperature for various
result of liquid sodium’s use as a HTF in nuclear reactors surface temperatures. The subscripts S and H will be used
(Olivier et al., 2007). to refer to sodium and Hitec respectively.
2296 N. Boerema et al. / Solar Energy 86 (2012) 2293–2305

Properties for sodium have been taken from “Thermo- types, as well as the heat transferred. For this to be
dynamic and Transport Properties of Sodium Liquid achieved in a given receiver design the fluids would need
And Vapor” (Fink and Leibowitz, 1995) and are suitable to have different mass and volumetric flow rates. These
for temperatures 97.7–873 °C. Properties for Hitec have can be calculated as follows:
been taken from the Solar Advisor Model (NREL, 2009)
and are intended for temperatures 142–535 °C. For the Cp S
m_ H ¼ m_ S ð1Þ
dynamic viscosity an equation has been fitted to the curve Cp H
given by Bohlmann (1972), pp. 25, Fig. 6.
Fig. 1a shows that liquid sodium has a density of almost Wp S
V_ H ¼ V_ S ð2Þ
half that of Hitec. Fig. 1b shows that the specific heat Wp H
capacity of liquid sodium and Hitec are quite similar.
Fig. 1c shows that liquid sodium has a viscosity 30–5 times where m_ is the mass flow rate (kg/s), Cp is the specific heat
lower than Hitec and Fig. 1d shows the large difference in (J/kg °C), V_ is the volumetric flow rate (m3/s) and Wp is the
thermal conductivity between the two fluids, with Sodium heat capacity. Thus we can observe that the required mass
being 160–60 times more thermally conductive than Hitec. flow rate for Hitec is that of the sodium scaled by a factor
The heat capacity is calculated as the product of density of C P S =C P H . This ratio ranges from about unity to 1.35
and the specific heat. Hitec has a higher heat capacity than over the temperature range and thus the mass flow rate
liquid sodium which means that, for equivalent operating of Hitec will be greater than that using sodium (see ratio
temperatures, Hitec will require a lower thermal storage in Fig. 1b). The volumetric flow rate for Hitec is the sodium
volume to store the same amount energy. Thus the ratio volumetric flow rate scaled by a factor of WP S =Wp H . This
of volumetric specific heat capacities of liquid sodium ratio ranges from about 0.45 to 0.65 over the temperatures
and Hitec is important as an indicator for the increase in and thus the volumetric flow rate for Hitec will be less than
storage volume to be expected. that using sodium. For the same diameter pipe the Hitec
For this comparison we want the receiver inlet and out- will thus require lower velocities (see ratio in Fig. 2). This
let temperatures of the HTF to be the same for both fluid ratio also signifies the increase in volume that will be

2000 1 1.5 1.5

1800 (a) 0.9 (b)


1.4 1.4
1600 0.8
Ratio: Sodium/Hitec

Ratio: Sodium/Hitec
Density (kg/m3)

1400 0.7 1.3 1.3


Cp (kJ/kg.°C)

1200 0.6
1.2 1.2
1000 0.5
1.1 1.1
800 0.4
600 0.3 1 1
400 0.2
Sodium Sodium
0.9 0.9
200 Hitec 0.1 Hitec
Ratio Ratio
0 0 0.8 0.8
100 200 300 400 500 600 100 200 300 400 500 600
Temperature (°C) Temperature (°C)

0 2 3
10 10 10

10
-2 (c) (d)
Dynamic Viscosity (kg/m.s)

Ratio Sodium/Hitec

Ratio Sodium/Hitec

1
10
k (W/m.k)

-1 2
10 10
-3
10
0
10

Sodium Sodium
Sodium
Hitec Hitec
-4
Hitec
Ratio -2 -1 Ratio 1
10 10 10 10
100 200 300 400 500 600 100 200 300 400 500 600
Temperature (°C) Temperature (°C)

Fig. 1. Properties of liquid sodium and Hitec at temperature 350–900 K. (a) Density (kg/m3), (b) specific heat capacities at constant pressure (kJ/kg °C),
(c) dynamic viscosity (kg/ms) and (d) thermal conductivity (W/m °C). From equations in Tables 1 and 2.
N. Boerema et al. / Solar Energy 86 (2012) 2293–2305 2297

0.8 Nusselt number equations for liquid sodium do not include


a friction factor, however Cebeci (2002, pp. 205) has shown
0.7 0.65 that for low Prandtl number fluids (Pr < < 1) the Nusselt
Volumetric Cp (kWh/m3.° C)

number is much less dependent on the pipe roughness.

Ratio: Sodium/Hitec
0.6
0.6 For liquid sodium, the equations are given by (Sleicher
0.5 and Rouse, 1975):
0.55  
0:004  Pr  0:01
0.4 Constant Heat Flux Nu ¼ 6:3 þ 0:0167Re0:85 Pr0:93
S
104  Re  106
0.5
0.3 ð3Þ
 
0:004  Pr  0:01
Sodium 0.45 Constant Surface Temperature Nu ¼ 4:8 þ 0:0167Re0:85 Pr0:93
S
0.2 104  Re  106
Hitec
Ratio ð4Þ
0.1 0.4
80 130 180 230 280 330 380 430 480 530 580 630
For higher Prandtl number fluids, such as Hitec, the
Temperature (°C)
Nusselt number is given by (Gnielinski, 1976, pp. 359–368):
Fig. 2. Heat capacity of liquid sodium and Hitec from equations in Tables  
1 and 2. ðf =8ÞðRe  1000ÞPr 0:5  Pr  2000
Nu ¼
1 þ 12:7ðf =8Þ0:5 ðPr2=3  1Þ 3  103  Re  5  106
required when using liquid sodium compared to Hitec for ð5Þ
storage.
where e is the pipe roughness height, h is the heat transfer
2.1. Storage cost coefficient, V is the average fluid velocity, D is the pipe
diameter (a value of 10 mm has been used), Re is the Rey-
Storage costs for 4–6 h of full capacity operation are nolds number and Nu is the Nusselt number.
estimated to make up a relatively small percentage of cap- Fig. 3 shows that Hitec has a high Prandtl number
ital costs for molten salts. Hinkley et al. (2011) have calcu- (45–1), depending on temperature, whilst liquid sodium
lated values of about 4% for 6 h (Molten salt AUS$12/ has a very low Prandtl number (0.01–0.004), with only
kW hth), whilst Gil et al. (2009) predict a higher 8%. How- relatively small variation over the temperature range.
ever, Herrmann and Kearney (2002) have estimated prices The heat transfer coefficient can be calculated by taking
for sodium and Hitec at US$2/kg and US$0.5/kg, respec- the product of the Nusselt number (constant heat flux
tively. The effect of the fourfold price increase for sodium equation used for sodium) with the thermal conductivity
is made up for partially by sodium’s higher specific heat. of the fluid divided by the tube diameter. This has been cal-
However, the result is that the fluid costs for sodium are culated for mass flow rates of 0.05, 0.1, 0.2 and 0.5 kg/s,
approximately 3.2 times the costs of Hitec and requires which equate to velocities of approximately 0.75, 1.5, 3
about 1.8 times the storage volume. The storage amount and 7.5 m/s for liquid sodium, and to about 0.35, 0.7,
will be somewhat reduced if sodium’s ability to handle 1.45 and 3.6 m/s for Hitec for a 10 mm diameter pipe.
higher temperatures (and hence the difference between the As a reference for the following figures, the IEA-SPSS
hot and cold fluid temperatures) is utilised. This greater billboard receiver had a total design flow rate, using liquid
temperature difference will also reduce the volume of sodium, of 7.3 kg/s. The receiver was made up of five
HTF that needs to be circulated, resulting in a reduction
in the pumping power required. These large increases in 2
10 10
-2

storage costs need to be justified by an increase in electric-


ity production from utilising higher temperatures (higher 1
10
turbine efficiencies) or alternative storage methods need
Ratio Sodium/Hitec

to be found.
Prandtl No.

0
10 Sodium
3. Heat transfer and pump work Hitec -3
Ratio 10
-1
10
3.1. Heat transfer
-2
To determine the internal pipe heat transfer coefficient 10
the Prandtl number, Nusselt number, Reynolds number
and friction factor are needed. The friction factor has been -3
10 10
-4

solved using the implicit Colebrook equation ( Cengel, 100 150 200 250 300 350 400 450 500 550 600
Temperature (°C)
2007, pp. 475).
The following equations for calculating the Nusselt Fig. 3. Prandtl number (Pr = lCp/k) for liquid sodium and Hitec.
number are recommended by Cengel (2007, pp. 474). The Calculated using equations in Tables 1 and 2.
2298 N. Boerema et al. / Solar Energy 86 (2012) 2293–2305

5 5 5 5

Pressure Drop (Pa/m)

Pressure Drop (Pa/m)


10 10 10 10
h (W/m 2.°C)

h (W/m 2.°C)
4 4
10 10
4 4
10 10
3 3
10 10

3
(a) 3
(b) 2 (a) 2 (b)
10 10 10 10
200 400 600 200 400 600 200 300 400 500 600 200 300 400 500 600
Temperature ( °C) Temperature ( °C) Temperature (°C) Temperature (°C)
5 5 5

Pressure Drop (Pa/m)

Pressure Drop (Pa/m)


5
10 10 10 10
h (W/m 2.°C)

h (W/m 2.°C)

4 4
10 10
4 4
10 10
3 3
10 10

3 (c) 3 (d) 2 (c) 2 (d)


10 10 10 10
200 400 600 200 400 600 200 300 400 500 600 200 300 400 500 600
Temperature ( °C) Temperature ( °C) Temperature (°C) Temperature (°C)

Fig. 4. Heat transfer coefficient (using equations in Tables 1 and 2) for Fig. 5. Pressure drop (Pa/m) for mass flow rates: (a) 0.05 kg/s, (b) 0.1 kg/
mass flow rates: (a) 0.05 kg/s, (b) 0.1 kg/s, (c) 0.2 kg/s and (d) 0.5 kg/s. s, (c) 0.2 kg/s, (d) 0.5 kg/s. e = 0.002 mm (Stainless steel).
e = 0.002 mm Stainless steel e = 0.002 mm, D = 10 mm.

2.85 m high panels connected in series, with each panel in heat transfer coefficients Hitec would require a longer
consisting of 39 vertical tubes, connected in parallel. This pipe length or a higher temperature difference (across the
equates to a design mass flow rate per tube of 0.187 kg/s tube wall) to achieve the same outlet temperature. Whilst
and a series tube length of approximately 14 m. The tube heat transfer coefficients are quite stable for various mass
internal diameter was 12 mm resulting in a design flow flow rates using liquid sodium, higher Reynolds numbers
velocity of about 2 m/s, a Reynolds number of 8  104 are needed to increase the heat transfer coefficients using
and a pressure drop of approximately 3300 Pa/m (Schiel Hitec (see Fig. 4). The high pressure losses that occur at
and Geyer, 1988). high fluid velocities demonstrate the difficulty in achieving
Fig. 4 shows that the heat transfer coefficient of liquid these high Reynolds numbers (and thus high heat transfer
sodium is about an order of magnitude higher than that coefficients) when using Hitec.
of Hitec except when using rough pipes at very high fluid Figs. 6 and 7 allow the heat transfer coefficient and pres-
velocities. It also shows that the heat transfer coefficient sure losses to be compared for different pipe diameters and
for liquid sodium is almost independent of fluid tempera- mass flow rates (note that the heat transfer increases with
ture and fluid velocity for these mass flow rates. The large mass flow rate). The fluid temperature used was 500 °C
difference in heat transfer coefficients is predominantly due
to the large difference in Reynolds numbers for the two flu-
ids at the same mass flow rate or velocity. For equivalent 10
6

Reynolds numbers the difference between the heat transfer


coefficients is reduced, however an equivalent Reynolds
5
number is much more difficult to achieve using Hitec. 10
Pressure Drop (Pa/m)

3.2. Pressure loss considerations 4


10

Whilst increasing the mass flow rate will increase heat


transfer, the compromise is the resultant increase in pres- 3
10
sure loss and pump work. The pressure drop due to friction
D=6 mm
per metre of pipe for a given diameter is given by: D=8 mm
2
qV 2 10 D=10 mm
DP L ¼ f ðPa=mÞ ð6Þ D=12 mm
2D D=15 mm
D=20 mm
where DPL = pressure drop per metre. 10
1

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6


Fig. 5 shows that liquid sodium and Hitec will have sim-
2 5
ilar pressure losses for a given mass flow rate, and that Heat transfer coefficient (W/m .°C) x10
pressure losses are much more dependent on the pipe diam- Fig. 6. Heat transfer coefficient (W/m2 °C) versus pressure drop per metre
eter and fluid velocity than on fluid temperature. Note that (Pa/m). Mass flow rates are 0.05–0.8 kg/s. Heat transfer coefficients
pressure losses are per metre and that due to the differences increase with increasing flow rate. e = 0.002 mm (Stainless steel).
N. Boerema et al. / Solar Energy 86 (2012) 2293–2305 2299

1.8 3.3. Film temperatures


D=6 mm
D=8 mm
D=10 mm Hitec degrades at temperatures above 550 °C. For this
1.7
D=12 mm reason the film temperature (the fluid temperature at the
D=15 mm
wall, or the inner pipe temperature) is important as it will
Pressure loss ratio

D=20 mm
1.6 be hotter than the bulk fluid temperature, meaning that
the surface temperature will need to be limited and flow
1.5
conditions maintained such that the film temperature limits
are not exceeded. The possibility of hotspots on the recei-
ver from heliostat aiming errors and mirror image quality
1.4 and the possibility of fluid pump failures also increase the
risk of exceeding the film temperature limits when using
1.3 Hitec and must be considered in the receiver design and
system operation.
0 5 10 15 20 25 30 35 40 45 50 For steady state conditions, the inside wall temperature
Heat transfer coefficient ratio of a pipe can be calculated using:
Fig. 7. Sodium/Hitec: ratio of heat transfer coefficients versus pressure T S  BT 11 r1 h
loss ratios. e = 0.002 mm (Stainless steel). T ðr1 Þ ¼ ; where B ¼ lnðr1 =r2 Þ ð7Þ
1B k
where TS is the pipe outside surface temperature; T11 the
and pump efficiency has not been included. These losses are fluid temperature away from the wall; r1 and r2 the pipe in-
for a single pipe, however in a receiver there would be side and outside radii, respectively.
many pipes running in parallel, multiplying the mass flow Eq. (7) can then be solved for a given surface tempera-
rate but not the pressure drop. ture to yield the film temperature. The results are shown
The pressure drop across the receiver is an important in Fig. 8 for various mass flow rates and demonstrate the
design parameter and must be kept to a minimum to min- temperature difference that must be considered when
imise the required upstream pressures and the pump work. designing a receiver using a molten salt.
These constraints place limits on the heat transfer enhance-
ment that can be achieved simply by increasing fluid veloc- 4. Effect of heat transfer fluid on high temperature receiver
ity, or through using roughened pipe. Other heat transfer design and efficiency
enhancement methods, such as using fins, are also possible
for Hitec. The heat transfer enhancement to a molten salt It has now been shown that, compared to Hitec, liquid
using a spiral tube has been investigated by Yang and Yang sodium achieves higher heat transfer coefficients for an
et al. (2009) who concluded that the heat transfer rate equivalent pressure drop. To investigate the implications
could be increased approximately three fold compared to of these findings on receiver design and efficiency, a simple
a smooth pipe, although the corresponding increase in model of a billboard receiver has been developed for liquid
pressure drop was not given. There are some difficulties sodium and Hitec.
in enhancing the already high heat transfer rates achieved
using liquid sodium. Liquid sodium has very high thermal 4.1. Receiver model
conductivity and thus, even in turbulent flow, conduction
still contributes strongly to the heat transfer. As a result, For the receiver model an inlet temperature of 285 °C
the increase in heat transfer due to increasing turbulence and a desired outlet temperature of 565 °C have been used.
is not as significant as for higher Prandtl number fluids, These inlet/outlet temperatures were chosen as they are the
such as Hitec. The conductivity of liquid sodium is also expected temperatures to be used in the Gemasolar central
higher than most metals that sodium is compatible with, receiver plant (Ortega et al., 2008). The design specification
and thus increasing the pipe surface area will not result is to achieve 1MWth of heat transfer into the HTF whilst
in increased heat transfer if the energy must travel through not exceeding a pressure drop of 0.5 bar. This pressure
an increased amount of pipe material. Hitec’s low conduc- drop is equivalent to the design pressure drop used for
tivity again means that this heat transfer enhancement the IEA/SSPS receiver (Schiel and Geyer, 1988). The recei-
method will be more effective for Hitec. ver consists of a number of parallel pipes. No spacing has
Fig. 5 shows that at a given mass flow rate the pressure been included between the pipes. A 1 mm pipe wall thick-
loss is greater using liquid sodium (due mainly to liquid ness has been used along with a pipe thermal conductivity
sodium’s low density), however the pressure loss ratio of 22.6 (W/m °C) (AISI 304 stainless @ 800 K ( Cengel,
(1.25–2.05) is much smaller than the ratio of heat transfer 2007)). A pipe surface roughness of 0.002 mm was used
coefficients (1–50) (Fig. 7). This means that for an equiva- (stainless steel). An adiabatic backing has been assumed
lent pressure loss the rate of heat transfer using liquid and thus the receiver loses heat to the environment
sodium will be greater than that using Hitec. (25 °C) from only one side. The view factor for this geom-
2300 N. Boerema et al. / Solar Energy 86 (2012) 2293–2305

Temperature (°C)

Temperature (°C)
600 600

400 400

200 200
(a) (b)
0 0
200 400 600 200 400 600
Fluid Temperature (°C) Fluid Temperature (°C)
Temperature (°C)

Temperature (°C)
600 600

400 400

200 200
(c) (d)
0 0
200 400 600 200 400 600
Fluid Temperature (°C) Fluid Temperature (°C)

Fig. 8. (i) Wall temperatures for liquid sodium and Hitec for a surface temperature of 650 °C. Mass flow rates are: (a) 0.05 kg/s, (b) 0.1 kg/s, (c) 0.2 kg/s
and (d) 0.5 kg/s. Diameter = 10 mm. pipe thermal conductivity = 21 W/m2 °C. Wall thickness = 1 mm (ii) temperature locations.

etry has been calculated to be 0.6366. The incident radia- Qrad ¼ erF 12 Aseg ðT 4S  T 41 Þ Radiative heat loss ð9Þ
tion is assumed to be evenly distributed over the exposed Qconv ¼ hconv Aseg ðT S  T a Þ Convective heat loss ð10Þ
pipe surface, which results in a pipe surface radiation inten-
sity (W/m2 pipe) of Ipipe = Iap(2/p). Where Iap is the inten- Qabs ¼ I pipe Aseg ð1  qR Þ Energy being absorbed ð11Þ
sity over the receiver area and is taken as 900CR (W/m2), Qabs ¼ Qf þ Qrad þ Qconv ð12Þ
where CR is the concentration ratio.
where r is the Stefan-Boltzmann constant, Aseg is the seg-
ments surface area, T1 is the temperature of the surrounds,
4.2. Energy balance and pressure loss calculations hconv is the convective heat transfer coefficient (a value of
30 W/m2K has been used) and Ta is the ambient
For a selected intensity and pipe diameter, the rate of temperature.
energy transfer to the HTF, the radiative and convective The thermal losses, energy transferred to the HTF, and
losses and the pressure drop can be calculated. A fixed sur- pressure drop for the segment can then be calculated. The
face reflectivity, qR, of 7% was used along with a constant temperature at the end of the segment T2 is then given by:
emissivity, e, for long wavelengths (both 0.3 and 1 used).
In order to determine the flow rate, pipe length and Q_ f
T2 ¼ þ T1 ð13Þ
number of pipes in parallel such that the desired outlet tem- _ p
mC
perature is achieved without exceeding the maximum pres-
where m_ is the mass flow rate (kg/s), Cp is the specific heat
sure drop, an iterative procedure was used to solve Eqs.
of the fluid at the segments bulk mean temperature and T1
(8)–(14).
is the temperature at the start of the segment.
To start the iterations an initial length and mass flow
These calculations can be repeated for all segments. If
rate are assumed. The pipe length is divided into segments
the final fluid temperature at the end of the final segment
of length dx. Using the fluid inlet temperature, the fluid
does not match the desired exit temperature the length
properties, friction factor and heat transfer coefficient hf
can be adjusted. The losses, energy transferred to the
can be calculated for the first segment. The pipe wall ther-
HTF, and pressure drop for the entire pipe length can be
mal resistance, the thermal convective resistance and the
calculated by summing the segments:
total resistance (RT) can be calculated using standard one
dimensional pipe resistance equations ( Cengel, 2007, pp. X
m
qi V 2i
146–152): DP L ¼ fi dx ð14Þ
2D
The segment’s surface temperature Ts can then be solved i¼1

using the following five equations:


where m is the number of sections and fi, qi and vi are the
ðT S  T f Þ friction factor, fluid density and fluid velocity, respectively,
Qf ¼ Heat transfer to fluid ð8Þ
RT for section i.
N. Boerema et al. / Solar Energy 86 (2012) 2293–2305 2301

Table 1
Properties of liquid sodium. T is the fluid bulk temperature in Kelvin.
Property Equations Units
0:5
Density qS ¼ 219 þ 275:32ð1  T =2503:7Þ þ 511:58ð1  T =2503:7Þ kg/m3
Specific heat capacity at constant pressure C P S ¼ 1:6582  8:4790  104 T þ 4:4541  107 T 2  2992:6T 2 kJ/kg °C
Thermal conductivity k S ¼ 124:67  0:11381T þ 5:5226  105 T 2  1:1842  108 T 3 W/m °C
Dynamic viscosity logðls Þ ¼ 6:4406  0:3958 logðT Þ þ 556:835=T Pa s

Table 2 is reduced. This reduces the temperature difference between


Properties of Hitec. T is the fluid bulk temperature in Kelvin. the wall and the fluid and thus reduces the increase in the
Property Equations Units rate of heat transfer to the fluid (see Fig. 9).
Reducing the surface temperature reduces the radiation
Density qH ¼ 0:733ðT  273:15Þ þ 2080 kg/m3
Specific heat capacity at C P H ¼ 1:560  ðT  273:15Þ=1000 kJ/kg °C and convection losses. However, the much higher heat
constant pressure transfer coefficient of liquid sodium only decreases the lost
Thermal conductivity k H ¼ 0:78  1:25  103 T þ W/m °C efficiency by about 1% relative to the Hitec losses. This is
1:6  106 T 2 due to the difference in surface temperatures between the
Dynamic viscosity b ¼ 5:9ðT  9:638Þ=990:362 Pa s
b b two fluids being quite small compared to the overall surface
lH ¼ eeb þe
eb
 0:999
temperature difference from ambient.
The minimum pipe surface temperature that can be
achieved is the fluid temperature. This means that the min-
The pressure drop along the pipe is compared to the
imum efficiency loss due to thermal losses is about 1.4%
design pressure drop while the mass flow rate is adjusted
(for a blackbody and hconv = 30 W/m2 °C). To decrease
accordingly. Changing the mass flow rate will change the
the radiation losses further, a selective surface can be used.
end fluid temperature and thus a new pipe length to achieve
For a decrease in emissivity from 1 to 0.3 the efficiency loss
the required end temperature must be calculated. The itera-
from re-radiation were reduced by about 1% and 1.8% at
tions continue until the desired end temperature and design
CR = 1300 and by 0.6% and 1.4% for liquid sodium and
pressure drop are achieved. The required number of parallel
Hitec respectively, at CR = 3000.
pipes needed to achieve 1 MWth is then calculated. As this
number must be an integer the heat transfer will not be
exactly 1 MW and thus the receiver area can be scaled by 4.3.1. Thermal losses
the ratio of Qdesired/Qactual for a better comparison. The efficiency loss from re-radiation from a blackbody
at a given surface temperature and under various sun inten-
sities can be seen in Fig. 10. At a given temperature the re-
4.3. Receiver model results and discussion radiation will be the same however its fraction of the
incoming energy changes. This figure shows that radiative
From Table 3 we can see that although the heat transfer losses are particularly important at higher surface temper-
coefficient of liquid sodium is 6–10 times greater than that atures and for part sun conditions. Fig. 11 shows the large
of Hitec for these diameters, the areas required to achieve reduction that can be achieved using a selective surface.
1MW heat transfer are all near to 0.95 m2 for a concentra- Fig. 12 shows that, with the calculated heat transfer
tion of 1300 and a sun intensity of 900 W/m2. coefficients, a blackbody surface would have efficiency
This can be explained due to the fact that as the heat losses from re-radiation on the order of 2–9% for
transfer coefficient increases the pipe surface temperature CR = 1300. Using a cut-off wavelength of 2.51 lm these

Table 3
Inputs and results of receiver model using liquid sodium and Hitec. Greyed cells show input changes.
Inputs Sodium Hitec Sodium Hitec Sodium Hitec Sodium Hitec Sodium Hitec Hitec
Di (mm) 9 9 9 9 9 9 9 9 16 16 16
Emissivity 0.3 0.3 1 1 0.3 0.3 1 1 0.3 0.3 0.3
Pipe roughness (mm) 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.045
Concentration ratio 1300 1300 1300 1300 3000 3000 3000 3000 1300 1300 1300
Results
Area corrected (m2) 0.94 0.95 0.95 0.97 0.40 0.41 0.41 0.41 0.94 0.95 0.95
Mass flow rate (kg/s) 0.169 0.200 0.169 0.197 0.226 0.276 0.234 0.274 0.559 0.644 0.610
h average (W/m2 °C) 62,821 6523 62,868 6436 66,853 8642 67,466 8575 41,141 6072 6985
TS @ exit (°C) 607 696 603 691 675 823 671 816 614 696 679
Efficiency (%) 90.77 89.98 89.74 88.17 91.85 91.15 91.23 89.72 90.74 90.00 90.10
Eff. loss: Radiation (%) 0.45 0.79 1.50 2.60 0.27 0.63 0.90 2.07 0.47 0.77 0.73
Eff. loss: convective (%) 1.77 2.24 1.76 2.23 0.88 1.22 0.87 1.21 1.79 2.22 2.17
2302 N. Boerema et al. / Solar Energy 86 (2012) 2293–2305

(a) (b)
Fig. 9. (a) Receiver model geometry and (b) cross section of receiver.

25 9
OneSun=300 W/m2 OneSun=300 W/m2
OneSun=500 W/m2 OneSun=500 W/m2
OneSun=700 W/m2 8
20 OneSun=700 W/m2
OneSun=900 W/m2
OneSun=900 W/m2
OneSun=1100 W/m2
Radiation Losses (%)

7 OneSun=1100 W/m2

15 Radiation Losses (%)


6

10 5

4
5
3

0 2
400 450 500 550 600 650 700 750 800 850 900 2 3 4 5 6 7 8 9 10
Surface Temperature (°C)
1/R T (W/m 2.°C) x105
Fig. 10. Receiver efficiency loss from re-radiation (%) as a function of
surface temperature and sun intensity (absorbtivity/emissivity = 0.95), Fig. 12. Receiver efficiency losses from re-radiation for 1/RT (changing
concentration ratio = 1300. heat transfer coefficient). Absorbtivity/emissivity = 0.95, C = 1300.

8 1.6
OneSun=300 W/m2 OneSun=300 W/m2
OneSun=500 W/m2 OneSun=500 W/m2
7 1.4
OneSun=700 W/m2 OneSun=700 W/m2
OneSun=900 W/m2 OneSun=900 W/m2
6
Radiation Losses (%)

OneSun=1100 W/m2 1.2 OneSun=1100 W/m2


Radiation Losses (%)

5
1
4
0.8
3
0.6
2

0.4
1

0 0.2
2 3 4 5 6 7 8 9 10
400 450 500 550 600 650 700 750 800 850 900
Surface Temperature (°C) 1/R T (W/m 2.°C) x105

Fig. 11. Near ideal receiver: absorbtivity = 0.95 until cut-off wavelength Fig. 13. Near ideal receiver: absorbtivity = 0.95 until cut-off wavelength
of 2.51 lm, then absorbtivity/emissivity = 0.05. of 2.51 lm, then absorbtivity/emissivity = 0.05, C = 1300.

losses can be reduced to below about 1% (Fig. 13). This can that can be achieved through using a fluid with a higher
be compared to the far smaller reduction in lost efficiency heat transfer coefficient (Figs. 12 and 13). Thus, increasing
N. Boerema et al. / Solar Energy 86 (2012) 2293–2305 2303

the absorbtivity of solar wavelengths whilst reducing the of 2400 leading to a required area of 0.51 m2 (47% area
emissivity of the receiver (selective surface) is a more effec- decrease) and an efficiency of 90.9% (1.5% efficiency
tive means of increasing efficiency than increasing the heat increase).
transfer coefficient of the fluid (for a given concentration The area reduction will allow cost savings due to mate-
ratio). rial reduction, lower design conditions for wind loading
Using the receiver model, the effect of different concen- and reduced manufacturing costs. However to achieve an
trations can be determined. The effect of increasing the increased concentration ratio, it will be necessary to
concentration ratio is that efficiencies increase. Particularly increase the heliostat accuracy and to increase the number
important though is that the receiver efficiencies (for the of heliostats per receiver (field size). Furthermore, the
same fluid) at different sun intensities become less sensitive increased thermal strains on the receiver will increase mate-
to sun intensity due to less energy being lost through re- rial wear. It should also be noted that the stagnation tem-
radiation. This is a result of the increase in surface temper- perature at this intensity is about 1800 °C and the total
atures being relatively low compared to the concentration mass of this receiver is only 11 kg (including HTF). Thus
ratio increase (nonlinear). This effect is true for both liquid if the mass flow of the HTF is stopped, the receiver surface
sodium and Hitec (Fig. 14). temperature will initially be increasing at a rate of 125 °C/s
and will increase by a 1000 °C in under 20 s.
4.3.2. Surface temperatures In the receiver programme, it was assumed that the pipe
For a given concentration, the receiver will have similar length was not limited (as the receiver could be made up of
efficiencies and receiver areas for both fluids (Table 3). The panels connected in series to achieve the required length
difference between the two receivers is that the surface tem- whilst maintaining a useable geometry). Fig. 15 shows that
peratures of the Hitec receiver are about 90 °C greater than increasing the pipe length, which will also increase the mass
for the sodium receiver. It is likely that the design concen- flow rate, has almost no effect on the efficiencies and max-
tration ratio chosen for a receiver will be restricted by an imum surface temperatures using liquid sodium, whilst it
upper limit of the surface temperature, due to receiver results in increased efficiencies and thus decreased surface
material and HTF fluid limitations. Thus a maximum tem- temperatures for Hitec. This is a result of the predominant
perature limit should be used in the design. To account for heat transfer mechanism for liquid sodium being conduc-
this, the receiver optimisation can be re-performed but with tion, whilst for Hitec, the predominant heat transfer mech-
the concentration ratio for the sodium receiver being anism is forced convection, and thus its heat transfer
increased until the maximum temperature is equal to the depends strongly on the Reynolds number. This is impor-
maximum temperature of the Hitec receiver. For the inten- tant as, for simplicity; the receiver may be designed to be
sity distribution used, the maximum temperature will be a single panel with a set length.
the surface temperature at the pipe exit. Using a fixed receiver length (4 m has been used) and
From Table 3 it can be seen that, for a concentration altering the mass flow rate such that the desired fluid exit
ratio of 1300, the exit surface temperature using Hitec is temperature is achieved, the efficiency and maximum sur-
approximately 690 °C. Using this value and 9 mm piping, face temperature (temperature at pipe exit) is able to be cal-
recalculating the receiver geometry using liquid sodium culated for various pipe diameters. Fig. 16 shows that,
and increasing the concentration ratio until the exit surface using liquid sodium, the efficiency and the pipe surface tem-
temperature is equal to 690 °C allows a concentration ratio perature at the receiver exit is quite stable for increasing
92 1300
94
90 1250 Pipe surface temperature at exit

92 1200
88
1150
Efficiency (%)

86
90 Hitec: Efficiency 1100
Efficiency (%)

84 Sodium: Efficiency
Hitec: Surface T 1050
88
82 Sodium: surface T
1000

86 80
950

Hitec: Concentration=1300 78 900


84 Sodium: Concentration=1300
Hitec: Concentration=3000 76 850
Sodium: Concentration=3000
2 3 4 5 6
82 Pipe Length (m)
300 400 500 600 700 800 900 1000 1100
Sun Intensity (W/m 2) Fig. 15. Hitec and sodium receiver efficiencies and maximum surface
temperatures for various pipe lengths. emissivity = 0.85, concentration
Fig. 14. Receiver emissivity/absorbtivity = 0.95, C = 1300. ratio = 1300, D = 9 mm.
2304 N. Boerema et al. / Solar Energy 86 (2012) 2293–2305

1200 ver systems. Using a receiver model developed herein, we


90 have identified several important differentiating parameters

Pipe surface temperature at exit (° C)


1100 for these fluids. The evident disadvantages of using liquid
85 sodium is that it is up to 200% more expensive per kilogram
1000 and that it must be completely isolated from the environ-
Hitec: Efficiency
Efficiency (%)

Sodium: Efficiency ment. Furthermore, its low volumetric specific heat capac-
80 Hitec: Surface T 900 ity results in higher storage volumes, although this is
Sodium: surface T
reduced if sodium’s higher temperature possibilities are uti-
800
75 lised. If these limitations can be overcome, however, this
article shows that there are several advantages. First, with
700
70 a heat transfer coefficient that is an order of magnitude
greater and a thermal conductivity that is two orders of
600
magnitude greater than that for Hitec, liquid sodium will
65
6 9 12 15 18 21 24 reduce the risk of hot spots and thus reduce pipe stresses.
Diameter (mm) This can also mean much slower flow rates for the same
level of heat transfer. Secondly, it was found that a receiver
Fig. 16. Hitec and sodium receiver efficiencies and maximum surface
temperatures for various pipe diameters. emissivity = 0.85 concentration using liquid sodium could achieve absorber area reductions
ratio = 1300 length = 4 m. of 57% and an absolute efficiency increase of 1.1% by uti-
lising higher concentration ratios.
Overall, the main advantage of employing liquid sodium
92 740 is its superior ability to stay liquid at both higher and lower
temperatures. Unless stable molten salts can be developed
Pipe surface temperature at exit ( ° C)

90 720
to operate over a similar range, liquid sodium is a viable
700
88 option. The greatly increased storage costs are a significant
680 deterrent to the use of this fluid, and thus alternative
Efficiency (%)

86
Hitec: Efficiency energy storage methods compatible with liquid sodium is
Sodium: Efficiency 660
84 Hitec: Surface T
another area recommended for further research.
640
Sodium: surface T
82
620 Appendix A
80
600 (1) The pipe surface radiation intensity can be calculated
78 580 as (W/m2 pipe):
76 560 I pipe ¼ I ap Aap =Apipe
400 600 800 1000 1200
2
Sun Intensity (W/m ) length  width
¼ I ap
Fig. 17. Hitec and sodium receiver efficiencies and maximum surface n  length  0:5pDo
temperatures for various sun intensities. emissivity = 0.85, concentration length  n  Do
ratio = 1300, D = 9 mm, length = 4 m. ¼ I ap
n  length  0:5pDo
pipe diameters whilst for Hitec they are highly dependent 2
on the pipe diameter. This is important as the use of larger I ap
p
diameter pipes reduces the number of pipes needed for the
receiver which thus reduces the manufacturing costs. The where Iap is the intensity over the receiver area, Aap is
stress due to thermal expansion is however proportional the receiver area, Apipe is the pipe surface area ex-
to the pipe diameter and thus must also be considered in posed to environment, Do is the pipe outer diameter
the receiver design to ensure that the receiver does not fail and n is the number of pipes in parallel.
before its design lifetime (Peng and Peng, 2009). Finally, (2) Pipe resistance calculations:
Fig. 17 shows that, for a fixed geometry and using liquid lnðDo =Di Þ
sodium, efficiencies will decrease less with a decrease in Rpipe ¼ 2
2pk pipe dx
sun intensity compared to when using Hitec. This is impor-
tant as throughout each operating day the receiver will be 1
Rf ¼ 2
exposed to different sun intensities. h f Af

Rr ¼ Rf þ Rpipe
5. Conclusion
where Di is the pipe inside diameter, kpipe is the pipe
This paper presents a comparison between the HTFs thermal conductivity and Af is the segments inside sur-
liquid sodium and Hitec molten salt for use in central recei- face area. Note that the prefix of two in the equations
N. Boerema et al. / Solar Energy 86 (2012) 2293–2305 2305

is due to the heat only being transferred from one side, Hinkley, J., Curtin, B., et al., 2011. Concentrating Solar Power-Drivers
thus resulting in a doubling of the resistances. and Opportunities for Cost-Competitive Electricity. Commonwealth
Scientific and Industrial Research Organisation (CSIRO), Australia.
Kearney, D., Herrmann, U., et al., 2003. Assessment of a molten salt heat
References transfer fluid in a parabolic trough solar field. J. Solar Energy Eng. 125
(2), 170–176.
Bignon, M.J., 1980. The influence of the heat transfer fluid on the receiver Luster, V.P., Freudenstein K.F., 1996. Feedback from practical experience
design. Electr. Power Syst. Res. 3 (1–2), 99–109. with large sodium fire accidents. In: Technical Committee Meeting on
Bohlmann, E.G., 1972. Heat Transfer Salt for High Temperature Steam Evaluation of Radioactive Materials Release and Sodium Fires in Fast
Generation. Reactor Chemistry Division. Reactors, International Atomic Energy Agency, Vienna, Austria.
Bradshaw, R.W., Siegel, N.P., 2008. Molten nitrate salt development for NREL, 2009. Solar Advisor Model. National Renewable Energy
thermal energy storage in parabolic trough solar power systems. In: Laboratory.
ASME Conference Proceedings, 2008(43208), pp. 631–637. Olivier, T.J., Radel, R.F., et al., 2007. Metal fire implications for advanced
Cebeci, T., 2002. Convective Heat Transfer. Horizons Publishing, reactors Part 1: Literature review. Sandia National Laboratiories.
Springer. Ortega, J.I., Burgaleta, J.I., et al., 2008. Central receiver system solar
Cengel, Y.A., 2007. Heat and Mass Transfer: A Practical Approach, third power plant using molten salt as heat transfer fluid. J. Sol. Energy Eng.
ed. McGraw-Hill, America. 130 (2), 024501.
CoastalChem., 2011. HITEC Heat Transfer Salt. Retrieved May 16th, Peng, L.-C., Peng, T.-L., 2009. Pipe Stress Engineering. Texas, American
2012, from http://www.coastalchem.com/PDFs/HITECSALT/ Society of Mechanical Engineers, ASME.
HITEC%20Heat%20Transfer%20Salt.pdf. Rodriquez, G., Sawada, M., 2006. Pocketbook on Sodium Technology.
Fink, J.K., Leibowitz, L., 1995. Thermodynamic and Transport Properties first ed. Retrieved June 17th, 2011, from http://www.cea-jaea-collab-
of Sodium Liquid and Vapour. Reactor Engineering Division, oration.net/napocket/pocketbook/page22.html.
Argonne National Laboratory. Schiel, W.J.C., Geyer, M.A., 1988. Testing an external sodium receiver up to
Fraas, A., 1989. Heat Exchanger Design. John Wiley & Sons, America. heat fluxes of 2.5 MW/m2: results and conclusions from the IEA-SSPS
Gil, A., Medrano, M., et al., 2009. State of the art on high temperature high flux experiment conducted at the central receiver system of the
thermal energy storage for power generation. Part 1 – Concepts, plataforma solar de Almeria (Spain). Sol. Energy 41 (3), 255–265.
materials and modellization. Renew. Sustain. Energy Rev. 14 (1), Sleicher, C.A., Rouse, M.W., 1975. A convenient correlation for heat
31–55. transfer to constant and variable property fluids in turbulent pipe flow.
Gnielinski, V., 1976. New equations for heat and mass transfer in Int. J. Heat Mass Transfer 18 (5), 677–683.
turbulent pipe and channel flow. Int. Chem. Eng. 16 (1), 8–16. Yang, M., Yang, X., et al., 2009. Heat transfer enhancement and
Herrmann, U., Kearney, D.W., 2002. Survey of thermal energy storage for performance of the molten salt receiver of a solar power tower. Appl.
parabolic trough power plants. Sol. Energy 124 (2), 145–152. Energy 87 (9), 2808–2811.

You might also like