Sie sind auf Seite 1von 11

17th IAHR International Conference on Cooling Tower and Heat Exchanger, 8-11 September

2015, Gold Coast, Queensland, Australia

COMPARISON OF DIRECT AND INDIRECT COOLING IN AN sCO2


BRAYTON CYCLE FOR CONCENTRATED SOLAR POWER PLANTS

Erond PEREZ, Sam DUNIAM, Yuanshen LU, Ingo JAHN, Kamel HOOMAN, and Anand
VEERARAGAVAN
The University of Queensland, Australia

Keywords: concentrated solar power, natural draft dry cooling tower, direct and indirect cooling
ABSTRACT: The supercritical carbon dioxide (sCO2) Brayton cycle has been identified as a
promising candidate for the power cycle in Concentrated Solar Power (CSP). It exhibits high
efficiency even with dry cooling, has smaller components and less complex cycle design. In this
paper, cooling of the sCO2 cycle is investigated by the use of Natural Draft Dry Cooling Towers
(NDDCTs). Two configurations are considered; direct and indirect cooling. Direct cooling is
where ambient air is used to cool the sCO2 that is sent to the NDDCT while in indirect cooling,
sCO2 rejects heat through a secondary water loop through an intermediate heat exchanger (IHX)
and the water is cooled by ambient air in the NDDCT. Direct cooling is investigated as a means to
reduce exergy losses to the additional heat transfer process in the IHX for indirect cooling. For
each case, size of the NDDCT are similar and their effect on cycle performance are discussed.

INTRODUCTION

Concentrating solar power (CSP) plants use heliostats to concentrate light from the sun onto a
receiver where it is converted to heat and used to generate electricity via a power generation cycle.
CSP power plants can be considered to consist of two main subsystems: the solar conversion
system and the power block or power cycle (Viebahn, Lechon, & Trieb, 2011). A lot of studies
and research is currently happening in the field of concentrating solar power systems, looking for
solutions to increase efficiency and lower overall cost to make CSP technology more competitive

to other methods of electricity generation thus making it more attractive to investors (Williges,
Lilliestam, & Patt, 2010).
One of the most prominent option is using supercritical CO2 in a closed loop Brayton cycle. This
cycle enjoys the benefit of having a higher efficiency compared to superheated steam Rankine
cycles when source temperature is above 550C (Dostal, 2004) and more compact components
can be achieved due to the low specific volume of carbon dioxide at the sCO2 cycles operating
pressure. It can also achieve relatively high cycle efficiency in dry cooling condition due to heat
rejection of sCO2 not being limited by the saturation temperature offering potentially cost-effective
cooling (Neises & Turchi, 2014) (J. Dyreby, Klein, Nellis, & Reindl, 2014). Brayton cycle also
has simpler control systems since there is no phase change in the heat exchangers unlike for
Rankine cycle where, to prevent pump damage due to cavitation, complicated control systems has
to be implemented to ensure fluid is fully condensed prior to entering the pump (Dostal, 2004) (C.
S. Turchi, Ma, Neises, & Wagner, 2013) (J. Dyreby et al., 2014). With regards to cooling, there
are several ways to achieve cooling in the CSP system, however, taking into consideration that
CSP power plants are most likely to be built in arid, dry, and with little to no access to water, dry
cooling is the best option.
Dry cooling can be achieved by either mechanical draft or natural draft. A mechanical draft cooling
tower uses a fan that is situated in either the air entrance or exit to move the air. This type of system
requires power to run the fan, is noisy and requires regular maintenance due to rotating
components. Natural draft dry cooling tower (NDDCT) on the other hand relies on the buoyancy
effects that exist between the less dense, heated air which is in contact with the finned tubes and
the denser, cooler ambient air. It does not require external power to move the air and is low
maintenance thus reducing running costs (SPX, 2015).
There are two configurations of dry cooling investigated in this paper; direct and indirect. These
are schematically shown in Fig 1. The main aim is to compare the two types of dry cooling
configuration of the same tower size, operating at one point, and assess how the cooling
configuration affect cycle performance. The first stage involves a cycle analysis to determine the
cooling system requirements for the reference case considered. These system requirements are
then used to perform sizing calculations on the indirect cooling system. The performance of the
tower sized for indirect cooling is then analyzed in direct cooling configuration. Finally, the
discussion compares the results to suggest the preferred option and identifies future work.

LITERATURE REVIEW

A number of studies in sCO2 Brayton cycle investigated the cycle performance using dry cooling.
J. J. Dyreby, Klein, Nellis, and Reindl (2013) stated that dry cooling performance can be easily
affected by weather conditions and the plant will not operate as per design point most of the time.
Thus, a model capable of predicting the off-design and part-load performance of the cycle is
necessary. The modelling approach utilizes thermodynamic models for the compressor, turbine
and heat exchanger and integrated into a system-level model. C. S. Turchi et al. (2013) investigated

the ability to dry cool more complex configurations such as pre-compression cycle, recompression
cycle and partial cooling cycle to achieve greater than 50% efficiency. The study concluded that
partial cooling cycles and recompression with main compression intercooling together with reheat
are able to achieve the target efficiency even with dry cooling. The paper by Padilla, Soo Too,
Benito, and Stein (2015) conducted a detailed energy and exergy analysis of four different sCO2
Brayton cycle configuration with dry air cooling for all the configuration studied. It was
determined that the simple Brayton cycle experiences the highest exergy destruction in the cooler
while recompression with main compression intercooling experiences the lowest exergy
destruction which is in agreement with result of Turchi. Another study by C. Turchi, Ma, and
Dyreby (2009) investigated two cycles, optimized at wet and dry cooling and analyzed at design
point for a 10MW net power cycle. With the same turbine inlet temperature, wet cooling is able to
achieve a compressor inlet temperature of 33C while dry cooling achieved 60C. This resulted in
cycle efficiency decrease of 4% at worst case condition.
Singh, Rowlands, and Ventura (2014) analyzed the dynamic performance of sCO2 Brayton cycle
as being direct heated and dry cooled. The fluctuations in heat input and ambient air temperatures
causes dynamic changes in CO2 mass flow rate, pressures and temperature thus also affecting the
cycles net power output. It was concluded that adding extra receiver section, extra thermal storage
volume or intermediate heat exchangers can ease the effect of such fluctuations. The simulation
model used in the study considered a cooling tower for sCO2-to-air heat exchange but no
discussion of dry cooling is presented as the study focused on direct heating of sCO2 and its effect
on dynamic performance.
A patent by Litwin, Zillmer, Hoffman, von Arx, and Wait (2010) about sCO2 Brayton cycle with
the heat rejection done by sCO2 rejecting heat to the water which is then sent to a cooling tower to
release heat to the atmosphere is studied. Litwin suggested an alternative configuration heat
rejection where it is accomplished by directly air cooling sCO2. No discussion is done as to how
different cooling configuration would affect performance of the cycle.
Conboy, Carlson, and Rochau (2015) investigated dry cooling of sCO2 for advanced nuclear
reactors and compared it with dry cooling of steam. It was determined that at same operating point,
steam is more difficult to dry cool since it experiences no temperature change thus requiring
significantly higher air flow rate that reached about 24 times compared to airflow required for
sCO2. This would consequently require a larger and more expensive cooling.
The investigation of Anton Moisseytsev (2014) focused on the use of direct fan assisted dry air
cooling for sCO2 Brayton cycle power converter for a 400MWe (1000MWth) Sodium-cooled fast
reactor and compared it to once-through water cooling at 20,000 kg/s. In spite of the optimization
conducted to air cooled sCO2 cycle, the study concluded that it will not be competitive with water
cooling and will result in at least 40% increase in electricity price thus limiting the application of
air cooling to situations where source of water is not available or expensive. Though the literature
reviewed here thoroughly investigated dry cooled Brayton cycle and were able to determine what
cycle configuration are best suited for dry cooling, no discussion or investigation is done on what
configuration would be best for cooling.

MODELLING APPROACH

The analysis is done in IPSEpro, a process simulation program, using the recuperated Brayton
cycle shown in Fig. 1. The power cycle components are the heat source heat exchanger,
compressor, turbine, recuperator and cooling tower. For indirect dry cooling, the components are
the same with direct, with the addition of an intermediate heat exchanger (IHX). The standard
component models in IPSEpro are used in the cycle modelling.

Figure 1: Cycle diagram for direct and indirect configuration

3.1

Cooling system modelling

Two configurations are considered; direct and indirect cooling. In both cases, the ambient air
temperature is assumed to be 25C. The minimum size of the dry cooling equipment which can
achieve the required compressor inlet temperature is determined for indirect cooling and the
calculated size is used to determine the performance of direct cooling.

Figure 2: Loss coefficients considered and notation used in NDDCT analysis (Krger, 2004).

The models for the NDDCT shown in Fig. 2 are developed based on the method presented by
Krger (2004). This method balancing the buoyancy force generated due to heat transfer to the air,
against the resistances to air flow through the tower. The heat transfer coefficient for water was
calculated using the Gnielinski correlation shown in Eq. 1.
=

( /8)( 1000) [1 + (/)0.67 ]


1 + 12.7( /0.8)0.5 ( 0.67 1)

(1)

where the friction factor, fD, is calculated using the Colebrook equation, d is the tube inner diameter
and L is the effective tube length. Re and Pr are calculated at mean water properties. The heat
transfer coefficient is then calculated using Eq. 2.

(2)

where k is the thermal conductivity. The indirect cooling NDDCT model, with water being cooled
in the NDDCT, based on the Krger method has been validated against the results presented in
Krger, the validation is presented in (Duniam, 2015). The approach in the water cooling model
involves the use of the mean properties of the inlet and outlet states. This method is suitable for
cooling of liquid water, but is not sufficient for sCO2 as the non-linear cooling profile, due to the
shape of the isobar of sCO2, and the rapidly changing properties of sCO2 for small changes in
temperature and pressure require a more detailed approach. Fig. 3 illustrates the variability of sCO2
Prandtl number and specific heat capacity for the heat exchanger working pressure considered in
this work, and this variation becomes much more pronounced as the temperature approaches the
critical point (31 oC, 7.35 MPa).

Figure 3: Variation of sCO2 specific heat and Prandtl number throughout cooling process at 7.35 MPa,
calculated using CO2 properties from REFPROP v 9.1.

The above approach is suitable for sCO2 cooling, however a more detailed approach is required to
accurately account for the heat transfer process; this is achieved by calculating the temperature
profile and properties at slices through the heat transfer process. This gives a more accurate value
for the mean temperature difference (MTD).
The selection of the heat transfer coefficient correlation for sCO2 requires additional consideration.
Tanimizu and Sadr (2015) found that the Dittus Boelter correlation over-predicted the Nusselt
number as compared to experimental results for convection heat transfer in sCO2 in a horizontal
pipe. The Wang Peng correlation is a modification of the commonly used Dittus Boelter
correlation which uses a lower experimental coefficient to predict a lower Nu. Fernndez and
Sedano (2013) found that the Wang Peng correlation shown in Eq. 3 found good agreement with
their CFD analysis results for sCO2 heat transfer. The Wang Peng correlation is given by:
= 0.0085 0.8 0.4

(3)

The sCO2 properties are calculated at each slice and the mean of each of the properties is then used
to calculate the Reynolds number and Prandtl number, which are used to determine the Nusselt

number using Eq. 3 above, which is then used to calculate the Tubeside heat transfer coefficient
using Eq 2.
The IHX and NDDCT distribution system are not geometrically defined to the level of detail
required to accurately quantify pressure loss for comparative purposes thus an assumed value is
used for pressure loss in each case. Furthermore, as shown by Fang, Xu, Su, and Shi (2012), the
currently available pressure loss correlations for supercritical fluids are not totally satisfactory and
do not suitably account for the combination of friction and acceleration pressure drop.
For both models fixed finned tube heat exchanger bundle geometry is used, based on Kroger
(2014), where the following relations are determined by experimental findings; the characteristic
heat transfer parameter is given by:
(4)
= 383.617313 0.523761
where Ry is the characteristic flow parameter, which is defined as = /34 , where the
subscript a34 refers to the mean airside properties in passage through the heat exchanger, and A fr
is the total frontal area of the heat exchangers. The heat exchanger loss coefficient for isothermal
normal flow is given by:
(5)
= 1383.94795 0.332458

RESULTS

4.1

Cycle Analysis

The cycle analysis uses the values in Tab. 1, to find the cooling system requirements, which will
be used in the analysis of the cooling configurations. The values for component efficiencies and
constraints are as per C. S. Turchi et al. (2013) and (Dostal (2004)). The values assume a largescale, commercial system of approximately 100MWe net power generation.
Design Parameters
Net power output

Value
100 MW

Design Parameters
Turbine inlet temperature

Turbine efficiency

93%

Compressor inlet temperature

Compressor efficiency
Heat exchanger efficiency
Heat exchanger
Heat exchanger pressure loss

89%
95%
5 oC
20 kPa

Turbine inlet pressure


Compressor inlet pressure
Ambient air temperature
Cycle efficiency

Value

700 oC
45 oC
25 MPa
7.35 MPa
25 oC

42.81 %

Table 1: Cycle design parameter

From the cycle analysis the pre-cooling system inlets are obtained and these are used in the
NDDCT sizing analysis for the indirect configuration. The results are summarized in Tab. 2.

Cooling System Parameters


Heat rejection
sCO2 cycle mass flow rate
Cooling system sCO2 inlet temperature
Cooling system sCO2 inlet pressure
Cooling system sCO2 outlet temperature

Value
115.5 MWth
763.5 kg/s
152.8 oC
7.37 MPa
45 oC

Table 2: Cooling system requirements

4.2

NDDCT Analysis

There are many geometric variables to be selected in designing the NDDCT, so to simplify the
NDDCT design process constant geometric ratios shown below in Tab. 3 were used throughout.
Variable
Aspect ratio (H5/d3)
Diameter ratio (d5/d3) (d5 = throat diameter)
Heat exchanger coverage of tower inlet (Afr/A3)
Inlet height (H3)
Number of tower supports (nts)*
Length of tower supports (Lts)*

Value
1.4
0.7
0.65
d3 / 6.5
d3 / (82.96/60)
H3 x (15.78/13.67)

Table 3: Fixed geometric ratios used for NDDCT analysis, *ratios based on those used in Krger (2014).

The first stage of the NDDCT analysis involves calculation to determine the minimum size of the
NDDCT for indirect configuration which can achieve cooling of the outlet sCO2 stream to the
desired temperature of 45oC. This was performed considering the cooling system in isolation and
by setting the IHX inlet values to those identified in Tab. 2 and then increasing the tower size that
cooled the sCO2 stream to below 45oC. The method used was to select a low starting value for
tower size and then increase the tower size and the cooling water mass flow rate proportionally to
find the smallest tower size which can achieve the desired outlet temperature. The cooling water
inlet temperature is also manually selected, and required a tradeoff to find the highest temperature
which still gives a cooling water pump outlet temperature of which satisfies the requirement
of 5oC. The cooling system configuration which satisfies the above criteria is shown below in Fig.
4. For this configuration the selection of the pinch point temperature difference in the IHX has a
very significant influence on the resultant tower size. The above process was performed for
ranging from 3 to 5oC, the results shown in Tab. 4.

Tower height, H5 [m]


Heat exchanger area [m2]
Air mass flow rate [kg/s]
Air outlet temperature

= 5oC
106.09
15,926
6310
42.16

= 4oC
103.10
15,041
6007
43.07

= 3oC
100.65
14,333
5768
43.91

Table 4: Influence of IHX Tmin specification on the required NDDCT size.

This method, with the use of fixed heat exchanger bundle geometry allows the entire geometry of
the NDDCT to be defined based on the specification of the number of heat exchangers.

Figure 4: Indirect cooling system configuration with smallest tower size to satisfy requirements given in
Tab. 2, using geometric relations in Tab. 3.

The Tmin specification increases the required IHX inlet temperature, which allows a higher
NDDCT inlet temperature, this leads to a higher air outlet temperature and therefore a reduction
in the required air mass flow rate. Since there is a lower air flow rate requirement, this can then be
achieved with a lower heat transfer surface area and a smaller NDDCT.
The same tower size that was determined for the indirect configuration is then used in the direct
configuration in Fig. 5 and a difference in outlet conditions is observed. The outlet conditions are
then used in the cycle analysis to determine the difference in the cycle performance.

Figure 5: Direct cooling NDDCT performance using size required for indirect cooling.

DISCUSSION

5.1

Effect of cooling configuration on cycle performance

The effect on cycle performance of the direct and indirect cooling systems are compared for the
cooling systems presented in Fig. 4 and Fig. 5 and summarized in Tab. 5. Since the indirect system

is sized to achieve the desired cooling system outlet temperature of 45oC, the cycle efficiency for
this case is as given in Tab. 1, which gives a cycle efficiency of 42.81%. For the direct case, a
lower NDDCT outlet temperature is achieved, when this value is used in the cycle analysis a cycle
efficiency of 43.12% is achieved. An increase in efficiency of 0.31%.
Cooling configuration
Compressor inlet temperature
Cycle efficiency
45.0C
42.81%
Indirect
42.2C
43.12%
Direct
Table 5. Effect of cooling configuration on cycle performance

Direct configuration improves cycle efficiency but it also reduces the cooling system complexity
and would therefore reduce cost since it does not require an IHX and water pump needed for
indirect configuration. The increase in cycle efficiency for direct configuration is in part due to the
reduced exergy losses present in the IHX for indirect configuration. Furthermore, even though
direct configuration has an overall heat transfer coefficient which is 3.501 times lower than
indirect, the mean temperature difference in the direct case is 3.583 times higher than for indirect,
and this enables a higher heat transfer rate for the same heat transfer area.
5.2

Varying tower size to achieve similar cycle performance

The sCO2 outlet temperature is lower than for the indirect case. However, it is noted that the outlet
temperature seems to be limited somewhat by the approach to the ambient temperature. Therefore,
for the sake of comparison the same methodology as for the indirect case sizing analysis was
applied to the direct case to determine the minimum tower size required to achieve the desired
outlet temperature.

Figure 6: Indirect cooling system configuration with smallest tower size to satisfy requirements given in
Tab. 2, using geometric relations in Tab. 3.

According to this modelling, looking at the bare tube area required for Fig.4 and 6, a significantly
smaller direct sCO2 cooling NDDCT (20% smaller) can achieve the same cooling performance
than for indirect cooling for the reference case considered here. Heat transfer is a product of heat
transfer coefficient (U), area and logarithmic mean temperature difference (LMTD). A smaller
tower for direct configuration is able to achieve the same heat transfer of a larger tower for indirect
configuration since even though it has a lower U, the LMTD is much higher thus resulting to a
slightly lower required area. However, this benefit of direct cooling would only be applicable for
high sCO2 inlet temperature. Thus, if the cycle operates in a lower sCO2 inlet temperature, the
relative benefit of direct configuration will be less.

CONCLUSION

Two cooling configuration are investigated, direct and indirect cooling for sCO2 Brayton cycles.
Their effect on the cycle performance is investigated and it was determined that, having same sized
tower, direct configuration for the reference point analyzed in this paper increased the efficiency
of the cycle by 0.31%. Moreover, if the cooling performance of the system is kept the same, the
resulting tower size for direct configuration would be 20% smaller than indirect configuration.
Thus for the commercial system of approximately 100MWe studied, implementing a direct cooling
configuration is preferred due to the improvements in cycle performance. However, as mentioned,
the values used in this study assumes a large-scale, commercial system of approximately 100MWe
net power generation only. The extent of effect of the cooling configuration could be different as
the net power generation of the cycle change thus further investigation is required. Also, a more
in-depth analysis is needed to fully compare the two cooling system and their effect in the total
running and capital cost of CSP.

ACKNOWLEDGEMENTS
This research was performed as part of the Australian Solar Thermal Research Initiative (ASTRI),
a project supported by the Australian Government, through the Australian Renewable Energy
Agency (ARENA)

REFERENCES
Anton Moisseytsev, J. S. (2014). Investigation of a dry air cooling option for an sCO2 cycle. The 4th
International Symposium - Supercritical CO2 Power Cycles.
Conboy, T., Carlson, M., & Rochau, G. (2015). Dry-Cooled Supercritical CO2 Power for Advanced Nuclear
Reactors. Journal of Engineering for Gas Turbines and Power, 137(1), 012901.
Dostal, V. (2004). A supercritical carbon dioxide cycle for next generation nuclear reactors. Massachusetts
Institute of Technology.
Duniam, S. (2015). Design Optimisation Of An Australian EGS Power Plant Using a Natural Draft Dry
Cooling Tower. (Master of Philosophy), University of Queensland.
Dyreby, J., Klein, S., Nellis, G., & Reindl, D. (2014). Design Considerations for Supercritical Carbon Dioxide
Brayton Cycles With Recompression. Journal of Engineering for Gas Turbines and Power, 136(10),
101701-101701. doi: 10.1115/1.4027936
Dyreby, J. J., Klein, S. A., Nellis, G. F., & Reindl, D. T. (2013). Modeling Off-Design and Part-Load
Performance of Supercritical Carbon Dioxide Power Cycles. Paper presented at the ASME Turbo
Expo 2013: Turbine Technical Conference and Exposition.
Fang, X., Xu, Y., Su, X., & Shi, R. (2012). Pressure drop and friction factor correlations of supercritical flow.
Nuclear Engineering and Design.
Fernndez, I., & Sedano, L. (2013). Design analysis of a leadlithium/supercritical CO2 Printed Circuit Heat
Exchanger for primary power recovery. Fusion Engineering and Design.
Krger, D. G. (2004). Air-Cooled Heat Exchangers and Cooling Towers (Vol. 2). Tulsa, Okl: Pennwell Corp.
Litwin, R. Z., Zillmer, A. J., Hoffman, N. J., von Arx, A. V., & Wait, D. (2010). Supercritical CO2 turbine for
use in solar power plants: Google Patents.

Neises, T., & Turchi, C. (2014). A comparison of supercritical carbon dioxide power cycle configurations
with an emphasis on CSP applications. Energy Procedia, 49, 1187-1196.
Padilla, R. V., Soo Too, Y. C., Benito, R., & Stein, W. (2015). Exergetic analysis of supercritical CO2 Brayton
cycles integrated with solar central receivers. Applied Energy, 148, 348-365. doi:
http://dx.doi.org/10.1016/j.apenergy.2015.03.090
Singh, R., Rowlands, A., & Ventura, C. (2014). Impacts of Collector Receiver Volume on Dynamic
Performance of a Direct-Heated Supercritical-CO2 Closed Brayton Cycle in a Solar Thermal Power
Plant. Paper presented at the Solar 2014: Scientific and Research Conference and Expo.
SPX. (2015). Cooling tower fundamentals.
Retrieved 7/16/2015, 2015, from
http://spxcooling.com/library/detail/cooling-tower-fundamentals
Tanimizu, K., & Sadr, R. (2015). Experimental investigation of buoyancy effects on convection heat transfer
of supercritical CO2 flow in a horizontal tube. Heat and Mass Transfer. doi: 10.1007/s00231-0151580-9
Turchi, C., Ma, Z., & Dyreby, J. (2009). Supercritical CO2 for application in concentrating solar power
systems. Paper presented at the SCCO2 Power Cycle Symposium, RPI, Troy, NY.
Turchi, C. S., Ma, Z., Neises, T. W., & Wagner, M. J. (2013). Thermodynamic Study of Advanced
Supercritical Carbon Dioxide Power Cycles for Concentrating Solar Power Systems. Journal of Solar
Energy Engineering, 135(4), 041007-041007. doi: 10.1115/1.4024030
Viebahn, P., Lechon, Y., & Trieb, F. (2011). The potential role of concentrated solar power (CSP) in Africa
and EuropeA dynamic assessment of technology development, cost development and life cycle
inventories
until
2050.
Energy
Policy,
39(8),
4420-4430.
doi:
http://dx.doi.org/10.1016/j.enpol.2010.09.026
Williges, K., Lilliestam, J., & Patt, A. (2010). Making concentrated solar power competitive with coal: The
costs of a European feed-in tariff. Energy Policy, 38(6), 3089-3097. doi:
http://dx.doi.org/10.1016/j.enpol.2010.01.049

Das könnte Ihnen auch gefallen