www.cet-journal.

com Page 1 Chemical Engineering & Technology

Hybrid cooling towers in a free cooling application - modelling and field

measurement verification
1 1 1
Philipp Puls *, Christopher Lange , Richard Öchsner
1
Fraunhofer-Institute for Integrated Systems and Device Technology (IISB), Schottkystr. 10, 91058 Erlangen,
Germany
*Correspondence: Philipp Puls (E-mail: [email protected]), Fraunhofer-Institute for Integrated
Systems and Device Technology (IISB), Schottkystr. 10, 91058 Erlangen, Germany
Abstract
In this paper, the energy-saving potential of free cooling in chilled water systems is investigated. As a
preliminary study, simulation was used to quantify the performance of free cooling in a given reference system.
In order to reduce computational effort for the simulation of long time periods, simplified models for dry and
evaporative cooling were developed. The simulation results indicate that free cooling reduces the electric
energy demand for cold supply almost by half. In a follow-up investigation, free cooling was implemented in
the reference system. The actual energy-savings showed good agreements with the simulation forecasts and
confirmed the feasibility of the concept.
Keywords: Energy efficiency, cooling system, free cooling, simulation, hybrid cooling tower
1 Introduction and motivation
Chillers account for a major part of the energy costs in facilities with cooling demand. In the winter
periods, the demand for cold energy is often low because of diminished needs for air conditioning. As
a consequence, the chillers are operated at part loads because they are usually dimensioned for the
peak loads that occur in the warmer period of the year. Load differences may also occur between
daytime and nighttime: The maximum cooling load is typically expected in the afternoon hours [1],
when ambient temperatures are high. The part load operation of chillers causes low efficiencies and
therefore unnecessary high chilled water supply costs [2]. An approach to avoid the operation of
chillers in low-demand periods is the use of free cooling. Free cooling means the exploitation of low
ambient temperatures to provide chilled water without the use of chillers. In Germany, the
temperatures needed for many server and process cooling applications are significantly higher than
the average annual ambient temperature. In those applications, long free cooling periods can be
achieved and potential energy-savings are high.
In this paper, the feasibility of free cooling was investigated for the chilled water supply system at the
Fraunhofer-Institute for Integrated Systems and Device Technology (IISB). The reference system has
a high cooling demand throughout the year and serves as a demonstration platform for energy
efficiency measures in an industry-oriented environment. In order to make predictions concerning

Received: December 10, 2018; revised: February 07, 2019; accepted: May 08, 2019

This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which
may lead to differences between this version and the final Version of Record (VOR). This
work is currently citable by using the Digital Object Identifier (DOI) given below. The final
VoR will be published online in Early View as soon as possible and may be different to this
Accepted Article as a result of editing. Readers should obtain the final VoR from the
journal website shown below when it is published to ensure accuracy of information. The
authors are responsible for the content of this Accepted Article.

To be cited as: Chem. Eng. Technol. 10.1002/ceat.201800712

Link to final VoR: https://doi.org/10.1002/ceat.201800712

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 2 Chemical Engineering & Technology

the performance of closed circuit cooling towers in a free cooling application, the reference system
was reproduced in a simulation. As numerical methods for the representation of the evaporative
cooling performance are very complex [3] and do not necessarily achieve a sufficient quantitative
prediction [4], a simple model was developed, that is suitable for use in simulations with long time
periods. The free cooling concept was implemented subsequently in order to validate the model
design.

2 System architecture
In most chilled water systems, chillers are used to cool the water in the consumer circuit. The waste
heat of the consumer circuit is transferred to the refrigerant circuit in the evaporator of the chiller. In
order to remove this excess energy from the refrigerant circuit, the condenser of the chiller needs to
be cooled. This can either be achieved by using air or water cooled condensers. Independent of the
condenser cooling system, the waste heat is always transferred to the surrounding air. The recooling
demand 𝑄̇rc is defined as the sum of the original cooling load 𝑄̇c and the electric power
consumption 𝑃el of the refrigerant compressor [5].
𝑄̇rc = 𝑄̇c + 𝑃el 1
Fig. 1 shows the topology of a chilled water system with a compressor chiller and a water cooled
condenser.

Figure 1: Schematic diagram of the energy flux and the components in a chilled water system
In fig. 1 the condenser is cooled via a separate cooling water circuit. This circuit can be connected to
a dry cooler, an open cooling tower or a closed circuit cooling tower (also referred to as hybrid
cooling tower). At the reference site, the condenser cooling system contains a single closed circuit
cooling tower as pictured in fig. 2.

Figure 2: Schematic diagram of a closed circuit cooling tower in the reference system
In closed circuit cooling towers, the cooling water and the ambient air are separated by the wall of a
heat exchanger. The present heat exchanger is a counter current shell-and-tube design. Hybrid
coolers usually can be operated in dry or in wet mode. In the dry mode, the cooling water in the
tubes of the heat exchanger is cooled via the passing ambient air. In the wet mode, water is sprayed
on the tube bundle of the heat exchanger. Due to the high airflow, the water evaporates and
saturates the passing air with water vapor. For the evaporation of water, large amounts of energy are
needed, which are mainly taken from the surrounding air. This reduces the temperature of the
passing air and enhances the cooling effect. The minimum achievable temperature is referred to as
wet bulb temperature. It is reached when the air is saturated with water vapor. In the present
system the spray nozzles are placed directly above the closed tube bundles. Therefore, adiabatic
cooling of the air before entering the tube bundle is not an option. As the reference system in this
study exclusively uses hybrid cooling towers, the further investigations will be restricted to this
technology.
In the illustrated design, evaporative cooling (wet mode) can be initiated via activation of the spray
water pump. However, at low ambient temperature levels, the dry mode is usually sufficient to meet
the recooling demand of the chiller. Independent of the spray water pump, the fan can be operated
in different stages in order to control the mass flow of air through the cooling tower (see supporting
information). In order to prevent the cooling water from freezing at sub-zero temperatures, water-
ethylene mixtures are used. If the ambient temperatures or the recooling demands are low, the
spray water reservoir is emptied and the wet mode is turned off manually.

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 3 Chemical Engineering & Technology

3 Free cooling potential

In this study, the use of ambient air for cooling without a chiller is referred to as free cooling. When
free cooling is used, the chiller is switched off and the electric power consumption of the refrigerant
compressor is saved. The recooling demand then equals the cooling demand of the users in the
chilled water circuit. This benefit has to be contrasted with the additional electric energy the cooling
tower needs to supply chilled water with a lower temperature level than that of the cooling water.
In order to determine the potential of free cooling for a specific location, the temperature levels of
the ambient air and the chilled water have to be considered. In Germany, the average annual
ambient temperature level usually lies around nine degrees Celsius, depending on the region. The
temperature level of the chilled water in the reference system at the IISB is about twelve degrees
Celsius. In order to make a first estimation on the possible operation time of free cooling, a simplified
approach was chosen: As a rule of thumb, dry cooling systems need to have a temperature
difference of four to seven degrees Celsius [5] between the cooling water and the ambient air in
order to be economically feasible. Lower temperature differences require larger surface areas for
heat exchange, which is equal to higher investment costs for the cooling tower. Fig. 3 shows the
temperature duration curves of the location Nuremberg. The dashed lines indicate the time period,
in which the ambient temperature level is below the required temperature level for free cooling as
defined by the given rule of thumb.

Figure 3: Cumulated time period of ambient temperature levels below a threshold value of 5 °C for
the region of Nuremberg
Fig. 3 shows that free cooling is possible between 2000 and 3000 hours per year when an ambient
temperature of five degrees Celsius is assumed as a threshold value.
At the site of the reference facility there is a separate chilled water system for air conditioning, which
is exclusively operated in the summer months. This system has two hybrid cooling towers that are
available for free cooling over the course of the winter. As the cooling towers were not designed for
the task of free cooling, the maximum ambient temperature for free cooling may deviate from the
given threshold value, based on the rule of thumb. It should be noted that the available cooling
towers cannot use evaporative cooling in the free cooling mode, as the circuit for the spray water
pump cannot be drained automatically and is emptied before the start of the winter period. By using
evaporative cooling, the possible free cooling period could be extended significantly, as the needed
difference between the wet bulb temperature of the air and the temperature of the consumer circuit
can be as low as 3 °C with appropriate cooling equipment [6]. Possible ways of integrating a cooler
into a chilled water system have been described in the literature [6, 7, 8]. In the simplest case, the
chillers are bypassed and the chilled water is redirected to the cooling tower of the free cooling
system. Since the implementation is associated with a high technical effort, simulation is used to
examine the economic feasibility of the concept in advance.

4 Modelling of components
As low ambient temperatures can occur over a large fraction of the year, a system model must be
derived that is capable of taking into account long time periods with acceptable computational
effort. The relevant components are represented by individual models, which are controlled by a
dedicated operating strategy. The main components are the chiller, the cooling tower, the main
pumps for the chilled water and the cooling water, as well as the free cooling unit, which is derived
from the model of the hybrid cooling tower. An energy-based approach was chosen. This means that
the components are controlled to meet the cooling demand in the chilled water system as well as the
recooling demand of the chiller. Different simulation scenarios are evaluated with respect to the

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 4 Chemical Engineering & Technology

amount of electric energy that was needed to meet the cooling demand of the consumers. The
quotient of the delivered cold energy and the needed electric energy is referred to as efficiency ratio
of cold supply.

4.1 Chiller
For the chiller, a simple data-based model was derived: The energy efficiency ratio 𝜖 of a screw
compressor chiller is represented with a 2nd degree polynomial function, which correlates the energy
efficiency ratio 𝜖 to the cooling load 𝑄̇c as shown in [9] (see supporting information). The model
equation is given by equ. 2.
2
𝜖 = 𝑎 ∙ 𝑄̇c + 𝑏 ∙ 𝑄̇c + 𝑐 2
with
𝑎 = −3.9 ∙ 10−5 𝑊 −2
𝑏 = 0.026 𝑊 −1
𝑐 = 0.47
In the reference system, the calculation of electric power consumption of the chiller at a given
cooling load is in good agreement with the measured values. The mean absolute percentage error
(MAPE) of the chiller model usually ranges between one and two pecent for a data set of a whole
year in minute resolution.

4.2 Pumps
In the original state of the reference system the pumps aren’t speed-controlled, so that slight
changes in volume flow and electric power consumption are only due to pressure differences in the
consumer circuits. For the representation of the pumps, the fluid flows and the electric power
consumptions were measured and implemented in the simulation in the form of fixed values.
Deviations from the measured values are considered minimal, as most of the consumers exhibit a
steady cooling demand.

4.3 Recooler
For the simulation of the cooling towers, a grey-box model was developed. The model uses empiric
correlations for the calculation of heat transfer between the passing air and the cooling water in
dependence of the Reynolds and the Nusselt numbers. For the representation of evaporative cooling,
the heat exchanger is separated into segments. The number of segments is used to match the model
with the measurement data.

4.3.1. Dry cooling mode

In order to determine the Reynolds numbers 𝑅𝑒 for the airflow through the tube bundle of the heat
exchanger, the individual air velocities of the different fan stages were measured with a hot-wire
anemometer. With knowledge of the Reynolds number 𝑅𝑒 in the tube bundle, the Nusselt
number 𝑁𝑢 for the heat transfer between the air and the wall of the pipe bundle is calculated
according to Gnielinski [10]. Subsequently the heat transfer coefficient 𝛼air on the outside of the tube
bundle is calculated in dependence of the thermal conductivity 𝜆air of the air and the characteristic
length 𝐿 (see equ. 3). The latter is given by the so-called streamed length, which is defined as one
half of the circumference of a single pipe in the bundle with the outer diameter 𝑑o [10].
𝜆air 2∙𝜆air
𝛼air = 𝑁𝑢 ∙ 𝐿
= 𝑁𝑢 ∙ 𝜋∙𝑑o
3

Subsequently, the heat transfer coefficient 𝛼we for the water-ethylene mixture on the inside of the
tube bundle is determined. Analogous to the airflow, the flow velocity of the cooling water was

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 5 Chemical Engineering & Technology

measured. As the pump in the condenser cooling system is not speed-controlled, the flow velocity is
assumed to be constant throughout the year. The Reynolds number in the pipes is calculated after
Gnielinski [11]. With help of the Reynolds number, the Nusselt number is calculated. The heat
transfer coefficient 𝛼we on the inside of the tubes is determined in accordance with equ. 3, whereby
the characteristic length 𝐿 is represented by the inner diameter of the pipe 𝑑i .
With the heat transfer coefficients on the inside and the outside of the heat exchanger, the overall
heat transmission coefficient 𝑘 is determined via equ. 4 [12]. Beside the heat transfer coefficients,
the thickness of the pipe wall 𝛿, the thermal conductivity 𝜆 as well as the respective heat
transmission surfaces 𝐴 on the inner (i) and the outer (o) side of the pipe wall are needed. 𝐴m
describes the mean surface area for heat conduction, as the cross-sectional area of the pipe wall
decreases from the outside to the inside [13].
1
𝑘= 𝐴i 𝛿∙𝐴i 𝐴i 4
+ +
𝛼we ∙𝐴i 𝜆∙𝐴m 𝛼air ∙𝐴o

In equ. 4, the heat transmission coefficient 𝑘 is given as an average value across the heat exchanger
of the cooling tower. With the knowledge of 𝑘, the recooling capacity 𝑄̇rc is calculated via equ. 5 [12].
𝑄̇rc = 𝑘 ∙ 𝐴i ∙ ∆𝑇m,log 5
In order to take into account the temperature profiles across the countercurrent heat exchanger, the
mean logarithmic temperature difference ∆𝑇m,log is used [12]. The temperatures 𝑇air and 𝑇we
represent the temperatures of the passing air and the cooling water at the inlet (in) and the outlet
(ex) of the heat exchanger.
in −𝑇 ex +𝑇 ex −𝑇 in
𝑇we
∆𝑇m,log = we air
ex
𝑇 −𝑇air
air
in 6
ln we
in ex
𝑇we −𝑇air

Equ. 6 requires the temperature of the passing air 𝑇air ex

at the outlet of the heat exchanger. In contrast
to 𝑇we , which is usually set to a fixed value, this temperature is not initially known and must be
ex

estimated. Therefore, the calculated recooling capacity 𝑄̇rc in equ. 5 is not accurate. In order to
determine the recooling capacity with higher accuracy, an iterative procedure is required: With the
calculated recooling capacity, the outlet temperature is determined backwards via equ. 7 [12]. For
this, the total mass flow of air 𝑚̇air for every fan stage and the isobaric heat capacity 𝑐p,air of air are
needed.
𝑄̇rc
𝑇air
ex
= 𝑚̇ + 𝑇air
in
7
air ∙𝑐p,air

The calculated outlet temperature of the air is used as new estimated value and the iteration
procedure starts anew until the difference between the two temperatures is below a predefined
error limit. In each iteration step, all temperature dependent material values (air, water-ethylene and
the pipes) have to be determined with help of the mean average temperature between the inlet and
the outlet of the respective medium. The presented method is suitable for the determination of the
dry cooling capacity of the cooling tower. It does not yet comprise evaporative cooling, which is
needed for the representation of the electric power consumption of the regular chiller/cooler
configuration.

4.3.2. Wet cooling mode

A new approach was developed to calculate the wet recooling capacity: The heat exchanger of the
cooling tower is separated into uniform dry segments, which are calculated with the procedure given
above. However, in between the dry segments, the air is assumed to be saturated with water vapor
due to irrigation of the heat exchanger. This cools the passing air until it reaches the wet bulb
temperature 𝑇wb . As the air is assumed to be saturated with water vapor, the water content 𝑋wb

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 6 Chemical Engineering & Technology

before each dry segment is at its maximum value and the relative humidity 𝜑wb is equal to one. This
increases the recooling capacity of the next dry segment, where the temperature of the air increases
and the relative humidity decreases again. The intensity of interaction between dry and wet
segments can be adjusted via the discretization parameter 𝑛.
In the following example the model for the heat exchanger starts and ends with a dry segment. The
total number of wet segments is therefore 𝑛, whereas the number of dry segments is 𝑛 + 1. Fig. 4
shows the concept of discretization for the proposed method.

Figure 4: Discretization of the heat exchanger for the representation of evaporative cooling in a
hybrid cooling tower
For the calculation of the wet bulb temperature an iterative procedure was used [4]. The number of
iterations for a dry segment usually ranges between 5 and 20. With knowledge of the mass flow of
air, the amount of evaporated water is calculated. The costs of treated water are taken into
consideration in order to determine the operating costs of the cooling tower in the simulations.
The proposed method presents a simple instrument to model the energy performance of a hybrid
cooler with adequate calculation effort. In order to determine the number of dry segments 𝑛 + 1 for
a given cooling tower, the measured values of the recooling capacity 𝑄̇rc and the electric power
consumption are compared with the simulated equivalents. The parameter 𝑛 is then varied until the
model matches the measurement data. In the present case 𝑛 = 3 showed the highest agreement
with the measurements. The maximum absolute error between the measured and the modelled
energy demand of the free cooling system was 3 % (see supporting information). In the current
condition of the reference system free cooling can solely be operated in dry mode. The model for
evaporative cooling is needed to determine the energy demand of the condenser cooling system
during normal operation of the chiller. When free cooling is active, the electric power consumption
of the regular system must be calculated in order to compare the energy performance of both
scenarios.
The presented model was used to calculate the recooling capacities for the two scenarios of cooling
the condenser of the chiller and free cooling. Fig. 5 holds the simulation results for the condenser
cooling application in dependence of the inlet temperature and the relative humidity of the passing
air. The temperature of the cooling water is 32 °C and the mass flow rate of the passing air is
-1
16 kg s .

Figure 5: Recooling capacities for dry (a) and wet (b) operation of the hybrid cooling tower
The diagram shows that the recooling capacity is increased significantly when the wet mode of the
cooling tower is engaged. In contrast to the dry mode, there is a noticeable dependence on the
relative humidity of the ambient air.
The free cooling system comprises two separate cooling towers with three fan stages 𝐿0 − 𝐿2 each.
Fig. 6 shows the cooling capacities (a) and the energy efficiency ratios (b) of the free cooling system
for all possible fan stage combinations in dry mode of operation. The performance of the wet mode
was not investigated, as the spray water reservoir is emptied at low temperatures in the reference
system.

Figure 6: Calculated cooling capacities (a) and efficiencies (b) of the free cooling system in depence
of the ambient temperature.

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 7 Chemical Engineering & Technology

The cooling capacities increase when the ambient temperature decreases. As the electric energy
demands for the individual fan stages are constant, the efficiency of cold supply increases
proportionally. The operation of both cooling towers on fan stage one (𝐿1𝐿1) represents the most
efficient cooling mode. However, the highest cooling capacities are achieved with both cooling
towers set on fan stage two (𝐿2𝐿2).

4.4 Simulation results

For the investigation of the energy saving potential of free cooling, a simulation of the reference
system over a time period of 24 weeks ranging from November 2015 to April 2016 was conducted.
The measured values of the ambient temperature, the relative humidity and the cooling load were
used as input data in a temporal resolution of one minute. The mean average ambient temperature
of the time range was 4.4 °C and the mean average cooling load was 126 kW. Free cooling was used
in dry mode only in order to avoid freezing of spray water in the piping system. In contrast to fig. 3,
the unit was engaged when the ambient temperature dropped below four instead of five degrees
Celsius. The threshold value was lowered, because the simulations showed that the cooling loads
were sometimes higher than the maximum cooling capacity of the free cooling system at five
degrees Celsius. At higher temperatures, the regular chiller/cooler configuration (regular system) was
operated. When free cooling was active, the mean average cooling load was 112 kW and the mean
average ambient temperature was −0.06 °C. Fig. 7 shows the results of the regarded time period.

Figure 7: Operating resources (a) and efficiencies (b) of the regular chiller/recooler configuration in
comparison to the configuration with free cooling
The simulation results imply that free cooling yields significant energy savings. The energy demand of
cold supply is 22 % lower in comparison to the regular system configuration in the given time period.
This translates to about 12 % electrical energy savings in a whole year, as free cooling cannot be used
in the summer periods. At times, where free cooling is actually in operation, the efficiency of cold
supply is raised by 47 % in comparison to the regular system.

5 Implementation of free cooling in the reference system

Fig. 8 shows the reference system before and after implementation of the free cooling unit. The
latter comprises two hybrid cooling towers (A), which were integrated into the chilled water circuit of
the reference system. A separate heat exchanger (B) for fluid separation was installed, as the
consumers do not accept ethylene as a cooling medium and the cooling towers need to be operated
with a water-ethylene mixture for frost protection.

Figure 8: Schematic diagram of the reference system before (a) and after (b) integration of free
cooling
In the course of the installation, three motor-driven valves were implemented. These are needed to
close the hydraulic circuits of the chillers, when free cooling is in operation. Likewise, the free cooling
system can be closed, when ambient temperatures are too high and the chillers are running. With
the given topology, any possible combination of chillers and free cooling can be realized.
The free cooling system was taken into operation in the winter season of 2016/17. In order to
prevent possible damage from malfunction, the operation was restricted to consolidated time
frames. Overall, 15 field tests were conducted with a total operation time of 111 h. The energy
efficiency ratio of the free cooling system was measured and compared to the calculated energy
efficiency ratio of the regular system, comprising the chillers and the related cooling tower. The
electric energy consumptions of the main pumps and periphery aggregates (such as valves) were also

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 8 Chemical Engineering & Technology

taken into consideration, so that the energy efficiency ratio is related to the system. Fig. 9 holds the
results for the first winter season.

Figure 9: Comparison of the energy efficiency ratios of the free cooling and the regular cooling
system
The average ambient temperature of all conducted field tests was −0.24 °C and the average cooling
load was 112 kW. Fig. 9 shows that the energy efficiency ratio was doubled with free cooling in
comparison to the regular system configuration. This reveals a significant energy saving potential of
free cooling in the regarded reference system and confirms the simulation results in section 4.4,
where an energy saving potential of 47 % was found. The range of energy efficiency ratios for the
free cooling system in fig. 9 is a direct consequence of differing cooling loads and ambient
temperatures (see fig. 6). The highest efficiency was achieved on January 19th, where the mean
ambient temperature was -3.7 °C. In contrast, the lowest efficiency was found on January 12th with a
mean ambient temperature of 2.9 °C.
Fig. 10 shows the field test from 11th of March in 2017, where free cooling was operated between
zero and six o’clock.

Figure 10: Cooling load and electric energy consumptions of free cooling and regular chiller/recooler
configuration on 11th of March in 2017
Due to the change of the cold supply system, there is an impact on the measured cooling load when
free cooling is activated and again when it is shut off at six o’clock. However, the accompanying peak
loads do not compromise the stability of the reference system and also occur during a weekly change
of the two redundant chillers. The electric power consumption of the regular system configuration is
about 70 kW. With the activation of free cooling, the overall electric power consumption for cooling
drops down to about 25 kW. At about 0:30 a higher fan stage is needed in order to meet the cooling
demand of the consumers. This causes an increased electric power consumption of about 40 kW. In
the further progression, the power consumption of the free cooling system drops down to 25 kW and
then increases to 40 kW again. This behavior is due to accommodation to the cooling load in the
chilled water system. Overall, the electric energy demand of the free cooling system is significantly
lower than that of the regular system. In the presented field test, free cooling was stopped
intentionally at 6 o’clock, although it would have been capable of further operation.

6 Summary and outlook

In this study, a simple approach for the simulation of cooling towers was described. The presented
method requires minimal computational effort and is suited for the utilization in long-term
simulations of energy systems. The simulation results are in good agreement with the
measurements: the forecast of 47 % energy savings was slightly exceeded with the implementation
of the free cooling system, which yielded about 50 % energy savings. The deviation between the
simulation forecast and the measurements may arise from the differences in cooling load and
ambient temperature between the two evaluation periods.
The simulation was initially set up as a non-invasive means to investigate the potential energy
savings of free cooling. As the simulation results showed promising energy savings, free cooling was
implemented in the reference system and is now in an automated operation. In the described
implementation, two external cooling towers are used for free cooling. By additional use of the third
cooling tower for condenser cooling, the heat transfer surface could be increased and the threshold
temperature for free cooling could be raised. This would extend the free cooling period and further

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 9 Chemical Engineering & Technology

reduce electric energy consumption of the reference system. Another option of enhancing the free
cooling performance is the use of evaporative cooling at low ambient temperatures. This would also
increase the threshold temperature for free cooling and extend the possible free cooling period.
However, the present cooling towers require a manual routine to drain the spray water circuit. This
prohibits any use of evaporative cooling at low temperatures in the current system configuration.
Further energy savings can be achieved by combining free cooling with cold storage. Excess cold
energy from free cooling in the night times can be stored in a chilled water tank and released during
the day in order to decrease the cooling load or even turn off the chiller during the time period with
the highest ambient temperatures. Further studies will show how the efficiency of cold supply can be
optimized with a combination of free cooling and cold storage.

Acknowledgement
The authors would like to thank the Bavarian Ministry of Economic Affairs, Regional Development
and Energy for financial support of the project SEEDs.
The authors have declared no conflict of interest.

Symbols used
𝐴 [m²] surface area
𝑐p [J kg-1 K-1 ] isobaric heat capacity
𝑑 [m] diameter
𝑘 [W m-2 K-1 ] heat transmission coefficient
𝐿 [m] characteristic length
𝑄̇ [W] thermal capacity, thermal load
𝑚̇ [kg s-1 ] mass flow
𝑁𝑢 [−] Nusselt number
𝑃 [W] power input
𝑅𝑒 [−] Reynolds number
𝑇 [°C] temperature
𝑋 − water content

Greek characters
𝛼 [W m-2 K-1 ] heat transfer coefficient
∆ [−] difference
𝛿 [m] wall thickness
𝜖 [−] energy efficiency ratio
𝜆 [W m-1 K-1 ] heat conductivity
𝜑 [−] relative humidity

Sub- and superscripts

1-2 indication of element 1-2
air air
c cooling
el electric
ex outlet
i inner
in inlet
log logarithmic
m mean

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 10 Chemical Engineering & Technology

o outer
rc recooling
wb wet bulb
we water-ethylene

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 11 Chemical Engineering & Technology

References
[1] S. M. Hasnain, Energy Convers. Manage. 1998, 39(11), 1139-1153. DOI: 10.1016/S0196-
8904(98)00024-7.
[2] Q. Zhang, D. Yan, J. An, T. Hong, W. Tian, K. Sun, Energy and Buildings 2017, 139, 407-416.
DOI: 10.1016/j.enbuild.2017.01.044.
[3] W. Polifke, J. Kopitz, Wärmeübertragung – Grundlagen, analytische und numerische
Methoden, Person Education Deutschland GmbH , München 2009.
[4] M. Zeller, U. Busweiler, in VDI Heat Atlas, 2nd ed., (Ed: VDI-GVC), Springer-Verlag Berlin
Heidelberg, Düsseldorf 2010, 1343-1361.
[5] T. Urbaneck, Kältespeicher – Grundlagen, Technik, Anwendungen, Oldenbourg
Wissenschaftsverlag GmbH, München 2012.
[6] B. Costelloe, D. Finn, Energy and Buildings 2003, 35, 573-591. DOI: 10.1016/S0378-
7788(02)00161-5.
[7] R. Jaramillo, J. E. Braun, W. T. Horton, in Proc. of the 3rd Int. High Performance Buildings
Conf., West Lafayette, Indiana. June 2014, 1-10.
[8] H. Zhang, S. Shao, H. Xu, H. Zou, C. Tian, Renewable Sustainable Energy Rev. 2014, 35, 171-
182. DOI: 10.1016/j.rser.2014.04.017.
[9] P. Puls P, S. Linhardt, R. Öchsner, in NEIS Conf. 2016 Nachhaltige Energieversorgung und
Integration von Speichern (Ed: D. Schulz,), Springer Vieweg, Hamburg 2017, 123-129. DOI:
10.1007/978-3-658-15029-7.
[10] V. Gnielinski, in VDI Heat Atlas, 2nd ed., (Ed: VDI-GVC), Springer-Verlag Berlin Heidelberg,
Düsseldorf 2010, 725-729.
[11] V. Gnielinski, in VDI Heat Atlas, 2nd ed., (Ed: VDI-GVC), Springer-Verlag Berlin Heidelberg,
Düsseldorf 2010, 693-699.
[12] H. D. Baehr, K. Stephan, Wärme- und Stoffübertragung, 6th ed., Springer-Verlag Berlin
Heidelberg, 2008.
[13] W. Roetzel, B. Spang, in VDI Heat Atlas, 2nd ed., (Ed: VDI-GVC), Springer-Verlag Berlin
Heidelberg, Düsseldorf 2010, 67-74.

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 12 Chemical Engineering & Technology

Figures

Figure 1: Schematic diagram of the energy flux and the components in a chilled water system 2

Figure 2: Schematic diagram of a closed circuit cooling tower in the reference system 2

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 13 Chemical Engineering & Technology

Figure 3: Cumulated time period of ambient temperature levels below a threshold value of 5 °C for
the region of Nuremberg 3

Figure 4: Discretization of the heat exchanger for the representation of evaporative cooling in a
hybrid cooling tower 6

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 14 Chemical Engineering & Technology

Figure 5: Recooling capacities for dry (a) and wet (b) operation of the hybrid cooling tower 6

Figure 6: Calculated cooling capacities (a) and efficiencies (b) of the free cooling system in depence
of the ambient temperature. 6

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 15 Chemical Engineering & Technology

Figure 7: Operating resources (a) and efficiencies (b) of the regular chiller/recooler configuration in
comparison to the configuration with free cooling 7

Figure 8: Schematic diagram of the reference system before (a) and after (b) integration of free
cooling 7

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 16 Chemical Engineering & Technology

Figure 9: Comparison of the energy efficiency ratios of the free cooling and the regular cooling
system 8

Figure 10: Cooling load and electric energy consumptions of free cooling and regular chiller/recooler
configuration on 11th of March in 2017 8

This article is protected by copyright. All rights reserved.

www.cet-journal.com Page 17 Chemical Engineering & Technology

Table of contents
Research Article: This article Energy performance of hybrid
investigates the use of free cooling towers in a free cooling
cooling in chilled water systems. It application - modelling and field
introduces a new model for the measurement verification
simulation of hybrid cooling towers P. Puls, C. Lange, R. Öchsner
and presents experimental results Chem. Eng. Technol. 20XX, XX
of the integration of free cooling in (X),
a reference system. The identified xxxx…xxxx
energy savings with free cooling
were significant and showed good
agreement with the forecasts of
the simulations.

This article is protected by copyright. All rights reserved.