Es

SLa

b

a

ARRAA

KIIECD

1

tqts2bccairprc

t

h0

Energy and Buildings 121 (2016) 130–138

Contents lists available at ScienceDirect

Energy and Buildings

j ourna l ho me pa g e: www.elsev ier .com/ locate /enbui ld

xperimental analysis of a cross flow indirect evaporative coolingystem

tefano De Antonellis a,∗, Cesare Maria Joppolo a, Paolo Liberati b, Samanta Milani a,uca Molinaroli a

Dipartimento di Energia, Politecnico di Milano, Via Lambruschini, 4, 20156 Milan, ItalyRecuperator S.p.A., Via Valfurva, 13, 20027 Rescaldina (MI), Italy

r t i c l e i n f o

rticle history:eceived 22 December 2015eceived in revised form 15 February 2016ccepted 29 March 2016vailable online 30 March 2016

eywords:ndirect evaporative cooling

a b s t r a c t

Indirect evaporative cooling is an effective way to increase energy efficiency of air conditioning systems.This technology is particularly suitable for data centers applications, where the indoor temperature canbe higher than the one adopted in residential and commercial buildings. In this work an indirect evapo-rative cooling system based on a cross flow heat exchanger has been widely tested. The system has beendesigned in order to minimize water consumption, with water mass flow rate between 0.4% and 4% of thesecondary air one. On the whole, 112 experiments have been carried out in different working conditionsof data centers. The effects of variation of water flow rate, humidification nozzles setup and secondary air

ECxperimental testross flowata center

temperature, humidity and flow rate have been widely investigated. Results put in evidence that perfor-mance is slightly dependent on nozzles number and size but it is strongly influenced by the water flowrate. In addition, nozzles in counter flow arrangement perform better than in parallel flow configuration.Depending on working conditions and equipment setup, the wet bulb effectiveness varies between 50%and 85%.

© 2016 Elsevier B.V. All rights reserved.

. Introduction

In the last 15 years data centers, which consist of specific facili-ies containing ICT devices as well as cooling and power equipment,uickly increased in number and size [1]. As a result, in 2010he total electricity used by data centers was 1.3% of world con-umption and, in particular, in the US it increases from 0.13% in005 to 2% in 2010 [2]. Heat fluxes dissipated in data centers varyetween 0.5 kW m−2 to 10 kW m−2: as a consequence electricityonsumption for cooling is relevant and it can reach 50% of the totalonsumption [1,2]. Therefore, at present design, manufacturingnd management of cooling system is one of the most challeng-ng aspects of data centers. Many research activities deal with theeduction of primary energy consumption in data centers, and inarticular with energy efficiency measures and the integration ofenewable sources [3], with waste heat recovery [1] and with free

ooling options [2].

Recently ASHRAE updated the thermal guidelines for data cen-ers [4], suggesting appropriate temperature and humidity ranges

∗ Corresponding author.E-mail address: stefano.deantonellis@polimi.it (S. De Antonellis).

ttp://dx.doi.org/10.1016/j.enbuild.2016.03.076378-7788/© 2016 Elsevier B.V. All rights reserved.

for IT equipment operation. On the whole 4 classes are defined,namely A1, A2, A3 and A4. Classes A1 and A2 are identical to class1 and 2 reported in the previous edition of guidelines [5], insteadclasses A3 and A4 are new and represent an extension of suggestedtemperature and humidity limits. More precisely, in class A3 andA4 the maximum allowable temperature is respectively 40 ◦C and45 ◦C. The appropriate class is selected by data center’ operators inorder to achieve desired energy savings and IT equipment reliabil-ity.

It is well known that an increase in the indoor data centertemperature leads to an increase in free cooling working hoursand, therefore, to relevant energy savings [2,6]. As a consequence,research activities about free cooling technologies and energy effi-ciency measures in data centers are rapidly increasing. One of themost promising technologies is based on the indirect evaporativecooling (IEC) principle. In IEC systems an air stream is first cooledthrough an adiabatic humidifier, and then it is used to cool a work-ing fluid, typically a water stream [7] or an air stream [8]. Thesecond configuration is of particular interest, because the processair stream is directly supplied to the data center facility.

It is put in evidence that in case of data centers applications, thesystem is arranged in recirculation mode: the primary (or process)air stream is extracted from the building, it is cooled in the indi-

S. De Antonellis et al. / Energy and B

Nomenclature

AHE Heat exchanger cross area [m2]A, B, C Test conditionscp Specific heat [kJ kg−1 K−1]h Net channel height [m]L Gross plates length and width [m]L∗ Net plates length and width [m]m Specific flow rate [kg s−1 m−2]M Flow rate [kg s−1]M Mass [kg]NHE Number of heat exchanger plate [-]pt Plates pitch [mm]Q Volumetric flow rate [m3 h−1]t Time [s]T Dry bulb temperature [◦C]Twb Wet bulb temperature [◦C]v Channel air velocity [m s−1]xi Measured quantity [-]X Humidity ratio [kg kg−1]y Calculated quantity [-]

Greek letters�P Pressure drop [Pa]�T Temperature difference [◦C]�X Humidity ratio difference [kg kg−1]εdb Dry bulb effectiveness [-]εwb Wet bulb effectiveness [-]ı Plates thickness [m]

SuperscriptsN Nominal condition (� = 1.2 kg m−3)

Subscriptsa Aireva Evaporated waterin Inletnet Netout Outletp Primary air streams Secondary air streamw Waterxi Measured quantityy Calculated quantity

Acronyms

rthvb

thtm[t

mso

vided by manufacturer are:

IEC Indirect evaporative cooling

ect evaporative cooling system and, finally, it is supplied back tohe facility. Instead a secondary air stream at outdoor conditions isumidified and used to cool the previous one. In addition, a con-entional cooling system should be installed in order to provideackup and peak load cooling capacity.

Ongoing researches about IEC systems mainly deal with newhermodynamic cycles, heat exchanger materials and geometries,umidification systems and evaluation of energy savings comparedo conventional devices [9]. In particular, experimental works are

ainly focused on performance evaluation of different prototypes10–16], of heat exchanger orientation [16] and of flows configura-ion [17–20].

In data centers applications, as previously described, the pri-ary air flow of the IEC system is recirculated and it is completely

eparated from the secondary air stream. Therefore, indirect evap-rative coolers based on M-cycle heat exchangers [12] or in

uildings 121 (2016) 130–138 131

regenerative configurations [13] are not suitable for the investi-gated application. In fact, in such systems the primary air is split intwo streams: the first one is supplied to the building and the secondone is humidified and used as secondary air stream. As a conse-quence, only systems with independent air flows can be considered[21].

At present there is a lack of extensive experimental studies ofsuch indirect evaporative cooling systems, especially in typical datacenter working conditions. Based on the aforementioned consider-ations, the aim of this work is:

- To provide a detailed experimental analysis of the indirect evap-orative cooling system in typical working conditions of datacenters.

- To evaluate the effect of water nozzle arrangement on IEC perfor-mance.

- To analyze the effect of water mass flow rate on IEC performance.

The apparatus has been designed in order to minimize waterconsumption, with water mass flow rate between 0.4% and 4% ofthe secondary air one. It is shown in many operating conditions ahigh fraction of evaporated water is achieved

(Meva/Mw,in > 40%

):

in those cases such system can be manufactured without a pump forwater recirculation, leading to a compact apparatus and minimizingrisks of bacterial contamination.

2. Experimental set up

2.1. Description of the investigated indirect evaporative coolingsystem

The analyzed indirect evaporative cooling system consists of(Fig. 1):

- A commercial cross-flow plate heat exchanger.- Water spray nozzles installed in the upper part of the system.- An apparatus to increase pressure of water supplied to the noz-

zles.

The heat exchanger is made of aluminium plates with cross flowarrangement. Main characteristics are:

- Number of plates NHE = 119.- Plates thickness ı = 0.14 mm.- Plates pitch pt = 3.35 mm.- Net channel height h = pt − ı = 3.21 mm.- Gross plate length and width L = 500 mm.- Net plate length and width L∗ = 470 mm.- The plates spacing is obtained through dimples with semi-

spherical shape.

As shown in Fig. 1, two horizontal water manifolds are symmet-rically installed in the upper part of the heat exchanger casing. Ineach of them up to 4 nozzles (n◦ 8 in total) can be installed: thedistance between each nozzle along the water manifold is around8 cm. Instead the distance between the two manifolds is 18 cm andboth of them are installed 15 cm from the heat exchanger face.

Two different axial flow—full cone nozzles, characterized by adifferent orifice diameter, have been adopted. Nominal data pro-

- Nozzle A: water flow of each nozzle equal to 3.52 l h−1 at 10 bar.- Nozzle B: water flow of each nozzle equal to 7.50 l h−1 at 9 bar.

132 S. De Antonellis et al. / Energy and Buildings 121 (2016) 130–138

Fig. 1. Scheme of the different configurations of the in

Table 1Sensors main data.

Abbreviation Type of sensor Accuracya

Tb PT 100 Class A ±0.2 ◦CRHb Capacitive ±1% (between 0 and 90%)p Piezoresistive ±0.5% of reading ±1 Pa

up

2

dtstehisoraapiota

scaaz±

3

ttc

a At T = 20 ◦C.b Temperature and relative humidity probe.

Water is supplied to nozzles through a commercial pumpingnit with maximum flow rate equal to 110 l h−1 and maximumressure equal to 15 bar.

.2. Description of the test rig

Performance of the indirect evaporative cooling systemescribed in Section 2.1 has been evaluated through a specificest rig (Fig. 2), which has been designed to provide primary andecondary air streams at controlled conditions [22,23], in ordero evaluate performance of rotative heat exchangers, plate heatxchangers and desiccant wheels. Supply air temperature andumidity ratio are properly controlled through heating coils, cool-

ng coils and adiabatic humidifiers. In the air handling unit ofecondary air flow, an additional electrical heater is installed inrder to adjust the flow temperature up to 100 ◦C. Volumetric flowates of primary and secondary air streams are controlled by vari-ble speed fans and can be adjusted respectively up to 1400 m3 h−1

nd 2000 m3 h−1. Each of them is measured through two orificelates, constructed according to technical standards [24,25] and

nstalled in two different parallel ducts. Pressure drop across therifices is measured by piezoresistive transmitters. Depending onhe desired supply air conditions, the system can work in outdoorir mode or in recirculation air mode.

Temperature and relative humidity of each air stream are mea-ured at the inlet and outlet sections of the indirect evaporativeooling system casing through RTD PT100 probes coupled with rel-tive humidity capacitive sensors. Main data of calibrated sensorsre summarized in Table 1. The water flow Mw supplied to the noz-les is measured through a turbine flow sensor, whose accuracy is3% of reading.

. Adopted methodologies

In this work 112 tests have been carried out in order to evaluatehe indirect evaporative air cooler performance. Primary air condi-ions have been kept constant in representative conditions for dataenter applications, with Tp,in = 35 ◦C (assumed equal to data cen-

vestigated indirect evaporative cooling system.

ter set point temperature) and vpN equal to 3.7 m s−1 (QNp around1200 m3 h−1). Instead, water humidification setup and secondaryair working conditions have been widely modified. More precisely,tests have been organized in order to evaluate the effect of:

- Water nozzles orientation (parallel and counter flow).- Water nozzles characteristics (Type A and B).- Water nozzles number (4 or 8).- Water flow rate (Mw from 5 to 60 l h−1 or mw,in from 0.015 to

0.19 kg s−1 m−2).- Secondary air flow rate (vsN equal to 3.7 and 5.7 m s−1 and QNs

around 1200 and 1800 m3 h−1).- Secondary air inlet conditions (as reported in Table 2).

In all tests secondary air conditions have been properly set inorder to compare directly different experimental results, evaluatingeffects of inlet air dry bulb temperature, wet bulb temperature andhumidity ratio on system performance. As shown in Table 2, threedifferent secondary air conditions, namely A, B and C, have beenconsidered. It is highlighted that in conditions A and B there is thesame dry bulb temperature (Ts,in = 30 ◦C), in B and C the same wetbulb temperature (Twb,s,in = 22 ◦C) and in A and C the same humidityratio (Xs,in = 10.6 g kg−1).

In experimental tests data are collected in steady state condi-tions: in each session at least 300 samples of every physical quantityare logged with a frequency of 1 Hz. The uncertainty of evaluatedquantities is estimated in accordance with the work of Moffat [26].More precisely the experimental uncertainty uxi of each direct mon-itored variable xi (T, ϕ and p) is:

uxi = ±√u2xi,inst

+ (t95�xi )2 (1)

Where uxi,inst is the instrument uncertainty of the measured quan-tity, t95 is the student test multiplier at 95% confidence and �xi is thestandard deviation of the mean. The methodology and the assump-tions are described in detail in the reference international standard[27].

Instead the generic combined uncertainty uy of calculated quan-tities y, such as εwb, �Tp, �Ts, �Xs and Meva/Mw,in is calculatedas:

uy =

√√√√∑i

(∂y∂xiuxi,inst

)2

+ t295

∑i

(∂y∂xi�xi

)2

(2)

S. De Antonellis et al. / Energy and Buildings 121 (2016) 130–138 133

Fig. 2. Scheme of the test rig.

Table 2Classification of test conditions.

Test Condition Ts,in [◦C] Twb,s,in [◦C] Xs,in [g kg−1] ϕs,in[%] vsN [m s−1] Tp,in[◦C] Xp,in [g kg−1] vpN [m s−1]

40.50.27.

Imb

ε

ε

smp

rtp

A 30.0 20.0 10.6

B 30.0 22.0 13.4

C 36.8 22.0 10.6

n the experimental analysis performed in section 4, results areainly compared in terms of dry bulb effectiveness εdb and wet

ulb effectiveness εwb, respectively defined in the following way:

db =Mpcpp

(Tp,in − Tp,out

)Mpcpp

(Tp,in − Ts,in

) (3)

wb =Mpcpp

(Tp,in − Tp,out

)Mpcpp

(Tp,in − Twb,s,in

) (4)

In both Eqs. (3) and (4) it is assumed that thermal capacity ofecondary air stream is equal or higher than the one of the pri-ary air flow (Mscps ≥ Mpcpp), according to the conditions of the

erformed tests.

Further quantities that have been evaluated in the experimental

esults analysis are the specific water mass flow rate mw,in, the frac-ion of evaporated water Meva/Mw,in, the temperature variation ofrimary and secondary air streams, namely �Tp and �Ts, and the

0 3.7, 5.7 35.0 10.0 3.70 3.7, 5.7 35.0 10.0 3.73 3.7, 5.7 35.0 10.0 3.7

variation of the humidity ratio of the secondary air stream �Xs.Such terms are defined in the following Eqs. ((5)–(10)):

mw,in = Mw,inAHE,net

(5)

Meva

Mw,in=Ms

(Xs,out − Xs,in

)Mw,in

(6)

�Tp = Tp,in − Tp,out (7)

�Ts = Ts,in − Ts,out (8)

�Xs = Xs,out − Xs,in (9)

The net heat exchanger cross area, equal to 0.089 m2, is evaluatedas: ( ) ∗

AHE,net =

HHE − NHEı LHE2

(10)

Where the heat exchanger height can be calculated asHHE = (NHE − 1) pt + ı.

134 S. De Antonellis et al. / Energy and Buildings 121 (2016) 130–138

Table 3Preliminary tests in dry conditions.

Test vpN [m s−1] vsN [m s−1] Ts,in [◦C] Xs,in [g kg−1] Tp,in [◦C] Xp,in [g kg−1] εdb [-]

1 2.84 2.84 46.2 11.0 30.4 11.0 0.6262 3.69 3.69 57.2 11.0 32.9 11.0 0.621

Fig. 3. Wet bulb effectiveness as a function of m˙w,in for parallel and counter flow nozzles arrangement. Condition A, 8 nozzles type A, vsN equal to 3.7 m s−1 (Figure A) and5.7 m s−1 (Figure B).

F ,in for

tc

v

Wa

-

-

-

ig. 4. Wet bulb effectiveness and fraction of evaporated water as a function of m˙w

Air velocities vaN reported in the experimental analysis of Sec-ion 4 are referred to normal air conditions (�a = 1.2 kg m−3) andalculated in this way:

Na = QNa

3600 AHE,net(11)

here QNa is the volumetric air flow rate in m3 h−1, referred to theforementioned reference conditions.

Finally, the following considerations are pointed out:

In all tests the water temperature supplied to nozzles was around20 ◦C.

In all experiments the difference between total heat exchangedby primary and secondary air streams was within 5%.

In a few tests it has been verified the correctness of the water mass

balance by measuring the amount of water at the bottom of theheat exchanger in a determined period. After the evaluation of theaverage drained water flow Mdrain = Mdrain/tdrain, in the analyzedtests the ratio Mdrain/(Mw,in − Meva) was between 94% and 107%.

different counter flow nozzles arrangement. Condition A and vsN equal to 3.7 m s−1.

4. Experimental results

4.1. Preliminary considerations in dry conditions

Two preliminary tests have been performed in dry conditions, inorder to determine the dry bulb effectiveness of the heat exchanger.Tests have been carried out with balanced flows: inlet air conditionsand measured effectiveness are summarized in Table 3.

The nominal pressure drop across the heat exchanger is 100 Pawhen va = 3.2 m s−1 and Ta = 35 ◦C.

4.2. Effect of water nozzles arrangement

In Fig. 3 it is shown the wet bulb effectiveness as a function ofthe specific water mass flow rate, for different nozzles orientationand secondary air velocity. Tests are carried out in conditions A,

as reported in Table 2. The counter flow arrangement is clearlymore effective than the parallel flow one, due to the mixing ofwater droplets in the air stream, which leads to a uniform dis-tribution of water in the heat exchanger. Instead, in the parallel

S. De Antonellis et al. / Energy and Buildings 121 (2016) 130–138 135

Fig. 5. Wet bulb effectiveness and fraction of evaporated water as a function of m˙w,in for different type of nozzles. Condition A, 8 nozzles, counter flow arrangement and vsN

equal to 3.7 m s−1 and 5.7 m s−1.

F ,in for

e

fldbtwia(

wtcnfinmatp

witito

ig. 6. Wet bulb effectiveness and fraction of evaporated water as a function of m˙wqual to 3.7 m s−1 and 5.7 m s−1.

ow configuration, water mainly flows between plates positionedirectly in front of nozzles, with a consequent poor water distri-ution. The higher mw,in, the lower differences in εwb between thewo arrangements. In particular when mw,in > 0.7kg s−1m−2 theet heat exchanger surface increases in both configurations, lead-

ng to close wet bulb effectiveness. Finally, it is highlighted that theforementioned trend does not depend on secondary air velocityvsN = 3.7 and 5.7 m s−1), as shown in Fig. 3a and b.

The effect of nozzles quantity and type on effectiveness andater consumption is investigated in Fig. 4. It is possible to state

hat performance (tests in condition A, vsN = 3.7 m s−1, counter flowonfiguration) strongly depends on water flow rate and slightly onozzles arrangement. More precisely, at given water flow rate con-guration with 4 nozzles is slightly better than configuration with 8ozzles. In fact, in the first configuration nozzles are installed in theiddle of the plenum (Fig. 1) and a lower amount of water drops

gainst boundary walls of the casing. The slight increase in effec-iveness is around 1% and it is very limited compared to effects onerformance related to the variation of water mass flow rate.

In Fig. 4a and b it is shown, quite obviously, that the higher theater mass flow rate, the higher the wet bulb effectiveness. Anyway

t is highlighted that in case of low mw,in, (<0.04 kg s−1 m−2), despite

he low effectiveness (εwb < 0.65) the fraction of evaporated waters higher than 60%. Instead, in case of high mw,in (>0.1 kg s−1 m−2)he effectiveness is almost higher than 0.75 but the fraction of evap-rated water drastically reduces, being always lower than 0.3.

different type of nozzles. Condition B, 8 nozzles, counter flow arrangement and vsN

Finally, it is possible to state that in the investigated setup εwband Meva/Mw,in mainly depend on water flow rate, regardless oftype and number of nozzles.

4.3. Effect of secondary air conditions

In Figs. 5–7 the effect of different secondary air conditions(Table 2) and velocity on system performance is evaluated. All testshave been performed with 8 nozzles and different water flow rates.

Results of experiments in condition A (Ts,in = 30.0 ◦C,Xs,in = 10.6 g kg−1) are shown in Fig. 5: the higher vsN , the higherεwb and Meva/Mw,in. It should be noticed that in this case Ts,in < Tp,inand, therefore, even in dry conditions an increase in the secondaryair flow rate would lead to a higher effectiveness. Instead theincrease in Meva/Mw,in is mainly related to the higher mass trans-fer coefficient and to the higher humidity ratio difference betweenair and wet plates surface. Therefore, in condition A unbalancedflows (vsN > vpN) perform better than balanced ones.

In Fig. 6 secondary air is set to condition B (Ts,in = 30.0 ◦C,Xs,in = 13.4 g kg−1): compared to condition A, it is characterized bythe same dry bulb temperature and a higher humidity ratio. In these

tests, at a given water flow rate, εwb and Meva/Mw,in are respec-tively higher and lower than condition A. The increase in εwb isrelated to the increase in Twb,s,in and not in the reduction in Tp,out ,as discussed in detail in the following of this section. Instead the

136 S. De Antonellis et al. / Energy and Buildings 121 (2016) 130–138

Fig. 7. Wet bulb effectiveness and fraction of evaporated water as a function of m˙w,in for different type of nozzles. Condition C, 8 nozzles, counter flow arrangement and vsN

equal to 3.7 m s−1 and 5.7 m s−1.

Fig. 8. Wet bulb effectiveness and �Tp as a function of m˙w,in for different type of nozzles and secondary air conditions. 8 nozzles, counter flow arrangement and vsN equalto 3.7 m s−1.

ndary

rc

ith

Fig. 9. �Xs and �Ts as a function of m˙w,in for different type of nozzles and seco

eduction in Meva/Mw,in depends on the higher Xs,in and on theonsequent reduced mass transfer driving force.

Performance in condition C of the indirect evaporative coolers reported in Fig. 7. In this case wet bulb temperature is equalo condition B (Twb,s,in = 22.0 ◦C) while dry bulb temperature andumidity ratio are respectively higher (Ts,in = 36.8 ◦C) and lower

air conditions. 8 nozzles, counter flow arrangement and vsN equal to 3.7 m s−1.

(Xs,in = 10.6 g kg−1). In Fig. 7a it is shown that an increase in vsN doesnot lead to an increase in εwb. In fact in condition C it is Ts,in > Tp,in:

in dry condition the primary air stream would be heated by thesecondary air flow. Therefore the secondary air stream should beproperly humidified in order to reduce Ts along the heat exchangerand to cool the primary air flow: this effect can be achieved by

and B

rita

mrtcostcSat

M

baeeATi

iviac

5

sor

-

-

-

-

-

-

A

oMf

[

[

[

[

[

[

[

[

[

[

[

S. De Antonellis et al. / Energy

educing vsN or by increasing mw,in. If a limited amount of waters used in the humidification process

(mw,in < 0.15kg s−1m−2

), in

he investigated tests balanced flows are more effective due to theforementioned reasons.

In condition C the wet bulb effectiveness is lower than the oneeasured in condition B at the same water and secondary air flow

ate. The reduction in effectiveness is related to the reduction ofemperature drop of the primary air stream, while Twb,s,in is keptonstant (and, consequently the denominator of Eq. (4)). Quitebviously the increase in Ts,in lead to a decrease in the primary airtream cooling. At balanced air flows a wet bulb effectiveness equalo 0.7 is reached in condition B with mw,in ≈ 0.05kg s−1m−2 and inondition C with mw,in ≈ 0.12kg s−1m−2 (around 2.4 times higher).uch increase in the water flow rate is directly related to the highermount of water that should evaporate in heat exchanger in ordero reach saturation conditions.

Finally, also in this condition a higher vsN leads to a higher˙ eva/Mw,in, as previously discussed.

In Figs. 8 and 9 system performance in conditions A, B and C atalanced flow conditions are compared in terms of εwb, �Tp, �Tsnd �Xs. In Fig. 8 it is clearly shown that an increase in wet bulbffectiveness is not directly related to an increase in �Tp. In fact,ven if in condition B the effectiveness is higher than in condition, the decrease of process air temperature is higher in condition A.herefore care should be taken when results about different work-

ng conditions are compared.In Fig. 9 the variation of secondary air conditions across the

ndirect evaporative cooling system are reported. The maximumariations of �Xs and of �Ts occur in condition C, due to the lowestnlet relative humidity (ϕs,in = 27.3%). For the same reason the vari-tion of such quantities is higher in condition A (ϕs,in = 40%) than inondition B (ϕs,in = 50%).

. Conclusions

In this work an indirect evaporative air cooling system is exten-ively tested in order to evaluate performance under differentperating conditions and nozzles arrangement. Most significantesults of this research are hereinafter reported:

Nozzles installed in counter flow arrangement provide higher wetbulb effectiveness.

Performance depends strongly on water flow rate and slightly onnozzles number and size.

In condition A and with balanced flows the fraction of evaporatedwater Meva/Mw,in varies between 0.98 (mw,in ≈ 0.02kgs−1m−2)and 0.18 (mw,in ≈ 0.17kgs−1m−2). Such values increase in case ofhigh flow rate and low inlet relative humidity of the secondaryair stream.

At a given water flow rate, wet bulb effectiveness is stronglyinfluenced by secondary air conditions.

An increase in secondary air flow rate leads to an increase in wetbulb effectiveness if its inlet temperature is lower than the one ofthe primary air stream.

In case of low inlet relative humidity and high inlet temperature ofthe secondary air stream, the water flow rate should be increasedin order to have satisfactory wet bulb effectiveness (εwb > 0.7).

cknowledgments

The authors would like to acknowledge Mr Leone and Mr Bawaf Recuperator S.p.A. for the economical and technical support andr Favaretto, Mr Martello and Mr Boscaro of Carel Industries S.p.A.

or their technical support.

[

uildings 121 (2016) 130–138 137

References

[1] Khosrow Ebrahimi, Gerard F. Jones, Amy S. Fleischer, A review of data centercooling technology, operating conditions and the corresponding low-gradewaste heat recovery opportunities, Renew. Sustain. Energy Rev. 31 (March)(2014) 622–638, http://dx.doi.org/10.1016/j.rser.2013.12.007, ISSN0321-1364.

[2] Hainan Zhang, Shuangquan Shao, Hongbo Xu, Huiming Zou, Changqing Tian,Free cooling of data centers: a review, Renew. Sustain. Energy Rev. 35 (July)(2014) 171–182, http://dx.doi.org/10.1016/j.rser.2014.04.017, ISSN1364-0321.

[3] Eduard Oró, Victor Depoorter, Albert Garcia, Jaume Salom, Energy efficiencyand renewable energy integration in data centres. Strategies and modellingreview, Renew. Sustain. Energy Rev. 42 (February) (2015) 429–445, http://dx.doi.org/10.1016/j.rser.2014.10.035, ISSN 1364-0321.

[4] ASHRAE, Thermal Guidelines for Data Processing Environments—ExpandedData Center Classes and Usage Guidance, 2011.

[5] ASHRAE, Thermal Guidelines for Data Processing Environments Guidance,2008.

[6] Kuei-Peng Lee, Hsiang-Lun Chen, Analysis of energy saving potential ofair-side free cooling for data centers in worldwide climate zones, EnergyBuild. 64 (September) (2013) 103–112, http://dx.doi.org/10.1016/j.enbuild.2013.04.013 (ISSN 0378-7788).

[7] B. Costelloe, D. Finn, Indirect evaporative cooling potential in air–watersystems in temperate climates, Energy Build. 35 (July (6)) (2003) 573–591,http://dx.doi.org/10.1016/S0378-7788(02)00161-5, ISSN 0378-7788.

[8] Shahram Delfani, Jafar Esmaeelian, Hadi Pasdarshahri, Maryam Karami,Energy saving potential of an indirect evaporative cooler as a pre-cooling unitfor mechanical cooling systems in Iran, Energy Build. 42 (November (11))(2010) 2169–2176, http://dx.doi.org/10.1016/j.enbuild.2010.07.009, ISSN0378-7788.

[9] Zhiyin Duan, Changhong Zhan, Xingxing Zhang, Mahmud Mustafa, XudongZhao, Behrang Alimohammadisagvand, Ala Hasan, Indirect evaporativecooling: past, present and future potentials, Renew. Sustain. Energy Rev. 16(December (9)) (2012) 6823–6850, http://dx.doi.org/10.1016/j.rser.2012.07.007, ISSN 1364-0321.

10] Ana Tejero-González, Manuel Andrés-Chicote, Eloy Velasco-Gómez, FranciscoJavier Rey-martínez, Influence of constructive parameters on the performanceof two indirect evaporative cooler prototypes, Appl. Therm. Eng. 51 (March(1–2)) (2013) 1017–1025, http://dx.doi.org/10.1016/j.applthermaleng.2012.10.054, ISSN 1359-4311.

11] Eloy Velasco Gómez, Ana Tejero González, Francisco Javier Rey Martínez,Experimental characterisation of an indirect evaporative cooling prototype intwo operating modes, Appl. Energy 97 (September) (2012) 340–346, http://dx.doi.org/10.1016/j.apenergy.2011.12.065, ISSN 0306-2619.

12] Sergey Anisimov, Demis Pandelidis, Andrzej Jedlikowski, Vitaliy Polushkin,Performance investigation of a M (Maisotsenko)-cycle cross-flow heatexchanger used for indirect evaporative cooling, Energy 76 (November)(2014) 593–606, http://dx.doi.org/10.1016/j.energy.2014.08.055, ISSN0360-5442.

13] Frank Bruno, On-site experimental testing of a novel dew point evaporativecooler, Energy Build. 43 (December (12)) (2011) 3475–3483, http://dx.doi.org/10.1016/j.enbuild.2011.09.013, ISSN 0378-7788.

14] F.J. Rey Martıınez, E. Velasco Gómez, R. Herrero Martıın, J. MartıınezGutiérrez, F. Varela Diez, Comparative study of two different evaporativesystems: an indirect evaporative cooler and a semi-indirect ceramicevaporative cooler, Energy Build. 36 (July (7)) (2004) 696–708, http://dx.doi.org/10.1016/j.enbuild.2003.10.010, ISSN 0378-7788.

15] T.R. Tulsidasani, R.L. Sawhney, S.P. Singh, M.S. Sodha, Recent research on anindirect evaporative cooler (IEC) part 1: optimization of the COP, Int. J. EnergyRes. 21 (1997) 1099–1108, http://dx.doi.org/10.1002/(SICI)1099-114X(19971010)21:12<1099:AID-ER311>3.0.CO;2-S.

16] W.Y. Saman, S. Alizadeh, An experimental study of a cross-flow type plateheat exchanger for dehumidification/cooling, Sol. Energy 73 (July (1)) (2002)59–71, http://dx.doi.org/10.1016/S0038-092X(01)00078-0, ISSN 0038-092X.

17] Ghassem Heidarinejad, Mojtaba Bozorgmehr, Shahram Delfani, JafarEsmaeelian, Experimental investigation of two-stage indirect/directevaporative cooling system in various climatic conditions, Build. Environ. 44(October (10)) (2009) 2073–2079, http://dx.doi.org/10.1016/j.buildenv.2009.02.017, ISSN 0360-1323.

18] Min-Hwi Kim, Jin-Hyo Kim, An-Seop Choi, Jae-Weon Jeong, Experimentalstudy on the heat exchange effectiveness of a dry coil indirect evaporationcooler under various operating conditions, Energy 36 (November (11)) (2011)6479–6489, http://dx.doi.org/10.1016/j.energy.2011.09.018, ISSN 0360-5442.

19] Aftab Ahmad, Shafiqur Rehman, Luai M. Al-Hadhrami, Performanceevaluation of an indirect evaporative cooler under controlled environmentalconditions, Energy Build. 62 (July) (2013) 278–285, http://dx.doi.org/10.1016/j.enbuild.2013.03.013, ISSN 0378-7788.

20] B. Riangvilaikul, S. Kumar, An experimental study of a novel dew pointevaporative cooling system, Energy Build. 42 (May (5)) (2010) 637–644,http://dx.doi.org/10.1016/j.enbuild.2009.10.034, ISSN 0378-7788.

21] Sergey Bolotin, Borys Vager, Vladimir Vasilijev, Comparative analysis of thecross-flow indirect evaporative air coolers, Int. J. Heat Mass Transfer 88(September) (2015) 224–235, http://dx.doi.org/10.1016/j.ijheatmasstransfer.2015.04.072, ISSN 0017-9310.

1 and B

[

[

[

[

Standardization, ISO IEC Guide 98-3, Geneva Switzerland, 2008.[26] R.J. Moffat, Describing the uncertainties in experimental results, Exp. Therm.

38 S. De Antonellis et al. / Energy

22] Stefano De Antonellis, Manuel Intini, Cesare Maria Joppolo, Desiccant wheelseffectiveness parameters: correlations based on experimental data, EnergyBuild. 103 (September) (2015) 296–306, ISSN 0378-7788.

23] Stefano De Antonellis, Manuel Intini, Cesare Maria Joppolo, Luca Molinaroli,Francesco Romano, Desiccant wheels for air humidification: an experimental

and numerical analysis, Energy Convers. Manage. 106 (December) (2015)355–364, http://dx.doi.org/10.1016/j.enconman.2015.09.034, ISSN0196-8904.

24] DIN EN ISO 5167-2 Standards, Measurement of fluid flow by means ofpressure differential devices inserted in circular cross-section conduits

[

uildings 121 (2016) 130–138

running full—Part 2: Orifice plates (ISO 5167-2:2003).25] ISO IEC Guide 98-3, Uncertainty of Measurement—Part 3: Guide to Expression

of Uncertainty in Measurement, International Organization for

Fluid Sci. 1 (1) (1988) 3–17.27] Evaluation of measurement data—Guide to the expression of uncertainty in

measurement JCGM 100:2008 (GUM 1995 with minor corrections).