Experimentally validated model for atmospheric water generation using a solar assisted desiccant dehumidification system

Page 1

Eu

Da

b

a

ARRAA

KPPMCDW

1

qNstdmpeInvma

h0

Energy and Buildings 77 (2014) 236–246

Contents lists available at ScienceDirect

Energy and Buildings

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

xperimentally validated model for atmospheric water generationsing a solar assisted desiccant dehumidification system

ia Milania, Abdul Qadira, Anthony Vassalloa, Matteo Chiesab, Ali Abbasa,∗

School of Chemical and Biomolecular Engineering, The University of Sydney, NSW 2006, AustraliaInstitute Center for Energy (iEnergy), Masdar Institute of Science and Technology, Abu Dhabi, United Arab Emirates

r t i c l e i n f o

rticle history:eceived 27 September 2013eceived in revised form 6 March 2014ccepted 8 March 2014vailable online 28 March 2014

eywords:sychrometricVTodelling

oolingehumidificationater condensation

a b s t r a c t

This paper examines an alternative solution for emergency situations where freshwater and utilities areoften interrupted. Generating freshwater from the atmosphere using a small-scale air-cooled desiccantwheel dehumidifier was experimented. Condensed water was collected and systematically recordedagainst local meteorological data. A synthetic model simulating the actual lay-out of the experiment wasbuilt in TRNSYS. The model validated the experimental results and generated approximately 52 litres in9 days. The validated model is then run for three different climates; Sydney, Abu Dhabi and London toestimate annual water production. Abu Dhabi showed the best results compared with Sydney and Londonby generating 18.5 kL of water per year. The model is further developed to evaluate thermodynamicbenefits of using dehumidified processed air as a feed stream for a proposed small-scale air-conditioningsystem. The energy required for the wheel regeneration process is met by thermal gain of modified solarPVT panels where the stagnant heat at the back is used to heat up the regeneration air stream. It was foundthat the dehumidification process can significantly decrease the latent load of the air-conditioner and

easily bring indoor humidity level to the human comfort zone. However, dehumidified process air is alsoincreasing the sensible load because of the higher temperature associated with dehumidification process.Rooftop solar PV panels can easily meet the power demand of appropriate lighting, a computer and minirefrigerator for extended hours if an appropriate set of batteries are fitted, but unable to exclusively meetthe air-conditioner power demand to maintain indoor temperature within human comfort zone.

© 2014 Elsevier B.V. All rights reserved.

. Introduction

In the last century, there has been a dramatic increase in the fre-uency and intensity of natural disaster rates across the globe [1].atural disasters are declared when the physical hazards cause a

ubstantial damage or loss to social, economical and environmen-al assets that directly or indirectly threaten people’s lives [2]. Thisramatic increase of natural disaster rates is mainly caused by cli-ate change phenomena which directly influence the precipitation

attern and extreme weather events [3]. When the catastrophexceeds a society’s coping threshold, an emergency is declared.n most emergency situations, local water, sanitation and poweretworks are disrupted and no longer can be used. Therefore, pro-

iding safe and clean drinking water along with food, shelters andedication to prevent the spread of waterborne diseases within

ffected communities becomes a priority. In the past, the effort

∗ Corresponding author. Tel.: +61 2 9351 3002; fax: +61 2 9351 2854.E-mail address: [email protected] (A. Abbas).

ttp://dx.doi.org/10.1016/j.enbuild.2014.03.041378-7788/© 2014 Elsevier B.V. All rights reserved.

was to continuously deliver large quantities of clean water viawater tankers or small containers. However, delay in the deliveryof this costly operation and/or the collapse of road network in somecases, pressed on crisis handling personnel to consider easier andmore effective solutions. In this context, the idea of establishingonsite decentralised water generation technology might be moreattractive. However, treating water up to drinking standards dur-ing an emergency situation is challenging because of inadequateinfrastructure and the interruption that is often associated withthe chaos of the disaster. Therefore, an emergency water genera-tion technology must be safe, easy to establish and independentof utility networks to permit their application in disaster zones.Furthermore, the water generation device must also be a mobileor portable small-scale unit to provide a location-specific solu-tion. Immediate and adequate medical relief is also demanding arapid establishment of small independent field hospitals. Today,

first-aid cabins similar to that shown in Fig. 1 are easy to trans-fer and install in disaster zones to replace old-style medical tents.These cabins are often used for initial first-aid, medical assess-ment and immunisation services. They contain one or two beds,

Page 2

D. Milani et al. / Energy and Buil

COP coefficient of performanceMRC moisture removal capacitySER sensible energy ratioTEC thermoelectric coolerCpda dry air specific heat capacity coefficient [kJ/kgda K]Cpf circulating fluid specific heat capacity coefficient

[kJ/kg K]hda dry air enthalpy of the air [kJ/kgda]hma moist air enthalpy of the air [kJ/kgma]hwv water vapour enthalpy of the air [kJ/kgwv]ω humidity ratio [kgwv/kgda]� the specific volume of the air [m3

ma/kgda]dv the absolute humidity of the air [kgwv/m3

ma]�da dry air density [kgda/m3

ma]�ma moist air density [kgma/m3]Fvma moist air volumetric flow rate [m3

ma/h]Fma moist air mass flow rate [kgma/h]Fda dry air mass flow rate [kgda/h]mw water condensate mass flow rate [kgw/h]mf circulating fluid mass flow rate [kg/h]� specific humidity of the air [kgwv/kgma]�d dehumidifier efficiency�e emperical PV curve-fitting parameterς solar thermal collector efficiencyIL PV module photocurrentIL,ref PV module photocurrent at reference conditionsImp current at maximum power point along IV curveIo Diode reverse saturation current [Amp]Io,ref diode reverse saturation current at reference condi-

tions [Amp]Rs PV module series resistanceNs number of individual cells in the moduleTc solar module temperature [◦C]k Boltzmann constant [J/K]q electron charge constant [C]GT total solar radiation incidence [W/m2]Qu total thermal gain by the collector [kJ/h]

aaupc(

into cooling. Sorption cycle works by way of interaction of two or

Fig. 1. Solar cabin where PV panels are flat mounted on the rooftop.

table and a computer, a small refrigerator, small air-conditionernd adequate lighting. These cabins are designed to accommodatep to four medical personnel and patients at a time. The major

ortion of energy requirement for these cabins is met by five mono-rystalline silicon photovoltaic panels flat mounted on the rooftopFig. 1). A simple vapour-compression air-conditioning unit is used

dings 77 (2014) 236–246 237

to bring the temperature and humidity level inside the cabin tomeet human comfort level. In hot humid regions such as tropi-cal and coastal areas, the thermal load often exceeds the capacityof the designated air-conditioning unit. This is mainly because ofhigh atmospheric humidity ratio which possesses a high latent loadresulting from atmospheric water vapour condensation. Thereforethe idea of using a pre-treatment dehumidification unit to mitigatethe humidity content of the feed air and harvest the condensatecould be attractive.

This paper examines the possibility of using a dehumidificationsystem run by solar thermal energy for two specific purposes; (i) topre-treat feed air stream for the air-conditioning unit and reducelatent heat and consequently electrical power consumption. (ii) tocondense atmospheric moisture and use it as an additional renew-able source of water and further enhance the sustainability andindependence of first-aid cabins. The objective is to experimen-tally quantify collected water and energy consumption to comparewith a synthetic mathematical model built in a TRNSYS 17 plat-form using the real-time weather data recorded at the same timeand location where the experiment took place. Agreement betweenexperimental and mathematical model can validate the modellingprocess to predict the techno-economics of the atmospheric watergeneration anywhere and at anytime.

2. Atmospheric water dehumidification

The atmosphere surrounding the earth is estimated to containa total of over 12.9 × 1012 m3 of renewable water. This amountis even greater than the total available freshwater in marshes,wetlands and rivers around the world [4]. Water generation fromatmospheric air is considered as a renewable water source andhas sporadically appeared in the literature [5–14]. Most of theseworks have praised the viability of the process, especially near trop-ical, temperate and coastal areas where temperature and humiditylevels are typically high. However, their arguments were aboutthe energy intensive operation of water generation. High energyrequirement was declared as the biggest obstacle of the process andbecame a driving force for a variety of innovations. Dehumidifica-tion is a widely used technology in many industrial processes wherea stream of dry air is needed. It is also an essential part for heat-ing and air-conditioning process where humidity level of the air iscontrolled to meet human comfort requirements. The heating, ven-tilation and air conditioning (HVAC) system consumes the primeshare of the energy consumption and can account in some cases for70% of the total energy consumption of buildings [15]. Traditionally,condensed water is drained away of air-conditioning systems andthe energy consumed for condensation is wasted. Fig. 2 presents adetailed classification of various dehumidification techniques withat least one example of each technique being given. The first cat-egory of methods is represented by cooling surfaces that cool themoist air feed to below dew point temperature and condense themoisture content over specially designed cooling surfaces. This cat-egory was comprehensively addressed in a previous study [16]where a generic no-refrigerant TEC dehumidification techniquewas modelled and evaluated. In the second category, sorption isa mechanism in which water vapour molecules are captivated bya distinctive solvents or solid mediums that are specially affiliatedtoward water vapour molecules. In a later stage, after specific dif-fusion time, captured water vapour molecules will be desorbedby subjecting the carrier medium to heat. A sorption system hasan exceptional capability of transforming thermal energy directly

more substances, one of them at least is changing phase duringthe operation from liquid to vapour and vice versa. Among the pairof substances, the substance with the lower boiling temperature

Page 3

238 D. Milani et al. / Energy and Buildings 77 (2014) 236–246

r dehu

isdcralontdstwfeobbteggprasafcp

Fig. 2. Classification of moist ai

s the sorbate which plays the role of the refrigerant and the otherubstance is called the sorbent. Two main sub-categories are evi-ent in sorption cycles; open cycles and closed cycles. In openycles, the moisture is sorbed from a humid air stream using evapo-ative cooling principles. By capturing water molecules, the exitingir stream will be relatively drier. The adsorbed moisture accumu-ate in the desiccant (solid bed or liquid solvent) until saturationccurs where the desiccant no longer can uptake more water. In theext phase of the cycle, the desiccant enters the regeneration stageo re-evaporate water via heating. Closed sorption cycle does notiffer thermodynamically from an open cycle. Regardless of ves-el configuration (open or closed) and internal enthalpy changes,he same amount of energy must be consumed to remove 1 kg ofater under similar psychrometric conditions. The key advantage

or closed cycle is easier circulation of the chilled water productspecially for applications that require decentralised heat loadsr chilled ceilings. In the third category, gas separation via mem-rane compared to other membrane applications is relatively new,ut the rate of growth of this technology has been substantial. Inhe last two decades, gas mixture separation by membranes hasmerged from being a laboratory curiosity to becoming a rapidlyrowing, commercially viable alternative to traditional methods ofas separation. The development of new composite materials androcess fabrication methods is expected to dominate other sepa-ation methods. Vapour permeation through membrane also hasdvanced. Vapour permeation technology denotes the transport ofpecific condensable vapours through a membrane from a vapour

nd/or gas feed mixture. A constant and stable positive drivingorce will ensure efficient product penetration. Various polymersan be used as a selective barrier material specifically for the trans-ort of water vapour. The main stimulus variables for water vapour

midification technologies [36].

transport through any polymer is the effect of water vapour activ-ity in the mixture, temperature, membrane type and thickness, feedpressure, flow rate and permeate properties. However, two mainobstacles still need to be overcome if membranes are to becomefeasible for ambient air dehumidification. The first problem is thevariation of temperature and relative humidity of feed air whichwill continuously change the composition of the mixture compo-nents and consequently the partial pressure of water vapour inthe feed stream. The second problem is the well-known mem-brane fouling phenomenon which is related to feed and permeateconcentration build-up in the vicinity of the membrane surface.Most water vapour separation membrane applications are stillin the R&D phase and are at early phases of commercialisation.The dehumidification by solid desiccant material is usually basedon a rotor, also called desiccant wheel, filled with an adsorbentmaterial, which is the main component of the desiccant systems.Desiccant dehumidifiers are energy efficient and environmentallybenign devices that remove moisture from the air without coolingthe air below its dew point [17]. For this reason, the performanceof this kind of wheel has been analysed as a function of the mostimportant operating conditions in many theoretical, numerical andexperimental research papers. In humid regions, desiccant dehu-midification could reduce total residential electricity demand by25% providing a drier, cleaner and more comfortable indoor envi-ronment with lower energy demand [18]. Desiccant cooling anddehumidification systems are being successfully used in variousindustrial and commercial markets offering clear advantages in

many HVAC applications. Desiccant cooling systems are used toimprove the indoor air quality of all types of buildings by effi-ciently controlling the moisture level of fresh or ventilation air[19]. Furthermore, since desiccant cooling systems can also count

Page 4

d Buil

authfFuitriTtatcwsspiTfliioitrpft

cceiafaheithriteacaadurw(wss

Paa

D. Milani et al. / Energy an

s roof-top air handlers, they compete directly with the most pop-lar conventional air conditioning system currently in use, such ashe roof-top vapour compression system. Desiccant dehumidifiersave earlier been applied in thermally activated cooling systems

or processing the ventilation air to buildings in humid climates.or systems that are powered by gas or electricity, the energysage can be reduced by adding solar thermal collectors. In roof-top

nstallations the collectors can be located near the air conditionero simplify the installation process. In these systems, a desiccantemoves moisture from the air via a process called sorption, whichs often associated with heat release and temperature increase.his thermal gain could also be used for space heating in wintero provide heat and indirectly heat the supply air. Temperaturend humidity loads are effectively and efficiently met by separatinghem in this way. The desiccant is then dried out (regenerated) toomplete the cycle using thermal energy supplied by natural gas,aste heat, or solar heat. Commercial desiccants are made of hygro-

copic materials such as silica gel, activated alumina, natural andynthetic zeolites, titanium silicate, lithium chloride, and syntheticolymers. Henning [20] reported a COP equal to 0.7 for solid des-

ccant systems was achievable under normal operating conditions.he COP of liquid desiccant dehumidification does not vary widelyrom solid desiccants. Similar COPs have also been reported foriquid desiccant systems [21]. Most of the available desiccant stud-es are for the purpose of air-conditioning and ventilation. Theres no significant desiccant based work optimised for the purposef water condensation and collection found in the literature. Thiss because conventional desiccant dehumidification is most likelyo diffuse back the moisture content of processed air in the bedegeneration process. For water condensation and collection pur-oses, this will need extra measures to contain hot humid air exitingrom the regenerator to be cooled again by using one of the coolingechniques.

In this study, a prototype dehumidification system is set toontinuously dehumidify an atmospheric air stream in an openircuit. To reduce the energy footprint of water production, a lownergy-consumer silica gel dehumidification system is chosen. Des-ccant dehumidification does not require consumption of excessivemounts of energy to lower the temperature below the dew pointor water condensation. This is a particularly advantageous char-cteristic for moderate coastal climates such as Sydney where theumidity level is usually high but the ambient temperature is mod-rate. Fig. 3 shows a schematic of the proposed process where thentake fresh air is split into two streams. The major stream goeso the desiccant wheel (stream 2) and the minor stream goes to aeat exchanger (stream 4) where it exchanges enthalpy with theegeneration stream (stream 6) and condenses water. The humid-ty set-point on the wheel is lowered to force the dehumidifiero work at its full capacity at all times. The major air (stream 2)nters the dehumidifier at a certain state (usually relatively coolnd moist compared to the outlet). As water adsorbs onto the desic-ant, the heat of sorption is released and warms up the surroundingir. When the process air leaves the dehumidifier, it is often driernd warmer than its entry status. Water vapour molecules areeposited on the dehumidification wheel and begin to accumulatentil saturation. The wheel is rotating at very slow rate to increaseesidual time for water vapour molecules. Once that part of theheel is saturated it will ultimately reach the regeneration phase

point 9). In the regeneration phase, the saturated portion of theheel is exposed to a hot dry air stream that comes from a heat

ource. In this case, the heat is supplied by the thermal part of hybridolar collector.

Photovoltaic thermal hybrid solar collectors, known as hybridVT systems, are systems that simultaneously convert solar radi-tion into thermal and electrical energy. These systems combine

photovoltaic cell, which converts electromagnetic radiation

dings 77 (2014) 236–246 239

(photons) into electricity, with a solar thermal collector, which cap-tures the remaining energy and removes waste heat from the PVmodule. It is well-known that conventional PV cells suffer from adrop in efficiency with the rise in temperature. Hybrid PVT Systemscan be engineered to carry heat away from the PV cells thereby cool-ing the cells and improving their efficiency by lowering the heatresistance [22,23]. Although this is an effective method, it causesthe thermal component to under-perform compared to a full-scalesolar thermal collector. Recent research showed that photovoltaicmaterials with low temperature coefficients such as amorphoussilicon (a-Si:H) PV allow the PVT to be operated at high tempera-tures, creating an even more synergetic PVT system [24]. The hotair stream (stream 9) that is coming from hybrid PVT system wouldbe sufficient to remove water vapour molecules and regenerate thewheel for the next round of moisture uptake. In this process, theregeneration air stream will also lose part of its enthalpy to thewheel and moderate its temperature. The moderate humid regen-eration air exiting from the wheel (stream 6) is guided to a heatexchanger where it can be further cooled down by another streamof fresh air (stream 4). In a well-designed heat exchanger, due toenthalpy reduction, the major portion of water vapour moleculesare condensed. The exiting relatively drier air stream (stream 7) ispumped back to the PVT system for reheating (stream 8). The con-densed water is collected and drained out to a condensate collectorusing a small pump.

3. Mathematical framework

3.1. Water capture

Most thermodynamic properties of moist air are tempera-ture related values which means they change with the ambienttemperature variation. Tsilingiris [25] simulated thermo-physicaland transport properties of moist air for temperature range of(0–100) ◦C at (0–100)% relative humidity and developed a polyno-mial correlation for each property. Tsilingiris’s correlations werecombined to calculate the inlet and outlet thermodynamic prop-erties such as moist air specific heat capacity (Cpma) in a distinctalgorithm inter-linked to this model. Fig. 4 shows the equivalentmodel built in TRNSYS that matches the experimental layout ofFig. 3. ASHRAE [26] empirical equations are also used to deter-mine the enthalpy of moist air. At the entrance to the system, theenthalpy of moist air is combined from the enthalpy of dry air andwater vapour at a temperature that is equivalent to T1 (Eq. (1)):

hma = hda + ω1.hwv (1)

where hma is the moist air enthalpy (kJ/kg); ω1 is the humidity ratio(kgwv/kgda); hda is the dry air enthalpy which can be calculated inEq. (2) and hwv is water vapour enthalpy that is calculated in Eq. (3)[26]:

hda ≈ CpdaT (2)

hwv ≈ 2501 + 1.805T (3)

The specific volume �1, absolute humidity dv, moist air density�ma, volumetric moist air flow Fvma, dry air density �da, specifichumidity �1 and dry air mass flow rate Fda are determined fromfundamental ASHRAE equations (Eqs. (4)–(10)) [26]:

�1 = 0.2871(T1 + 273.15)(1 + 1.6078ω1)P1

(4)

ω

dv = 1

�1(5)

�ma = 1 + ω1

�1(6)

Page 5

240 D. Milani et al. / Energy and Buildings 77 (2014) 236–246

idific

F

�

�

Fig. 3. Schematic of the proposed dehum

˙ vma = Fma

�ma(7)

da = P1

Rda(T1 + 273.15)(8)

1 = ω1

1 + ω1(9)

Fig. 4. TRNSYS model of the

ation system integrated with solar PVT.

Fda = Fma(1 − �1) (10)

Once all these properties at the inlet and outlet of the heat

exchanger were calculated, condensate flow rate mw can be deter-mined in Eq. (11). The difference in humidity ratios of the inlet andoutlet (ω1 − ω2) are the determining factors of condensate flowrate. The efficiency of dehumidifier �d is included in Eq. (11) to

experimental layout.

Page 6

d Buil

ap

m

3

PahoadpdPtwctaebhs

boilPletsoideFc“mDEtdmtIaelstu9twracba[

D. Milani et al. / Energy an

ccount for the condensate wasted in the heat exchanger geometry,ipe wetting or due to water re-evaporation

˙ w = Fda(ω1 − ω2)�d (11)

.2. Solar hybrid PVT

The total energy output (electrical plus thermal) of a hybridVT system depends on the solar energy input, ambient temper-ture, wind speed, operating temperature of the system and theeat extraction mode. The electrical output is of priority and theperating conditions of the system thermal unit must be adaptedccordingly. Van Helden et al. [27] observed that 80% of the inci-ent solar radiation is absorbed by PV collectors but only a smallortion of this is converted to electrical energy and the remain-er is dissipated as thermal energy. He et al. [28] used the hybridVT collector technology using water/air as the coolant as a solu-ion for improving the energy performance. The efficiency of a PVTater/air collectors depends upon several key parameters such as

oolant mass flow rate, specific heat and the geometry of the collec-or. The possibility of generating electricity and heat energy from

commercial PV module adopted as a PVT air solar collector withither forced or natural air flow in the channel was demonstratedy Tonui and Tripanagnostopoulos [29]. Nayak and Tiwari [30] alsoave made energy and exergy analysis of PVT integrated with aolar greenhouse.

In TRNSYS, the type 50 module conceptualises a PV module com-ined with the standard flat-plate collector. It describes the usef the modified version for combined collector simulation whichncorporates both the work of Florschuetz [31] for flat plate col-ectors operated at peak power, and the analysis of combinedhotovoltaic/Thermal System [32] for concentrating combined col-ectors. The circuit of a PV unit employs equations for an empiricalquivalent circuit model to predict the current–voltage charac-eristics of a single module. This circuit consists of a DC currentource, diode, and either one or two resistors. The power outcomef the current source is totally dependent on solar radiation ands temperature-dependent because of the IV characteristics of theiode. The results for a single module equivalent circuit are thenxtrapolated to predict the performance of a multi-module array.or Mono-Crystalline Silicon modules (similar to those used in thisase study), Type 94 is the most appropriate unit that employs afour-parameter” equivalent circuit. The “four parameters” in theodel are: Module photocurrent at reference conditions (IL,ref),iode reverse saturation current at reference conditions (Io,ref),mpirical PV curve-fitting parameter (�e), and Module series resis-ance (Rs). These “four parameters” are empirical values whichefine an equivalent circuit and are employed to find the PV perfor-ance at each time step. Type 94 calculates these values based on

ypical manufactures’ catalogue data. Three of these values (IL,ref,o,ref, �e and Rs) may be isolated algebraically and type 94 usesn iterative search routine in these four equations to calculate thequivalent circuit characteristics. The first step is to set upper andower bounds for the series resistance parameter Rs. Physical con-traints require the Rs value to lie between 0 and the value suchhat �e = Ns; where Ns is the number of individual cells in the mod-le. The initial guess for Rs is midway between these bounds. Type4 also includes an optional incidence angle modifier correlationo calculate how the reflectance of the PV module surface variesith the angle of incidence of solar radiation. It determines PV cur-

ent as a function of load voltage and other outputs such as currentnd voltage at the maximum power point along the IV curve, open-

ircuit voltage, and short circuit current. This model was developedy Townsend [33] and also detailed by Duffie and Beckman [34]nd was initially incorporated as a TRNSYS component by Eckstein35]. The “four-parameter” model assumes that the slope of the

dings 77 (2014) 236–246 241

IV curve at short-circuit conditions is zero which is a reasonableapproximation for crystalline modules

(dI

dV)V=0

= 0 (12)

The IV characteristics of a PV panel change with both insola-tion and temperature. The PV model employs these environmentalconditions along with the four module constants IL,ref, Io,ref, �e andRs to generate an IV curve at each time step. The current–voltageequation of the circuit is as follows:

I = IL − I0

[exp(

q

�ekTc(V + IRs)

)− 1]

(13)

where I is the current (Amp); I0 is diode reverse saturation cur-rent (Amp); k is Boltzmann constant (J/K); q is the electron chargeconstant; Tc is the module temperature (◦C); IL is the module pho-tocurrent and depends linearly on the solar incident radiation:

IL = IL,ref

(GT

GT,ref

)(14)

where GT is the total radiation incidence on a PV array; GT,ref isthe total radiation incidence at reference insolation and is usuallydefined to be equal to 1000 W/m2. The diodes reverse saturationcurrent Io is a temperature dependent quantity

I0I0,ref

=(

Tc

Tc,ref

)3

(15)

where Io,ref and Tc,ref are the saturation current and module temper-ature at the reference point respectively. Eq. (13) gives the currentimplicitly as a function of voltage. Once I0 and IL are found fromEqs. (14) and (15), Newton’s method is employed to calculate thePV current. In addition, an iterative search routine finds the cur-rent (Imp) and voltage (Vmp) at the point of maximum power alongthe IV curve. Type 94 also includes an optional incidence anglemodifier correlation to calculate how the reflectance of the PV mod-ule surface varies with the angle of incidence of solar radiation.Type 94 determines PV current as a function of load voltage andother outputs such as current and voltage at the maximum powerpoint along the IV curve, open-circuit voltage, and short circuitcurrent

ς = Qu

AIT= mf Cpf (To − Ti)

AIT= FR(�˛)n − FRUL

(Ti − Ta)IT

(16)

where A is the total collector array aperture or gross area (m2);Qu is the total thermal gain by the collector (kJ/h); mf is the massflow rate of the circulating fluid (kg/h); Cpf is specific heat capacitycoefficient of circulating fluid (kJ/kg K); IT is the global radiationincident on the solar collector (kJ/h m2); FR is the overall collectorheat removal efficiency factor, (�˛)n is the product of the covertransmittance and the absorber absorptance at normal incidence;UL is overall thermal loss coefficient of the collector per unit area(kJ/h m2 K); Ti, To and Ta are the temperature of the inlet, outletand ambient respectively (◦C). The loss coefficient UL is not exactlyconstant, so a better expression is obtained by taking into accounta linear dependency of UL versus (Ti − Ta):

ς = FR(�˛)n − FRUL(Ti − Ta)

IT− FRUL/T

(Ti − Ta)2

IT(17)

ς = a0 − a1(T)

IT− a2

(T)2

IT(18)

where UL/T is the thermal loss coefficient dependency on T

(kJ/h m2 K2); a0 is the optical efficiency of the collector; a1 is thefirst-order coefficient in collector efficiency equation (kJ/h m2 K);a2 is the second-order coefficient in collector efficiency equation(kJ/h m2 K2). Eq. (18) is the general solar collector thermal efficiency

Page 7

242 D. Milani et al. / Energy and Buildings 77 (2014) 236–246

r tank

etpa

Toat

Fig. 5. (a) Dehumidification system connected to the wate

quation that is commonly used. In Eq. (18), T is often definedo be equal to (Ti − Ta). However, collector test reports sometimesrovide the efficiency curve using different references for temper-ture difference:

T =

⎧⎪⎨⎪⎩

Ti = Ti − Ta

Tav = Tav − Ta

To = To − Ta

he fist formulation is usually preferred in the US, while the sec-nd one is used in most European documents. If the coefficientsre given in terms of the average or the outlet temperature, correc-ion factors are applied. Those correction factors have been derived

Fig. 6. Schematic of the exper

that is fastened on the load cell; (b) weather instruments.

for linear efficiency curves and Eq. (18) must first be converted tothat form by performing a modified first-order collector efficiencycoefficient defined as:

U ′L = UL + UL/T (Ti − Ta) (19)

That gives:

ς = Qu

AIT= FR(�˛)n − FRU ′

L(Ti − Ta)

IT(20)

iment layout (phase 1).

Page 8

D. Milani et al. / Energy and Buildings 77 (2014) 236–246 243

re, rel

[

F

F

wiTf

F

F

Foat

r

r

F

4

drttww(

to most coastal cities, the humidity levels are often high espe-cially at night times and early mornings. The average ambienttemperature of the experiment period was 14.2 ◦C while rela-tive humidity average was 64.8%. This possibly makes desiccant

Fig. 7. Recorded barometric pressure, ambient temperatu

The correction factors are then given by Duffie and Beckman34]:

R(�˛) = Fav(�˛)n

(mtestCpf

mtestCpf + FavU′L

2

)(21)

RU ′L = FavU ′

L

(mtestCpf

mtestCpf + FavU′L

2

)(22)

here mtest is the fluid flow rate in test conditions (kg/h) and Fav

s the modified value of FR when the efficiency is given in terms ofav instead of Ti. For outlet temperature references, the correctionactor will also be:

R(�˛) = Fo(�˛)n

(mtestCpf

mtestCpf + FoU ′L

)(23)

RU ′L = FavU ′

L

(mtestCpf

mtestCpf + FoU ′L

)(24)

or conditions when the collector is operated at a fluid flow ratether than the value at which it was tested, both FR(�˛)n and FRU ′

Lre corrected to account for changes in FR. The ratio r1, by whichhey are corrected, is given by:

1 = FR(�˛)n|use

FR(�˛)n|test(25)

1 =

mf CpfAF ′UL

(1 − e

− AF ′ULmf Cpf

)|use

mtestCpfAF ′UL

(1 − e

− AF ′ULmtest Cpf

)|test

(26)

In which F′UL is equal to:

′UL = mf Cpf

Aln

(1 − FRU ′A

mf Cpf

)(27)

. Experimental lay-out

To validate the mathematical framework, a prototype desiccantehumidifier (Comdry M170L, Munters) was used. The humidityatio set-point ωset was lowered to the minimum value forcinghe dehumidifier to work in full capacity. The condensed water in

he heat exchanger was drained and periodically pumped out to aater tank fastened to an electrical load cell (Fig. 5a). The collectedater was timely weighed and the data transferred to a PLC system

Stardom, Yokogawa) via a strain gauge bridge. Local weather data

ative humidity and collected water over 9 days in Sydney.

also were transferred to the PLC using (4–20 mA) output weatherinstruments (Fig. 5b). Consumed energy for dehumidification pro-cess was met by electrical power outlet and logged to the PLC viaa Power data logger. The experiment was set to quantify collectedwater and the consumed energy over time. As the dehumidifica-tion system used in this experiment had a built-in electrical heaterused for regeneration purpose, the consumed electrical energy waslogged for later comparison with solar thermal gain projected in themodel. Fig. 6 shows a schematic of the final project lay-out wherethe electronic signals from these instruments were routed to thePLC box that communicated with the computer.

5. Results and discussion

The experiment was run continuously over 9 days at the endof an Autumn season. The data for local barometric pressure, rel-ative humidity and ambient temperature were transmitted to thePLC system and then digitally interpreted and presented in Fig. 7.In moderate climates such as Sydney, air-conditioning and cool-ing systems are running less times compared with hot humidregions (i.e. Middle East Gulf region). In general, this is becausehigher ambient temperature during the day is often associatedwith lower relative humidity level as shown in Fig. 7. Similar

Fig. 8. Comparison between experimental and calculated cumulative water con-densate over 9 days period.

Page 9

244 D. Milani et al. / Energy and Buildings 77 (2014) 236–246

Fig. 9. The plot of hourly relative humidity, temperature, water condensation and weekly cumulative water collection for Sydney, Abu Dhabi and London, respectively.

Fig. 10. The variation of temperature and relative humidity of the proposed cabin indoor and outdoor over the 9 days period.

Page 10

D. Milani et al. / Energy and Buildings 77 (2014) 236–246 245

Table 1Sky water quality versus distilled and tap water. Values in brackets are Sydney waterfigures for water from Prospect filtration plant.

Parameter Tap water Distilled water Sky water

PH 7.4 (7.7–8.0) 5.90 5.50Conductivity (mS/m) 39.5 (19–22) 0.40 0.23

W(

tnvipilthahatarrpTSipoeermFgatidcwttcpsteadttsL

Piteptm5

occupants. Nonetheless, this amount of power is unable to meet

ater analysis: typical water analysis for Sydney Water’s drinking water supplywww.sydneywater.com.au, accessed 21.06.12).

echniques more water productive compared with cooling tech-iques as desiccants are less affected by ambient temperatureariations. Fig. 8 also shows the slope of water collection trendncrease with relative humidity level and is less affected by tem-erature variation. The cumulative mass of condensate over 9 days

s presented in Fig. 8. The equivalent model for the experimentayout is developed in two phases. In phase one, the solar collec-or is replaced by built-in electrical auxiliary heater that supplieseat for the regeneration stream (stream 9) to continuously achievedequate regeneration at the dehumidification wheel. This wouldelp identify a model that would have the best water productivitygreement with the experimental results. Phase one also charac-erised other parameters such as the ratio between heat exchangernd process air volumetric flow rates (stream 4/stream 2) and theatio between the regeneration and process air volumetric flowates (stream 9/ stream 2). Fig. 8 compares the cumulative waterroductivity of the experimental setup with those calculated inRNSYS model over the same period. The calculated values by TRN-YS model show a slight variation from actual collected water. Thiss probably because of lost water droplets due to re-evaporation oripe wettings. However, the TRNSYS model gives good estimationf the total collected water with total error of ≤1% for the entirexperiment period. From there, dehumidification technical param-ters such as the dehumidification effectiveness �deh, the moistureemoval capacity (MRC), the dehumidification coefficient of perfor-ance (COPd) and sensible energy ratio (SER) can also be calculated.

ig. 9 shows that this small dehumidification system is able toenerate approximately 52 litres of freshwater in 9 days which islmost 6 litres per day. Samples of the condensate ‘Sky water’ wereaken from the water collection tank for water quality analysis. Thisnvolved measurement of the relevant parameters of pH and con-uctivity. In this analysis tap water and distilled water were alsoharacterised for comparative purposes. Table 1 shows that Skyater is slightly more acidic than tap and distilled water. This is

otally expected because of the presence of acidic depositions inhe atmosphere of most modern cities. However, the low value ofonductivity of the sky water indicates that it is less affected by theresence of inorganic dissolved solids such as chloride, sulphate,odium, calcium and others. Applying this model for different loca-ions with various climate conditions can also give an appropriatestimate of annual water productivity. In Fig. 9, this model waspplied for the meteorological data of three different locations withissimilar climates; Sydney, Abu Dhabi and London. Fig. 9 showshat Abu Dhabi posses the highest rate of potential water collec-ion that reach a cumulative value of up to 18.5 kL a year. For theame dehumidification system, Sydney can generate up to 13.8 andondon up to 10 kL of water a year.

In phase two, the artificial auxiliary heater is replaced by solarVT collectors (Fig. 4). A total gross area of 5 mono-crystalline sil-con PV panels with a gross area equal to 10 m2 flat mounted onhe solar cabin rooftop (tilt angle equal to 0◦). The role of solarnergy contribution for both electrical and thermal applications isrojected. It is anticipated that electrical outcome will supply elec-

rical energy to a small air-conditioning unit if required (900 Watts),

ini refrigerator (90 Watts), computer and printer (230 Watts) and Compact Fluorescent integral ballast lamps (total 90 Watts) for

Fig. 11. Electrical and thermal gain from 10 m2 PVT array flat mounted on the cabinrooftop.

the cabin. Excessive solar power will be stored in 6 cell batter-ies connected in series to serve at night time or when the solarresource is not sufficient. In this phase, a detailed model of type56 is projected in the TRNBuild platform. In this model, a syntheticcabin with similar positioning, geometry and solar radiation sim-ilar to the actual cabin is projected and all parameters related tofiltration, ventilation, gain, comfort, heating and cooling require-ments are estimated. Heat gain is hourly calculated based on 1–2personnel and varying 0–4 visitors during the day. Other variablessuch as filtration, ventilation and cooling requirements are also cor-related to the number of people who occupy the cabin at a giventime. In this scenario, it is assumed that power demand is onlymet by rooftop solar panels. Setting the target temperature insidethe cabin to 25 ◦C, Fig. 10 shows that while the relative humidityinside the cabin can be significantly reduced to suit human com-fort level (≤60%), the set temperature cannot be always met dueto the limited power generation out of the designated solar pan-els. In this regard, the temperature of dehumidified air coming outof the dehumidification system is often higher than it enters and,therefore, posses more sensible load on the air-conditioner. Thisindicates that the energy consumed on dehumidification system topre-treat feed air for the air-conditioner, in fact has two side effects.While this energy can reduce the latent load of feed air significantly,it also increases the sensible load. The latter part is beneficial onlyin cold seasons where heating is required. The overall performanceis fundamentally dependant on the equilibrium of latent-sensibleloads and that can only be optimised for site specific parametersfor a particular season or the entire year. It is also notable that gen-erated solar power from the rooftop of the cabin is limited andcannot always meet the air-conditioning power demand for theprojected period. Fig. 11 shows that the final output power from theinverter can only reach approximately 500 Watts in the peak hours.However, by using a suitable set of batteries, the excess stored elec-trical power can meet the demand of lighting, mini refrigerator,computer and printer for extended hours based on usage condi-tions. Additionally if the regeneration thermal demand is met byPVT thermal gain, excessive electrical power can also run the restof electrical power demand of the air-cooled dehumidifier for thetarget of moderating the humidity level of the cabin indoor andgenerating an adequate amount of water for the survival of the

the air-conditioning unit (900 watts) requirements at the currentsettings. The thermal gain is found to be proportional to electricgain and can reach up to 68% of the electrical gain in peak hours.

Page 11

2 d Buil

6

awdnaFid

1

2

3

doi1odoptapb(aridIflcfca

A

rc

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[[

[

[35] J.H. Eckstein, Detailed modeling of photovoltaic components, Solar Energy Lab-

46 D. Milani et al. / Energy an

. Conclusions

Atmospheric water generation is considered to be a sustain-ble and environmentally benign process. In emergency situationshere the utility and water networks are often damaged andisrupted, dehumidification techniques might help to generateecessary water and/or air-conditioning. The hurdle of energyvailability might also be overcome by facilitating solar energy.or the scenario of using small field clinics in self-contained cab-ns supplied by rooftop flat mounted solar PVT arrays, a small-scaleehumidification unit can be installed for the following objectives:

. Generating water from the atmosphere for survival and/or san-itation purposes.

. Reducing the latent load by pre-treating the air feed stream ofan air-conditioning system.

. Improving the PV efficiency by removing the resistant heat layerat the back of PV panels.

It was found that a small-scale low energy-consumer silica gelehumidification system can generate more than 5.2 litres a dayf high quality freshwater in Sydney. A simulation model builtn TRNSYS calculated that Sydney can generate a cumulative of3.8 kL a year without further optimisation for a particular seasonr geometry. Similarly, Abu Dhabi showed the highest water pro-uctivity of 18.5 kL and London the lowest water productivity ofnly 10 kL a year. Most of the required energy for the regenerationrocess of dehumidification wheel can be met by thermal gain ofhe solar hybrid PVT array during the day. The outlet processedir (dehumidified air stream) incurs a smaller latent load on aroposed small-scale air-conditioning system which can help toring the humidity level inside the cabin within human comfort≤60% relative humidity) but experiences a rise in the processedir temperature which increases the sensible load. The equilib-ium of latent-sensible load will only be meaningful if the systems optimised for a particular season (summer or winter) and pre-efined based on site specific parameters and occupancy patterns.

t is also found that 5 mono-crystalline silicon photovoltaic panelsat mounted on the rooftop can meet the energy demand of theabin proposed lighting, mini refrigerator, computer and printeror typical operating hours but is unable to run the proposed air-onditioner (900 Watts) most of the time, consequently requiringn additional energy source.

cknowledgment

This project is partially funded by the NSW Environmental Trust,esearch seed fund grant (2009/RDS/0034). The solar poweredabin is kindly provided by Blue Planet Buildings.

eferences

[1] The International Disaster Database EMDAT, Disaster trends and relationshipperiod 1900–2010, July 2009 http://www.emdat.be/natural-disasters-trends

[2] A.Z. Keller, Engineering in emergencies: a practical guide for relief workers –Davis, J., Lambert, R., Disasters 21 (March (1)) (1997) 92.

[3] J.J. McCarthy, O.F. Canziani, N.A. Leary, D.J. Dokken, S. White, Climate change2001: impacts, adaptation, and vulnerability, in: Contribution of WorkingGroup II to the Third Assessment Report of the Intergovernmental Panel onClimate Change, Technical report, Cambridge University Press, 2001.

[4] P.H. Gleick, Global freshwater resources: soft-path solutions for the 21st cen-

tury, Science 302 (November (5650)) (2003) 1524–1528.

[5] A. Khalil, Dehumidification of atmospheric air as a potential source of fresh-water in the UAE, Desalination 93 (August (1–3)) (1993) 587–596.

[6] B. Hellström, potable water extracted from the air – report on laboratory exper-iments, Journal of Hydrology 9 (1) (1969) 1–19.

[

dings 77 (2014) 236–246

[7] H.E. Landsberg, Beneficial modifications of the marine environment, in:Proceedings of Symposium Sponsored by National Research Council & Depart-ment of the Interior, Washington, DC, National Academy of Sciences, March1972, pp. 78–81.

[8] V.P. Starr, D.A. Anati, D.A. Salstein, Effectiveness of controlled convectionin producing precipitation, Journal of Geophysical Research 79 (27) (1974)4047–4052.

[9] A.K. Rajvanshi, Large-scale dew collection as a source of fresh-water supply,Desalination 36 (3) (1981) 299–306.

10] T.H. Elmer, J.F. Hyde, Recovery of water from atmospheric air in arid climates,Separation Science and Technology 21 (3) (1986) 251–266.

11] R.V. Wahlgren, Atmospheric water vapour processing, Waterlines 12 (2) (1993)20–22.

12] T.M.J. Nilsson, W.E. Vargas, G.A. Niklasson, C.G. Granqvist, Condensation ofwater by radiative cooling, Renewable Energy 5 (1–4) (Aug 1994) 310–317.

13] V.S. Nikolayev, D. Beysens, A. Gioda, I. Milimouk, E. Katiushin, J.P. Morel, Waterrecovery from dew, Journal of Hydrology 182 (1–4) (Jul 1996) 19–35.

14] D. Beysens, I. Milimouk, V. Nikolayev, M. Muselli, J. Marcillat, Using radiativecooling to condense atmospheric vapor: a study to improve water yield, Journalof Hydrology 276 (1–4) (May 2003) 1–11.

15] W. Chakroun, N. Ghaddar, K. Ghali, Chilled ceiling and displacement ventila-tion aided with personalized evaporative cooler, Energy and Buildings 43 (11)(2011) 3250–3257.

16] D. Milani, A. Abbas, A. Vassallo, M. Chiesa, D. Al Bakri, Evaluation of using ther-moelectric coolers in a dehumidification system to generate freshwater fromambient air, Chemical Engineering Science 66 (12) (2011) 2491–2501.

17] J.W. Mitchell, J.E. Braun, Design Analysis, and Control of Space ConditioningEquipment and Systems, University of Wisconsin, Madison, 1997.

18] D.J. Houghton, R.C. Bishop, A.B. Lovins, B.L. Stickney, J.J. Newcomb, B.J. Davids,State of the Art Technology Atlas: Space Cooling and Air Handling, ESource, Inc.,Boulder, CO, 1992.

19] M.E. Hourani, K. Ghali, N. Ghaddar, Effective desiccant dehumidification systemwith two-stage evaporative cooling for hot and humid climates, Energy andBuildings 68 (Part A (0)) (2014) 329–338.

20] H.M. Henning, Solar-Assisted Air-Conditioning in Buildings: A Handbook forPlanners, Springer, New York, 2004.

21] S. Matsushita, S. Yamaguchi, K. Saito, K. Hongo, K. Inagaki, Performance eval-uation of open type absorption dehumidifier using LICL solution as liquiddesiccant, in: Proceedings International Sorption Heat Pump Conference, Den-ver, USA, 2005.

22] S.A. Kalogirou, Y. Tripanagnostopoulos, Hybrid PV/T solar systems for domestichot water and electricity production, Energy Conversion and Management 47(November (18–19)) (2006) 3368–3382.

23] D. Kamthania, S. Nayak, G.N. Tiwari, Performance evaluation of a hybrid pho-tovoltaic thermal double pass facade for space heating, Energy and Buildings43 (9) (2011) 2274–2281.

24] M.J.M. Pathak, J.M. Pearce, S.J. Harrison, Effects on amorphous silicon photo-voltaic performance from high-temperature annealing pulses in photovoltaicthermal hybrid devices, Solar Energy Materials and Solar Cells 100 (May) (2012)199–203.

25] P.T. Tsilingiris, Thermophysical and transport properties of humid air at tem-perature range between 0 and 100 ◦C, Energy Conversion and Management 49(May (5)) (2008) 1098–1110.

26] ASHRAE, The ASHRAE Handbook CD, American Society of Heating RefrigeratingAir-conditioning Engineers, Atlanta, 2002.

27] WG.J. van Helden, RJ.C. van Zolingen, H.A. Zondag, PV thermal systems: PVpanels supplying renewable electricity and heat, Progress IN Photovoltaics 12(September (6)) (2004) 415–426.

28] W. He, T.T. Chow, J. Ji, J.P. Lu, G. Pei, L.S. Chan, Hybrid photovoltaic and thermalsolar-collector designed for natural circulation of water, Applied Energy 83(March (3)) (2006) 199–210.

29] J.K. Tonui, Y. Tripanagnostopoulos, Improved PV/T solar collectors with heatextraction by forced or natural air circulation, Renewable Energy 32 (April (4))(2007) 623–637.

30] S. Nayak, G.N. Tiwari, Energy and exergy analysis of photovoltaic/thermalintegrated with a solar greenhouse, Energy and Buildings 40 (11) (2008)2015–2021.

31] L.W. Florschuetz, Extension of the Hottel–Whillier model to the analysis ofcombined photovoltaic thermal flat-plate collectors, Solar Energy 22 (4) (1979)361–366.

32] Report ERC-R-78017, Combined Photovoltaic/Thermal System Studies, 1978.33] T.U. Townsend, A method for estimating the long-term performance of

direct-coupled photovoltaic systems, Solar Energy Laboratory, University ofWisconsin, Madison, 1989 (Master’s thesis).

34] J.A. Duffie, W.A. Beckman, Solar Engineering of Thermal Processes, 3rd ed., JohnWiley & Sons, Inc., Hoboken, NJ, 2006.

oratory, University of Wisconsin, Madison, 1990 (Master’s thesis).36] D. Milani, Modelling framework of solar assisted dehumidification system to

generate freshwater from “Thin Air”, The University of Sydney, 2011 (PhD the-sis).