ra

vers

Ice-making

ad byntri-amomr re68,

ling insupplountriear ton. Thng areResear

order to reduce capital costs and make such systems commerciallyviable [1,2]. The sorption generator model has been validatedagainst experimental data from a gas red heat pump [3,4].Therefore, it is possible to explore with condence other sorptionapplications using the current model.

capacity (below 100 kg ice per day) and high cooling capacity(above 100 kg ice per day). The use of high pressure ammonia as therefrigerant is also favourable in comparison to sub-atmosphericwater or methanol for which the ingress of air can be difcult toprevent.

This paper presents simulation results for the application ofthis low cost sorption generator to a solar powered refrigerationsystem. Fromweather data (mainly ambient temperature and solarinsolation) provided by the Meteonorm package, simulations are

* Corresponding author. Tel.: 44 24 76522108; fax: 44 24 76418922.

Contents lists availab

Applied Therma

sev

Applied Thermal Engineering 31 (2011) 2197e2204E-mail address: z.tamainot-telto@warwick.ac.uk (Z. Tamainot-Telto).solid sorption refrigeration systems has been active since the late1970s. However, such systems have not been commercialised dueto their high capital cost. For vaccine storage a high capital cost maybe acceptable; however in food preservation or comfort air condi-tioning the capital costs must be reduced dramatically in order forsuch systems to become viable. Researchers at the University ofWarwick have developed a prototype plate heat exchanger sorptiongenerator for use in carbon-ammonia adsorption systems. Thissorption generator is designed to increase specic cooling power(SCP, the cooling power per unit mass or volume of machine) in

countrys exports. The adsorption system is highly suited to appli-cation in remote areas and developing countries, as it containsrelatively few moving parts and thus requires little maintenance.The adsorption system is driven by heat from conventional solarthermal collectors: the system could provide cooling requirementsfor sh preservation (typically 1 kg of ice per kg of sh) assustainable technology that is complementary or a substitution toa conventional refrigeration machine. Furthermore such systemsoff-load the electrical grid since there is little or no electrical powerrequirement. This technology is suitable for both low cooling1. Introduction

There is a large demand for coowhere there is no reliable electricityor too expensive too obtain. Such csolar insolation, and so would appethe application of solar refrigeratiowhich there is a demand for coolipreservation and air conditioning.1359-4311/$ e see front matter 2010 Elsevier Ltd.doi:10.1016/j.applthermaleng.2010.11.001developing countriesy and fuels are difcults tend to receive highbe ideal candidates fore three main areas invaccine storage, food

ch into solar powered

The city of Dakar and its surrounds hosts about 95% of theindustrial shery activities of Senegal (West Africa) ranging fromsh processing to ordinary stock storage for local consumption orexportation. Both artisanal and industrial shery activities countfor about 2.5% of GDP. The sector provides about 600,000 jobsand the sh contribute up to 70% of animal protein consumption(26 kg per person, per year) [5,6]. In 2005, the sh captureproduction was estimated to be about 400,000 tonnes with 30% forexport with a value of aboutV270M representing around 40% of theRefrigerationApplication of a compact sorption geneDakar (Senegal)

S.J. Metcalf, Z. Tamainot-Telto*, R.E. CritophSustainable Energy Engineering and Design Research Group, School of Engineering, Uni

a r t i c l e i n f o

Article history:Received 1 September 2009Accepted 1 November 2010Available online 5 November 2010

Keywords:Active carbonAdsorptionAmmonia

a b s t r a c t

The feasibility of applyingrefrigeration is investigatemaking in developing couemploys the active carbonDriving heat is provided fra one-off machine with fou(Senegal) is estimated at V

journal homepage: www.elAll rights reserved.tor to solar refrigeration: Case study of

ity of Warwick, Coventry, CV4 7AL, UK

low cost plate heat exchanger solid sorption reactor to solar poweredusing a validated reactor model. The proposed system is targeted at ice-

es for food preservation. The adsorption refrigeration machine modelledmonia working pair in both two-bed and four-bed regenerative systems.standard at plate and evacuated tube solar collectors. The capital cost ofgenerative beds which could produce up to 1000 kg of ice per day in Dakar000.

2010 Elsevier Ltd. All rights reserved.

le at ScienceDirect

l Engineering

ier .com/locate/apthermeng

a standard solar collector should be themost cost effective solution.

Greek lettersh Efciencyl Thermal conductivity (W m1 K1)D Difference

Subscriptsa Adsorbed phase; Ammoniaamb Ambientc Carbon

al Engineering 31 (2011) 2197e2204Details of the two types investigated are included in Table 1.The adsorption system does not operate on a diurnal cycle as

with most solar powered adsorption ice makers, but operates witha much shorter cycle time (circa 3 min) and produces ice duringcarried out for a complete year for the city of Dakar and the capitalcost of a solar sorption cooling system producing up to 1000 kg ofice per day is estimated.

2. Solar refrigeration system

The proposed system consists of a multiple-bed regenerativeadsorption system driven by heat from solar thermal collectors viaa heat transfer uid. Two solar collector types are considered: a atplate type and an evacuated tube type, in order to compare theircapital cost per unit cooling. Since the solar collector array isoften the most signicant part of the total system cost, the use of

Nomenclature

A Surface area (m2)c Specic heat (J kg1 K1)COP Coefcient of performanceG Solar radiation incident on collector (W m2)H Specic heat of sorption (J kg1)k Dubinin coefcientm Mass ow rate (kg s1)M Mass (kg)n Dubinin coefcientL Latent heat of vaporisation (J kg1)q Specic cooling production (J kg1)Q Heat (J)_Q Heat ux (W)SCP Specic cooling power (W kg1)t Thickness (m); Time (s)T Temperature (K)U Overall heat transfer coefcient (W m2 K1)x Concentration (kg Ammonia kg1 Carbon)

S.J. Metcalf et al. / Applied Therm2198daylight hours. This minimises the size and therefore the cost of theadsorption system. Both a two-bed and a four-bed regenerativeadsorption cycle with mass recovery are investigated. Although thefour-bed cycle has a higher COP, SCP is lower and so a larger andthus more expensive machine is required. If the adsorption systemcost were negligible in comparison to the solar collector cost thenthe four-bed system would always prove superior, since its higherefciency would minimise the required collector area. However,the adsorption machine cost is unlikely to be negligible in all casesand therefore the economics of the two cycles needs to beinvestigated.

The plate heat exchanger sorption generator is shown in Fig. 1. Itis a prototype unit which is currently under development and is tobe applied to adsorption systems for gas red heat pumping and airconditioning and mobile air conditioning.

The unit is constructed from nickel brazed stainless steel andcontains carbon adsorbent in 4 mm thick layers. The heat transferuid is pumped through chemically etched channels in the stain-less steel shims which form the plates of the heat exchanger. Therelatively thin layers of adsorbent and the large area for uid heattransfer promote rapid temperature cycling and a high SCP. Similarunits may be employed in the numerous applications outlined,thereby enabling economies of scale in their manufacture andfurther lowering the capital cost of the system.The cost of such a heat exchanger is difcult to estimate at thisearly stage in its development. However, it seems reasonable toestimate it from the cost of an existing conventional plate heatexchanger with the same size and number of plates. The onlyconceivable additional cost involved in the manufacture of the unitwould be that of the manufacture and insertion of the adsorbent.The cost of the carbon adsorbent itself will be negligible and thecost of adsorbent insertion in the heat exchanger will be considerednegligible for the purposes of this analysis. Table 2 gives the costof a stainless steel brazed plate heat exchanger with plates ofsimilar dimensions to the sorption generator, which will be used inthe cost estimates.

Fig. 2 presents an example of a layout of solar powered sorptionrefrigeration system. The ammonia loop consists of a condenser, anexpansion valve, an evaporator and a thermal compressor. Thisthermal compressor corresponds to two regenerative beds (withset of four check valves) and is driven by both solar collector andcooler.

col Collectorcool Coolingf Fluid (Liquid)in InletLM Log-Meano Opticalout Outletp Pressuresat Saturationv Volumew Wall3. System performance simulation

3.1. Sorption system model

The sorption system model was created in MATLAB and isa nite difference model. The schematic in Fig. 3 shows the cong-uration of the modelled generator bed.

The adsorbent is in contact with a wall that separates it fromthe heat transfer uid, which facilitates the heat transfer to theadsorbent. The model contains two heat transfer coefcientse thatbetween the uid and the wall, and that between the wall and the

Table 1Solar collector (Direct ow): cost and performance data.

Collector Type Cost(V/m2)

Opticalefciency,h0

Linear losscoefcient, k1(W m2K1)

Quadraticloss coefcient,k2 (W m2K2)

ViessmannVitosol100 S2.5

Flat plate 400 0.85 4.07 0.007

ThermomaxSolamax20 - TDS300

Evacuatedtube

636 0.769 1.61 0.0032

conductivity. The uid-wall heat transfer coefcient was calculatedfrom relationships given by Kays and London [7] for fully developedlaminar ow in rectangular channels. The ow was conrmed aslaminar by ensuring that the Reynolds number was less than the

Fig. 1. Prototype plate heat exchanger sorption generator.

Table 2Plate heat exchanger used in sorption system cost estimates.

Heat exchanger Cost (V) Plate size (m) Plate2

No of Cost per

S.J. Metcalf et al. / Applied Thermal Engineering 31 (2011) 2197e2204 2199carbon. Both the wall and adsorbent are isothermal, i.e. they areeach modelled by a single lumped element.

type area (m ) plates plate (V)

UK Heat exchangersSL23TL-AA-40

V220 0.312 0.076 0.0237 40 V5.5The thermal resistance of the separating wall is considerednegligible (since it is a highly conductive, thin metallic wall). Thethermal contact resistance between the wall and the carbon isaccounted for in the assumed value for the carbon thermal

Fig. 2. Layout of a solar powered sorptioncritical value. The control volume of the model has only a halfthickness of the overall carbon layer (tc) and the effective thermalconductivity (lc) includes the contact thermal resistance of thecarbon-wall. The wall-carbon overall heat transfer coefcienttherefore is calculated as:

Uwc lctc=2 2lctc

(1)

where lc is the effective thermal conductivity of the carbon(W m1 K1) and tc is the carbon layer thickness (m).refrigeration system (with two beds).

vxvt

vxvTc

P

vTcvt

(10)

since in an idealised cycle adsorption and desorption occur atconstant pressure. The concentration of adsorbed ammonia iscalculated using a modied DubinineAstakhov equation [8]:

x x0exp KTc=Tsat 1n (11)

and thus the rst partial derivative on the RHS of equation (10) canbe calculated analytically as:

vxvTc

P Knx

Tsat

TcTsat

1n1

(12)

In order to produce continuous cooling with better COP, at leasttwo beds are operating out of phase (180 rotation) with bothmass and heat recoveries between phases. For the mass recovery,the two beds are linked together through an ammonia loop until

al Engineering 31 (2011) 2197e2204Heat transfer between the uid and the wall is calculated usinga log-mean temperature difference

DTLMfw Tfin TfoutlnTfinTwTfoutTw

(2)

where DTLMfw is the log-mean temperature difference between theheat transfer uid and the wall. The governing equations for themodel are:

MwcpwvTwvt

UAfwTLMfw UAwcTw Tc (3)

Mf cpfvTfvt

_mcpfTfin Tfout

UAfwDTLMfw (4)

Mccpc xcpa

vTcvt

McHvxvt

UAwcTw Tc (5)

Where cpa is the specic of ammonia in adsorbed phase(4734 J kg1 K1), cpc is the specic of activated carbon and H is theheat of sorption. cpc (J kg1 K1) and H (J kg1) are given by thefollowing expressions:

cpc 175 2:245Tc (6)

H RA TcTsat

(7)

where R is the gas constant at the bed pressure P and temperatureTc (R 488 J kg1 K1); A is the slope of the saturated ammonia lineon the Clapeyron diagram (A 2823.4 K); Tc is the carbontemperature (K); Tsat is the saturation temperature corresponding

Fluid Wall Carbon AdsorbentTfin

Tf

Tfout

Tw x, Tc

(UA)wc(UA)fw

pfcm

tctwtf

Fig. 3. Schematic diagram of the sorption generator nite difference model.

S.J. Metcalf et al. / Applied Therm2200to the gas pressure (K).In the model the uid thermal mass is lumped with the wall

thermal mass e an approximation in order to simplify the solutionof the governing equations e so that equations (4) and (3) becomeequations (8) and (9) respectively.

_mcpfTfin Tfout

UAfwDTLMfw (8)

(corresponding to the effective heat rate input into uid-wallsystem during a xed cycle time)

Mwcpw Mf cpf

vTvt

UAfwDTLMfw UAwcTw Tc (9)

(corresponding to the effective heat rate input into the systemduring a xed cycle time). These equations are integrated throughtime using the explicit Euler scheme by substituting for vyvt inequation (5) with the following expression:Heat transfer uid thermalconductivity

0.14 W m K

1 1both saturating temperatures (Tsat1 and Tsat2 ) are equal. For the heatrecovery, the two beds are linked together through the coolantcirculation loop until the bed temperatures (TC1 and TC2) havea difference of 5 C, commonly called 5 C approach temperature.Both mass and heat recoveries have the effect of using the heatrejected by one bed during the cooling phase to pre-heat thesecond bed that is about to be heated: the effective heat requiredfor the heating phase is therefore reduced. The effective heat inputrequired per cycle Qin (J) is calculated after all temperature proles(carbon, wall and thermal uid) are established.

With the four-bed conguration, each bed operates with 90

phase with the neighbour bed with both mass and heat recoveriesbetween phases as previously with the two bed. To complete thefull four-bed cycle, this will require eight stages compared to fourwith the two-bed system [3]

The specic cooling production qcool (J kg1 carbon) is charac-terized by the latent heat of vaporisation of the ammonia liquidcollected during the condensation phase:

qcool Dx La (13)where Dx is the amount of liquid ammonia collected during thecondensation phase (kg ammonia/kg carbon) and La is latent heat ofvaporisation of the ammonia liquid (J kg1 ammonia. The coolingproduction Qcool (J) is therefore

Qcool Mcqcool (14)where Mc is the bed carbon mass (kg).

Table 3Modelled sorption generator parameters.

Parameter Value Unit

Carbon type Chemviron carbonSRD1352/3

e

Carbon thickness, tc 4 mmCarbon conductivity, lc 0.3 W m1 K1

Carbon density 435 kg m3

Limiting NH3 mass concentration, x0 0.5691 kg kg1

D-A equation coefcient k 6.6738D-A equation exponent n 1.1489Fluid channel depth, tf 0.5 mmWall thickness, tw 0.15 mmWall material 316 Stainless steelHeat transfer uid Mineral oil eHeat transfer uid density 800 kg m3

1 1

Heat transfer uid specic heat 2300 J kg K

capacity

The specic cooling power SCP (W kg1) corresponds to thespecic cooling power rate and is calculated from the followingexpression:

SCP qcoolDt

(15)

where: Dt is the cycle time (s)The coefcient of performance COP is as the ratio of cooling

production by the effective heat input:

COP QcoolQin

(16)

The advantage of this simplied modelling approach is thatresults can be obtained quickly for a large number of cases. Thestability and accuracy of the solutions was veried by varyingthe simulation time step by an order of magnitude and by manualchecking of the temperature history for smoothness and signs ofinstability. The simulations were run for several temperature cyclesuntil steady cyclic behaviour was obtained (usually after fourcycles). The validity of combining the uid andwall thermal massesand of the assumptions made was investigated using a moredetailed model outlined by Critoph [1,2] and the model is fullyvalidated against experimental data by Metcalf [3,4]. Furthermorethis model has already been used as the design tool of a heat drivenair conditioning system that is operating with satisfaction [9].

The sorption generator simulation parameters are given inTable 3. The adsorbent layer thickness, tc, of 4 mm is half of the totallayer thickness, since the other layer half will be heated and cooled

via heat transfer uid in the opposing plate. This is twice that ofthe layer thickness in the prototype shown in Fig. 1. The layerthickness was increased in order to reduce the thermal mass of thegenerator (which was optimised for a mobile air conditioningapplication) and thereby increase the COP. This reduces therequired collector area. However, SCP and therefore the sorptiongenerator costs are increased, and there is an optimum layerthickness dependent upon the relative cost of the collector and thesorption generator.

3.2. Sorption system simulations

The sorption system model was used to calculate performancein terms of COP and SCP over a range of ambient temperaturesand driving temperatures. The evaporating temperature was5 C,selected as appropriate for ice production. The condensingtemperature and the inlet temperature of the heat transfer uidduring nal cooling of the sorption generators were assumed tobe 5 C above ambient. Figs. 4 and 5 show the sorption systemperformance calculated from the model for driving temperaturesbetween 65 and 245 C and ambient temperatures from 10 to 50 Cfor two and four-bed cycles, respectively.

3.3. Solar collector calculations

Each solar collector is direct ow type and the efciency wascalculated using the coefcients (linear loss coefcient, k1 and

0.5

0.6

40

45

50

10oC0.8

40

45

50

50oC

10oC

k

30T

amb ( C)

S.J. Metcalf et al. / Applied Thermal Engineering 31 (2011) 2197e2204 2201Tcol (

oC)

POC

100 150 2000

0.1

0.2

0.3

0.4

10

15

20

25

30

35T

amb (oC)

50oC

gkW(

PCS1

-

)

Tcol (

oC)100 150 200

0

50

100

150

200

250

300

350

400

10

15

20

25

30

35

40

45

50

Tamb (

oC)

50oC

10oCFig. 4. Two-bed cycle sorption system performance.Tcol (

oC)

W(PCS

100 150 2000

50

100

10

15

20

2550oCTcol (

oC)100 150 200

0

0.2

10

15

20

g1

-

) 150

200

35

40

45

50

10oC

oPOC 0.4

0.6

25

30

35T

amb (oC)Fig. 5. Four-bed cycle sorption system performance.

where M is the mass of ice produced (kg), Q is the cooling (J),

22

23

24

25

26

27

28

29

30

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMonth

(e

ru

ta

re

pm

eT

o

)C

Fig. 6. Monthly mean ambient temperature for Dakar, Senegal.

S.J. Metcalf et al. / Applied Thermal Engineering 31 (2011) 2197e22042202quadratic loss coefcient, k2) and optical efciency (h0) given inTable 1 in the following equation:

h h0 k1Tcol Tamb

G k2

Tcol Tamb2G

(17)

The heat output from the solar collector is given by

_Qcol hTcol; TambGA (18)At steady state, this is balanced by the heat input to the sorption

refrigerator which is given by

_Qin SCPTcol; TambCOPTcol; Tamb

Mc (19)

For a given ambient temperature Tamb and solar insolation G, thecollector temperature which gives the correct balance _Qcol _Qinmust be found (heat balance within 2%). This is carried out inMATLAB using the fminsearch function minimisation routineand linear interpolation of the COP and SCP data from the sorptionsystem model.0

5

10

15

20

25

30

Jan Feb Mar Apr May Ju

M

m/

JM

(n

oi

ta

id

aR

ra

lo

Sn

ae

M2

)

Fig. 7. Monthly mean solar radice cool

Lwater is the latent heat of fusion of water (taken as 335.5 kJ kg1)and cpwater is the specic heat capacity of water (taken as4200 J kg1 K1).

3.5. Weather data

Weather data for Dakar (Senegal) was obtained from theMeteonorm 4.0 package [10]. The ambient temperature was takenas the air temperature and the solar insolation as the hemisphereradiation on a tilted plane. The solar collector was set south facingwith an inclination equal to the latitude of 14. Fig. 6 and Fig. 7 plot3.4. Ice production calculation

The quantity of ice produced was calculated using the followingexpression:

Mice Qcool

Lwater cpwaterTamb 273(20)n Jul Aug Sep Oct Nov Dec

onth

iation for Dakar, Senegal.

In Fig. 8, the four combinations of collector and cycle type are

2

4

6

8

10

12

m

re

p

ec

i

gk

yl

ia

d

na

em

la

un

nA

2

ro

tc

el

lo

c

Flat Plate Two-BedEvacuated Tube Two-BedFlat Plate Four-BedEvacuated Tube Four-Bed

0.0105

0.0110

0.0115

0.0120

0.0125

0.0130

0.0135

0.0140

0.0145

0.0150

ts

oc

eni

hc

am

la

to

tf

ot

in

ur

ep

ec

if

os

sa

my

li

aD

)/

gk

(

Flat Plate Two-BedFlat Plate Four-BedEvac Tube Two-BedEvac Tube Four-Bed

S.J. Metcalf et al. / Applied Thermal Engineering 31 (2011) 2197e2204 2203compared on the basis of the annual mean daily mass of iceproduced per square metre of collector area. The ratio of thecollector area to the total mass of carbon adsorbent in thethe monthly mean ambient temperature and solar radiationrespectively during operation of the system.

4. Results and discussion

00 0.5 1 1.5 2 2.5

Collector Area per kg Adsorbent (m2

kg-1

)

Fig. 8. Annual mean daily ice production for the four combinations of collector andcycle type.adsorption refrigerator is varied for each case and an optimumfound which maximises the ice production per square metre ofcollector. This optimal value is mainly driven by ammonia uptake.In fact increasing the collector area leads the increase of the heatinput. However there will be a point when the bed will almostcompletely release its ammonia content and any additionalheat input (therefore additional collector area) will be counter-

0.00

2.00

4.00

6.00

8.00

10.00

12.00

14.00

Jan Feb Mar Apr May Jun

mr

ep

ec

if

og

ky

li

aD

na

eM

2

ro

tc

el

lo

c

Fig. 9. Mean daily ice production for each month witproductive [11]. The optimum ratio is 0.625 and 1.5 m2 kg1 for thetwo-bed and four-bed cycles, respectively.

Fig. 9 plots the mean daily ice production for each month ofthe year with a four-bed cycle and the optimal area of evacuatedtube collectors. It can be seen from the gure that the maximum iceproduction is in April at 13.7 kg m2 and the minimum in July at8.6 kg m2. Although the solar insolation is higher in July, theambient temperature is also higher which reduces the performanceof the adsorption refrigerator. This would have to be consideredwhen sizing the machine along with the fact that the storage boxitself will have higher losses at higher ambient temperatures andtherefore require more cooling. This ice production is higher than

2

0.01000 0.5 1 1.5 2 2.5

Collector area per kg adsorbent (m2

kg-1

)

Fig. 10. Daily ice production per unit total system cost.the 4e7 kg m reported by Wang and Oliveira [12] for state-of-the-art solar powered adsorption ice makers.

The optimum collector areas in Fig. 8 will only be optimal interms of capital cost if the cost of the adsorption machine is negli-gible in comparison to that of the collector array. The cost of theadsorption machine is largely that of the sorption generators. Their

Jul Aug Sep Oct Nov DecMonth

h a four-bed cycle and evacuated tube collectors.

cost has been calculated at V70 per kg of adsorbent, which is basedon the cost of the plate heat exchanger in Table 2 assuming the samecost per plate. The price given in Table 2 is the retail price for a one-off unit and could be reduced if the systems were produced in largenumbers. The sorption generator cost is then added to the solarcollector cost from Table 1 to obtain an estimate of the total cost.

Fig.10 shows the annualmean dailymass of ice produced per unitcost of the system. It can be seen from the gure that the four-bedsystemmarginally outperforms the two-bed, producing 0.0147 kg ofice per day per V of total cost compared to 0.0129 kg/V with evacu-ated tube collectors. However, in practise the greater complexity ofthe four-bed system may make the two-bed system favourable.Evacuated tube collectors prove more economic than at platecollectors, their higher performance outweighing their higher capitalcost. It can also be seen from Fig. 10 that the cost of the adsorptionmachine is only high enough to alter the optimum collector area inthe case of at plate collectors with a four-bed system.

The four-bed systemwith evacuated tube collectors would havea capital cost of V68,000 for a machine which could produce anaverage of 1000 kg of ice per day, consisting of V10,200 for theadsorption machine and V57,800 for 91 m2 of evacuated tubecollectors. Although still relatively high, the running costs would beextremely low and a lifetime of 20 years could be expected, whichequates to an annual cost of V3400.

Work is currently being performed to increase the thermalconductivity of the carbon adsorbent, which could further increasethe SCP and reduce the cost of the adsorption machine. As solarcollector technology develops and production quantities increase,their price will also reduce signicantly.

countries has been investigated. Computational modelling haspredicted that for a four-bed adsorption system with evacuatedtube collectors, an annual average daily ice production rate of 11 kgper square metre of collector could be achieved. The capital cost ofa systemwhich could produce up to 1000 kg of ice per day in Dakar(Senegal) has been estimated at V68,000.

References

[1] R.E. Critoph, S.J. Metcalf, Specic cooling power intensication limits inammonia-carbon adsorption refrigeration systems, Applied Thermal Engi-neering 24 (5e6) (2004) 661e678.

[2] R.E. Critoph, Heat Exchanger. UK Patent Application No 0617721.6 (2006).[3] S. J. Metcalf, Compact high efciency carbon-ammonia adsorption heat pump,

University of Warwick (UK), PhD Thesis, 2009.[4] S.J. Metcalf, Z. Tamainot-Telto, R.E. Critoph. Development of domestic

adsorption heat pump, IEA 10th Heat Pump Conference, Tokyo (Japan), PaperNo 00201 (2011).

[5] A. Diallo, Status of sh stocks in Senegal, In: E.K. Abban, C.M.V. Casal, T.M. Falkand R.S.V. Pullin (eds.), Biodiversity and Sustainable use of sh in coastal zone,pp. 38e40, 2000.

[6] Sustainable management of sh resources project: Senegal, The World Bank,Report No 45462-SN, 2008.

[7] W.M. Kays, A.L. London, in: Compact Heat Exchangers, McGraw-Hill,1984.

[8] R.E. Critoph, Evaluation of alternative refrigerant-adsorbent pairs forrefrigeration cycles, Applied Thermal Engineering 16 (11) (1996)891e900.

[9] Z. Tamainot-Telto, S.J. Metcalf, R.E. Critoph, Novel compact sorption genera-tors car air conditioning, International Journal of Refrigeration 3 (4) (2009)727e733.

[10] METEONORM, Version 4.0, Global Meteorological Database Software,1999.

[11] Z. Tamainot-Telto, G.S.F. Shire, Compact sorption generator for solar and

S.J. Metcalf et al. / Applied Thermal Engineering 31 (2011) 2197e220422045. Conclusions

The feasibility of applying a novel plate heat exchanger solidsorption generator to solar powered refrigeration for developingwaste-heat applications: Ice making and Air Conditioning in tropical or hotclimate regions, In: K. Choon Ng and B.B. Saha (eds), Chap. 10, Advances inAdsorption Technology, ISBN: 978-1-60876-833-2, 2009.

[12] R.Z. Wang, R.G. Oliveira, Adsorption refrigeration e An efcient way to makegood use of waste heat and solar energy, Progress in Energy and CombustionScience 32 (2006) 424e458.

Application of a compact sorption generator to solar refrigeration: Case study of Dakar (Senegal)1 Introduction2 Solar refrigeration system3 System performance simulation3.1 Sorption system model3.2 Sorption system simulations3.3 Solar collector calculations3.4 Ice production calculation3.5 Weather data

4 Results and discussion5 Conclusions References