PERGAMON

Renewable Energy 16 (1999) 673678

RAPID CYCLING

SOLAR/BIOMASS

POWERED SYSTEM Dr. R.E. Critoph

ADSORPTION

REFRIGERATION

Engineering Department, University of Warwick, Coventry CV4 7AL, UK

ABSTRACT There is a need for heat operated refrigeration systems in areas of the world which have a warm climate and also have no firm electricity supply. Research is underway on carbon - ammonia refrigerators driven by the heat of steam condensing in a thermosyphon heat pipe. The heat source can be solar energy, biomass, or some combination of the two. New monolithic carbon adsorbent-aluminium composites have been developed which promise power densities of 1kW cooling per kg carbon. The high power density leads to much lower fust costs. 0 1998 Elsevier Science Ltd. All rights reserved.

KEYWORDS

Refrigeration, solar, biomass, carbon, ammonia, adsorption, heat transfer INTRODUCTION There is a demand for cooling in many parts of the world where there is no fum electricity supply and conventional fuels are difftcult or expensive to obtain. Requirements tend to be either for medical uses where a high capital cost per kW of cooling is acceptable, or for food (especially fish) preservation where the cooling power required is much greater and the acceptable cost per kW may be lower. Vaccine storage refrigerators have been sold at a cost of f60-170 /W cooling (Critoph, 1990). but in the case of a fish storage ice-maker for Zambia, the target capital cost was f8.5 /W (Harvey 1990). The nearest that an experimental system has come to achieving the target cost was a 100 kg/day ice-maker built in Thailand ( Exe11 and Kornsakoo, 1987). This was a 25 m2 ammonia - water refrigerator working on a diurnal cycle. The most expensive part of all solar refrigeration systems is the collector array. If the system is to be economic then ways must be. found to minimii the cost. The total area of collector can be reduced by utilising a back-up energy source during periods of low insolation. The source can be a renewable one such as the combustion of agricultural waste i.e. bagasse, rice husks, etc. In some locations there may be enough biomass to dispense with the solar input completely and use a biomass source. If collectors are used in what are bound to be large arrays, there is a case for using an intermediate heating fluid in order to reduce the cost. Many thermally driven refrigeration cycles use ammonia as the working fluid and the cost of constructing solar collectors which heat the fluid at pressures of up to 25 bar could be prohibitive. Other types of machine use water or methanol as refrigerants but in these cases the collectors suffer from the problems of having to be hermeticahy sealed against air ingress. Any inward leakage of air can stop the system working. The solution preferred here is to use collectors which will boil water in a thermosyphon heat pipe arrangement which does not require a pump and yet provides excellent heat transfer between the collector and refrigeration unit. Such a heating arrangement is also ideally suited to the use of burning biomass as a heat source, whether as a back-up or as 0960-1481/99/$-see front matter.0 PII: SO960-l48l(98)00250-X 1998 Elsevier Science Ltd. All rights reserved

674
sole source. Our research is concentrating thermosyphon heat pipe. THE ADSORPTION CYCLE

WREC 1998
on the use of adsorption cooling systems which receive heat from a

Adsorption refrigeration cycles rely on the adsorption of a refrigerant gas into an adsorbent at low pressure and subsequent desorption by heating. The adsorbent acts as a chemical compressor driven by heat. In its simplest form an adsorption refrigerator consists of two linked vessels, one of which contains adsorbent and both of which contain refrigerant as shown in Fig. I below.

w
Figure? Initially the whole assembly is at low pressure and temperature, the adsorbent contains a large concentration of refrigerant within it and the other vessel contains refrigerant gas (a). The adsorbent vessel (generator) is then heated, driving out the refrigerant and raising the system pressure. The desorbed refrigerant condenses as a liquid in the second vessel, rejecting heat (b). Finally the generator is cooled back to ambient temperature, readsorbing the refrigerant and reducing the pressure. Because the liquid in the second vessel is depressurised and boils, it takes in heat and produces the required refrigeration effect. The cycle is discontinuous since useful cooling only occurs for one half of the cycle. Two such systems can be operated out of phase to provide continuous cooling. Such an arrangement has a comparatively low Coefftcient of Performance (COP = Cooling I Heat Input). Also, the thermal conductivity of the bed is generally poor so the time taken for a cycle could be an hour or more and the cooling power per mass of adsorbent could be as low as IO W/kg. This is not a problem with solar powered vaccine refrigerators which produce a few kg of ice each day and operate on a diurnal cycle (Critoph, 1994). However, a refrigerator producing one tonne of ice in a diurnal cycle would need 5 tonnes of carbon and contain I.5 tonnes of ammonia. When contemplating larger icemakers it is obviously necessary to use a much faster acting cycle in order to reduce the mass of adsorbent and the cost of the system. Two beds, similar to the one shown above, can be heated and cooled out of phase to provide continuous cooling Good heat transfer is required to reduce the cycle time to a few minutes and thereby increase the power density of the adsorbent to the order of I kW/kg. We can also achieve a higher COP by maximising the quantity of heat regenerated. The heat rejected by one bed when adsorbing can provide a large part of the heat required for desorbing in other bed. This also requires good heat transfer. Early attempts to improve adsorbent bed heat transfer by using tinned heat exchangers within them (Zanife, 1988) were only of limited success, Attention turned to the use of composite adsorbents with either graphite flakes or metal foams to improve conductivity (Guilleminot et al 1993 , Mauran et al 1992). These do allow higher power densities but the problem of obtaining a high regeneration effectiveness between beds is still comparatively complex. The U.S. company Wave Au patented a Thermal Wave cycle (Miles et al 1990, 1992) in which a thermal fluid is pumped through a special heat exchanger in each bed. This sets up heating and cooling waves in each bed in a similar way to those produced in normal heat recovery regenerators and gives good COPS. MONOLITHIC CARBONS

The UK company, Sutcliffe Speakman Carbons, has developed processes for making solid blocks of almost any type of carbon in a wide variety of shapes. The details are confidential but may be summarised by saying that the powdered carbon is either compressed with an organic binder to make briquettes or extruded through a die. The green material is then fued in a furnace. The initial application that prompted this development was the storage of methane as a vehicle fuel, but it is equally useful in refrigeration applications. Monoliths up to 100 mm in diameter have been made. The ammonia adsorption properties of the monoliths have been measured by Critoph (1996a) and it has been shown that they offer greater storage density for ammonia than

WREC 1998

675

conventional

granular beds of carbon. The thermal conductivity

of the material is three times that of a granular path

bed. Whilst this is a large improvement, lengths are still needed.

if power densities of I kW/kg are to be achieved small conduction

Sutcliffe Speakman Carbons donated monolithic discs of 208C carbon, 100 mm in diameter and I5 mm thick, with a 25 mm central hole. An experimental generator was designed around the disc dimensions. The discs were machined down slightly coefficient thermosyphon to improve the surface fmish. This is necessary to ensure a good heat transfer of heat transfer, it in the phase and to cool between the disc and metal heat transfer surface. In order to maximise the intensity heat pipes were used both to heat the carbon in the desorption

adsorption phase. Fins were used to enhance heat transfer within the bed. Critoph performance of the design, which showed good agreement with computer simulation, optimum disc size. Computer geometry simulation of the dynamic heat and mass transfer in the carbon

(1996b) reported on the but was not based on the a

has been used to determine

that is closer to optimal

(Critoph

et al, 1997). An appropriate

figure of merit is the cooling

power

per unit mass of carbon (PC), since this will be influence strongly form of generator cooled externally in the effective used is as in Fig.

the capital cost of the whole machine. The is heated and is a laminate of carbon discs and aluminium without a major increase

1, which

shows the dimensions fmally used. The generator water. The interior

by condensing or evaporating

fins formed in situ. The aluminium thermal mass.

increases greatly the effective thermal conductivity

50mm

nalithic

carbon

Fig.2

Experimental

generator geometry aluminium sheet thickness was found to be of the order of 10% that of the carbon disc diameters are presented in Table I with an initial

The most desirable but practical carbon disc thickness.

The results of different

generator temperature of 50 The evaporating temperature T, is about 0 The specific power is calculated C. C. by assuming that the cooling process and the heating process have the same period (the heating period is taken when the maximum rate of concentration change occurs). The coefficient of performance (COP) figure does not include the effect of the thermal mass.

Table 1: Generator
evaporating temperature 9 (mm) 50 75 100

performance

with

condensing

temperature

50

C,

temperature 0C. and carbon thickness 2 mm with TE optimised for maximum cooling power. T, ( C) 143 142 139 (kg NH&g Ax Carbon/s) 0.310 0.304 0.292 COP

generating

P, (kW&g) 1.276 0.869 0.622

0.232 x I@ 0.158 x lU2 0.113x 10 2

676

WREC 1998

The mean carbon bed temperature and the mean ammonia concentration change for a generator with a disc diameter of 50 mm are shown in Fig. 3 and Fig. 4 respectively. The heat supply temperature is 2OO in all C cases. This is commensurate with burning biomass as a heat source, but not with conventional solar collectors. The use of evacuated tube collectors might enable a supply temperature of 150C to be used. Mean carbon bed temPe&re VC)

250

Fig. 3: Plot of mean carbon bed temperature against time From Fig. 3 it is apparent that the mean carbon bed temperature increases over time from the initial starting temperature of 5O up to the quasi-steady state value of 200C. The quasi-steady state temperature, C equivalent to the condensing temperature of the steam applied, is attained aRer approximately 150 seconds of heating. However, the generating temperature at which the specific cooling power is maximised is lower than the quasi-steady state value, as is shown in Fig. 4. The mean concentration shown in Fig. 4 can be seen to decrease during the heating phase as progressively more ammonia is des0rbe.d from the carbon bed. The maximum change in ammonia concentration predicted is approximately 10%. This value is attained after approximately I50 seconds corresponding to the maximum mean carbon bed temperature of 200C at the quasi-steady state condition. The heating time for maximised mean cooling power is 22 s, corresponding to a mean carbon temperature of 143C. EXPERIMENTAL SYSTEM A laboratory demonstration system capable of up to 3kW cooling is under construction. The two generators are as shown in Fig. 2. each being I m long. Heating and cooling of both is by a common thermosyphon heat pipe with water as the working fluid. The laboratory system has a large number of individual valves but a production version would have a single spool valve with two possible states. Steam from the boiler is diverted to one generator whilst the other is cooled by low pressure boiling water that rejects heat in the condenser. A schematic of the whole system is shown in Fig. 5. The layout was chosen for maximum flexibility in the laboratory. It would be possible to have up to four separate heat pipes to reduce the number of valves needed but this would also reduce the heat transfer area available to each. An appropriate design depends on the facilities available for manufacture, for maintenance and on the balance between efftciency and allowable capital cost.

WREC

1998

0.18 ,

Mom concentdon

0.14 .

+ 0
50 100 150 200 250 300 Time(s) Fig. 4: Plot
of mean ammonia concentration change against time

__

Ammonia

Water/steam

Ammonia check valve Water/steam shut-off valve

Fig. 5: Schematic diagram of complete test rig At the time of writing, the generator manufacturing technique has been established and all of the major components purchased. Commissioning should take place in May 1998.

678

WREC

1998

ACKNOWLEDGEMENTS This research was supported by the UK Engineering Speakman Carbons Ltd. and Alvis Vehicles Ltd. REFERENCES Critoph, R.E. (1990). Refrigeration in developing countries - the renewable options, Proc.lst World Renewable Energy Congress, Reading, UK. Critoph, R.E. (I 994). An ammonia carbon solar refrigerator for vaccine cooling, proc3rd World Renewable Energy Congress, Reading, UK. Critoph, R.E. (1996a).Evaluation of alternative refrigerant - adsorbent pairs for refrigeration cycles. Applied Thermal Engineering 16, II, pp 891-900. Critoph, R.E. (1996b) Towards a one tonne per day solar ice maker Proc 4th World Renewable Energy , Congress, Denver, pp. 626-63 I, Pergamon. Critoph, R.E., Tamainot-Telto Z., Davies, G. (1997). Design of an adsorption generator utilising a novel carbon-aluminium laminate. Proc. Heat Powered Cycles Nottingham, UK. 97, Exell, R.H.B., Komsakoo, S., Geapipatanakul, S(1987). A village size solar refrigerator. Asian Institute of Technology Report No. 173, Bangkok, Thailand. Guilleminot, J.J., Chalfen, J.B., Choisier, A. (1993). Heat and mass transfer characteristics of composites for adsorption heat pumps., AES - Vol 3 1, Int. Absorption Heat Pump Conf., ASME. Harvey, A.B. (1990). Study of an intermittent regenerative cycle for solar cooling. Ph.D. thesis, University of Warwick, UK, 1990. Mauran, S.. Prades, P., L Haridon, F., (1992). Transfer& de chaleur et de masse en milieux reactifs consolides pour systemes themochimiques. Proc. Solid Sorption Refrigeration, Paris, Nov. 1992, pub. Limsi, ISBN 2 903 633 592. Miles, D.J., Shelton, S.V., Wepfer, W.J. (1990). Ramp wave analysis of the solid vapour heat pump. J. of Energy Res. Tech., 112, pp 69-79. Miles, D.J., Sanbom,D.M., Nowakowski, G.A., Shelton, S.V. (1992). Gas tired sorption heat pump development. Proc. Solid Sorption Refrigeration, Paris, Nov. 1992, pub. Limsi, ISBN 2 903 633 592. Zanife, T.N. (1988).Etude de la regulation d une pomp-e a chaleur a adsorption a deux adsorbeurs : cas zeolithe eau. Proc. Pompes a chaleur de hautes performances, Perpignan, pub. Lavoisier. and Physical Sciences Research Council, Sutcliffe