Carbon Vol. 27. No. I, pp.63-70. 1989 Printedm Great Britain.
oW3-6223/8Y S3,lMl+.(XI 1989 Pcrgamon PTCSS plc
Copyright
ACTIVATED CARBON REFRIGERATION
ADSORPTION CYCLES AND HEAT PUMPING
R. E. Engineering Department,
D
FOR
CRITOPH
University of Warwick, Coventry, CV4 7AL. U.K
Abstract-Adsorption cycles were used for cooling before the advent of reliable mechanical compression refrigerators, and are now attracting renewed interest as the basis of solar powered refrigerators for vaccine storage etc. They also have potential for use in heat pump systems. The cycle thermodynamics are reviewed and the suitability of a range of refrigerants with active carbon adsorbents is assessed. Methanol, sulphur dioxide, ammonia, methyl amine, and formaldehyde are of particular interest. Experimental p-T-x data is presented for ammonia and methanol with a range of carbons, together with their calculated performance in refrigeration cycles. Currently available carbons can give good performance, but it is possible that specially developed carbons might show significant improvement. A major development area lies not in changing the porosity characteristics but in trying to improve heat and mass transfer to and from the adsorbent. Key Words-Adsorption,
refrigeration,
solar energy, activated carbon, heat pumps
1. INTRODUCTION
ture solar refrigerators at present. Two companies use adsorption cycles and the remainder are photovoltaic. The relative merits are discussed in[4]. Adsorption cycles have also been considered for use in heat-actuated heat pumps, as have many different absorption cycles. However, this market area is further away from the point of view of initial prototype testing, let alone commercialization. The simple adsorption cycle is best understood with reference to the p-T-x (pressure-temperatureconcentration) diagram of Fig. 1 and the schematic diagram of Fig. 2. The processes involved are as follows:
Adsorption cycles for refrigeration were first used in the early 1900s as reported by Plank and Kuprianoffll]. Machines using both the adsorbent-refrigerant pairs CaCl,-NH, and activated carbon-methanol were successfully manufactured. However, the development of both the hermetic compressor and the Platen-Munters absorption cycle led to the demise of these early systems. Interest in adsorption cycles has been rekindled by the need for solar-powered refrigeration. The World Health Organization has specified a requirement for solar refrigeration in its EPI (Expanded Programme on Immunisation), where it feels that solar refrigerators in Third World rural clinics have an important role to play in maintaining the vaccine cold chain[2]. On a larger scale, solar ice making for preserving fish catches looks economic in countries such as Zambia[3]. About 20 companies manufac-
1. Starting in the morning with the valve open and at ambient temperature of about 3O”C, the rich concentration adsorbent in the generator/absorber is heated by solar energy until the pressure reaches a level that enables refrigerant to desorb and be condensed in an air- or water-cooled condenser.
VALVE
I
Fig. 1. Schematic solar refrigerator. 63
/
/7/771 COLD BOX
64
R. E. CRITOPH
I -10°C
;30°c
100"c ;
) -+
(K-l)
Fig. 2. Simple adsorption refrigeration cycle. 2. Refrigerant is driven off at constant pressure, the adsorbent becoming more and more dilute until the m~imum cycle temperature of about 100°C is reached. The condensed liquid is collected in a receiver . 3. The valve is shut, and the adsorbent cools and reduces its pressure. At some stage of the evening or night, its pressure will be the saturated vapor pressure of refrigerant at - lO”C, which is sufficiently low for ice production. 4. The valve is now opened, and the liquid refrigerant starts to boil in the evaporator. Initially, the refrigerant within the evaporator and receiver simply cools itself, but having dropped below 0°C it can start to freeze water. Adsorption is completed by the following morning, completing the cycle. During this process heat is released in the adsorber and so the generatoriadsorber must be cooled by ambient temperature air or water. This ideal cycle can be used to calculate the coefficient of performance (COPS) with reasonable accuracy, but in practice an isolating valve is not necessary and processes c and d merge into a smooth curve and there are no sharp corners on the experimental p-T-x diagram. The main criteria of the suitability of a particular cycle is the coefficient of performance. For ref~geration, this is defined as COPC = Cooling obtained/high-grade
heat input
and for heat pumping as COPH = Heat output/high-grade
heat input
In solar refrigeration, there is a distinction to be made between the cycle COP and the solar COP. The former uses the heat actually entering the adsorbent as heat input, whereas the latter uses the soiar energy incident on the collector.
2. CYCLE THERMODYNAMICS It is possible to determine many characteristics of the cycle before fixing the adsorbent and refrigerant to be used. Consider the p-T-x diagram of Fig. 3. The temperatures shown are
Ta, = Temperature at start of adsorption Tu2 = Temperature at end of adsorption Te = Evaporating temperature Tc = Condensing temperature Tg, = Temperature at start of generation Tg, = Temperature at end of generation Te is determined by the cold box temperature and the heat transfer from it to the evaporator. Tc and Taz are temperatures of heat rejection to ambient, and so will be a little higher than ambient. For simplicity they will be assumed equal. Tg, is determined by the available solar radiation and the solar collector characteristics. It has been pointed out by Haseler et a1.[5] that the isosteres obey Trouton’s rule, and that the four temperatures specified by the intersection of two isobars and two isosteres are simply related. Specifically: Te Tg, = Tc Ta, and Tg, Ta, = Tg, Ta,
Critoph[6] goes on to derive an approximate value of the COP that neglects the sensible heating load of the adsorbent: COP = Evaporating
temperature/mean temperature
adsorption
= Condensing
temperature/mean temperature
generating
Activated carbon adsorption cycles A In P
I I I
I I I
Te
TW,TC
I I I
I I I
I I
Tgl Tar
T9z
w
-+(K-‘1
Fig. 3. Generalized adsorption cycle.
A base case of Te = - 10°C Tc = 3o”C, Tg (mean) = 90°C gives a COP of .83 compared to the Carnot cycle limitation of 1.09. The limit is further reduced when the sensible heat of the adsorbent and refrigerant are taken into account. However, further information is required-the latent heat (L) of the refrigerant and the rich and weak concentrations. The dependence of COP to these is shown in Figs. 4 and 5. taken from ref.[6]. They illustrate that provided L is greater than about 1000 kJ/kg and the change in concentration is greater than about lo%, then the possible COP gains due to further increase are not very large. These results are calculated using a suitable specific heat for a carbon adsorbent, but other adsorbents have been used, including zeolites and silica gel. However, Critoph and Vogel[4] evaluated several refrigerants with zeolites and active carbon and concluded that for solar cooling active carbon gives
a better COP. Zeolites give a greater temperature lift, that is, (Tu, - Te) is larger and so high ambient temperatures and low cooling temperatures can be tolerated, but (Tg2 - Tc) will be correspondingly high, requiring very high collector temperatures and low collector efficiencies. High desorption temperatures need not be a problem in applications such as gas-fired heat pumps for space heating, and the zeolite water pair has been suggested for this application[7]. Active carbon based pairs could be used for such applications, but the heat output temperature would have to be kept below about 50°C for reasonable efficiency. 3. REFRIGERANT REQUIREMENTS
To obtain a good COP, the heat of desorption should be very large compared to the sensible heating load of the adsorbent.
COP A COP X MEAN
0.8
ax.0
= 'I5
1
0.8 -
L =I200 kJ/kg
0.6
0.1
80
100
120
Fig. 4. COP v. Generating temperature
140
Tg/‘C)
and latent heat.
Fig. 5. COP v. Concentration
range and mean.
66
R. E. CRITOPH
Considering a particular carbon with many possible refrigerants, those refrigerants with a high latent heat per liquid volume should be best assuming similar pore volume filling in all cases. Apart from the latent heat criterion, the refrigerant needs to be thermally stable in the presence of the carbon, and preferably nontoxic and nonflammable. Another major consideration that affects refrigerant choice is whether the cycle is ever subatmospheric. If the working pressure is subatmospheric and even the smallest leak of air into the system occurs, then the refrigerator will cease functioning due to the inability of the refrigerant to diffuse through the air at a reasonable rate. However, with a high-pressure system a small outward leakage of refrigerant could be tolerated for some time and is easier to rectify. Table 1 shows the relevant properties of possible refrigerants. Water is included for completeness, but the practical difficulties preclude using it at ice-making temperatures. The table ranks the refrigerant in order of pL (liquid density x latent heat) and splits them into groups that are above and below atmospheric pressure when the saturation temperature is - 10°C (for ice making).
4. CYCLE PERFORMANCE
[-D
NO HEAT RECOVERY 0.6
0.2
0
($ln+)‘]
80
100
120
I40 Tg2("C)
Fig. 6. COP for various refrigerants.
where X= concentration P= liquid density (kg rne3) V” = pore volume (m3/kg) T= temperature (K) P= pressure (bar) Psa, = saturation temperature P= affinity coefficient of refrigerant (taking the affinity coefficient of benzene as un-
CALCULATIONS
Some experimental or theoretical p-T-x relationships are required to calculate the cycle COP. The Dubinin-Astakhov (D-A) or Dubinin-Radushkevich (D-R) equations are generally adequate in the lmited range required by refrigeration. Saturation pressure/pressure ratios (pSa,lp) vary from about 4 to 10 for different cycles and different refrigerants, and either equation gives a good fit providing capillary condensation can be avoided. The D-A equation is used in the form
x = pV,exp
0.8
ity) D,n = constants
characterizing
the carbon
The only other datum required on the carbon is its specific heat, which for general theoretical calculations was taken as 0.7 kJ/kg. To calculate the cycle COPS for a particular refrigerant in the absence of measured p-T-x data, p may be taken from the literature or calculated by various methods as outline in Smisek and Cerny[8]. The results of applying these methods to different refrigerants is shown in Fig. 6 taken from[6]. It can be seen that methanol is clearly the best refrigerant on COP grounds, followed by ammonia.
Table 1. Refrigerant properties Refrigerant Formula name NH, HCHO C2H3F SO, Cl,-
H,O &OH CzH50H CzHzN NO, CzHjN CH,NH:
Ammonia Formaldehyde Vinyl fluoride Sulphur dioxide Chlorine Water Methanol Ethanol Ethyl amine Nitrogen dioxide Acetonitrile Methyl amine
N.B.P., “C
M.W.
-34 - 19 -38 - 10 -34 100 65 79 51 21 81 -7
17 30 64 46 71 18 32 46 43 46 41 31
L(kJ/kg) 1368 768 389 605 288 2258 1102 842 746 415 766 836
p(kg/mY
p/L x 10-j
681 815 1455 883 1563 958 791 789 833 1447 782 703
932 626 566 534 450 2163 872 665 621 600 599 588
Activated carbon adsorption cycles Methanol is already well investigated as a refrigerant[9,10]. Possible difficulties with it are that it may not be stable above about 12O”C, the poor heat transfer in the packed bed of charcoal, the care with which pressure drops due to high flow velocities must be avoided, and the possible ingness of air over a long life. However, one solar refrigeration based on this pair is now on the market[ll]. In contrast, ammonia has received comparatively little attention. To validate the theoretical prediction that ammonia should prove a good refrigerant with activated carbon, experimental measurements were carried out with a range of charcoals supplied by Sutcliffe Speakman. 5. TESTS ON AMMONIA AND ACTIVATED CARBONS The experimental technique used was to measure the pressure and temperature of weighed quantities of the pair for a range of temperatures. A large (100 g) sample of activated carbon was used in a steel pressure vessel immersed in a temperature-controlled bath. The temperature at the center of the sample was measured and pressure and temperature recorded when the centre temperature came within 2 K of the outer temperature. Use of such a large sample gave a qualitative indication of the dynamics of a full-scale refrigerator. The results for both heating and cooling were fitted to the D-R equation, a typical set being shown in Fig. 7 for carbon 207C. In all, five carbons were tested and values of V,, and D calculated, assuming p to be 0.28. These results are shown in Table 2, together with results for methanol (p = 0.4) calculated from isotherms measured by Sutcliffe Speakman[ 171, and Sridhar[ 181. Having obtained these values, the COPS for a range of temperatures could be calculated, and a typical
67
Table 2. Activated carbon properties with methanol and ammonia V,,U/kg)
D x 10’
Ammonia 205c 207C 208C
0.286 0.291 0.381
0.781 0.797
610
0.408
1.054
Methanol 203c 207c 208C
0.250 0.246 0.530
0.832 0.608 1.74
1.110
set of results is shown in Fig. 8. Reasonable COPS can be obtained, but the sensitivity to the adsorption/condensing temperature should be noted. Different carbons at some set of temperature conditions also give a large range of COPS, as has been shown for methanol by Passos et a1.[12] and Critoph[6]. They have plotted curves of COP u. 7& for set adsorbing and condensing temperatures using various carbons. In order to generalize further, to a whole range of carbons, it is better to fix all the temperatures and to see how COP varies with the charcoal parameters V,, and D in the D-R equation. This has been done for both methanol and ammonia in Figs. 9 and 10. The methanol data is from refs.[12,17,18]. A high value of V,, and D gives a good COP, as would be expected. However, the gains to be made in trying to obtain very high values are comparatively limited. Combinations of V,, and D for different carbons are shown on the diagrams to illustrate the range currently available. The limit on COP if thermal masses are neglected is about .8, and to approach this more closely much higher concentration
4 -h(x)
NH, - 207C
Fig. 7. Typical D-R results for ammonia and activated carbon.
68
R. E. CKITOPH COP
COP NH3 - 610
h% - 610
Ta = Tc = 3O’C
0.8
Te = -5’C
0.8
-lO’C< Te
2S°C< Ta,TcsSO'C
0.6
0.6
i
> 80
100
120
140
t
Ts ('Cl
80
100
120
140
Tq (OC)
Fig. 8. COP as a function of cycle temperatures for ammonia and carbon 610.
changes must occur in the range 30 to 120°C.Itseems unlikely that very much greater gains can be made in this direction. 6.
PERFORMANCE OTHER
THAN
IMPROVEMENT CHANGES
BY MEANS
IN POROSITY
Rather than trying to achieve a performance gain by further improving the porosity characteristics, the cycle itself may be change. In double-effect cycles, refrigerant is condensed at a high enough temperature so that its heat output can be used to desorb more refrigerant from another generator. In practice, two cycles with different refrigerant adsorbent pairs would have to be used (e.g., active carbon-methanol and zeolite water as proposed by Meunier[ 131). A less complicated procedure is to utilize a regenerative cycle as is shown in Fig. 11. Two simple adsorption cycle machines are operated out of phase
COP
Te = -57 To= 3O'C Tg= 12O'C
0.8
so that when one is being heated the other is being cooled. Heat can be transferred from the cooling adsorbent to the heated adsorbent, reducing the external heat requirement and increasing the COP. No equivalent to the countercurrent heat exchanger exists for solids and so the effectiveness can never exceed 50%, but calculations suggest that in a typical case a COP of .4 could be boosted to S. There are major benefits to be gained from operation of a semicontinuous device, even if regeneration is not used. In solar refrigeration applications the simple cycle reaches its peak temperature by about 2 P.M. and then slowly cools down. This means that approximately half the daily available radiation is never used. A semicontinuous cycle would operate for the whole day, simultaneously heating and cooling and achieving roughly twice the ice production. In order to achieve this relatively large benefit, a simple cycle would have to be made to operate with perhaps a 30-min heating period followed by a 30-
COP
METHANOL
4 Te=-5°C Ta = 30°C Tg = 120°C
0.8
0.6
AMMONIA TEST RESULTS
0.6
0
I
I
I
I
0.2
0.4
0.6
0.8
m
V,llng
0
0.2
0.4
0.6
O.* V,l/b
Fig. 9. COP for a range of carbons using methanol refrig-
Fig. 10. COP for a range of carbons using ammonia re-
erant.
frigerant.
Activated carbon adsorption cycles PERIOD 1
2
69
3
4
PREHEAT
SOLARINPUT
COOLING
COOLTO AMBIENT
COOLING
COOLTO AMBIENT
PREHEAT
SOLARINPUT
Fig. 11. A semicontinuous
min cooling period. mass transfer.
This poses problems
regenerative adsorption cycle
of heat and
In the normal cycle, heating occurs over at least 4 h, and even then the temperature difference between the solar adsorber and the carbon can be as high as 25 K for the carbon methanol pair[ 141. Methanol appears to have a worse heat transfer problem than some other refrigerants, but heat transfer is seen as a major problem in all adsorption cycles. Several approaches may be considered to improve the conduction of heat through the packed bed of adsorbent.
1. Increase heat transfer area. Existing adsorption refrigerators all use fins to improve heat transfer, but this is at the expense of increasing the thermal mass of the reactor significantly. Work has been carried out by Guilleminot et a1.[15] to try to optimize fin depth, pitch, and thickness. 2. Use additives to improve conductivity. A patented method[l6] exists to improve the conductivity of the bed by a factor of 5 to 10 times. The additive has the form of a graphite powder, very finely dispersed in the adsorptive medium. 3. Producing active carbons with inherently higher conductivity. Although the microstructure precludes the conductivity of graphite combined with the adsorptive power of an activated carbon, it is not impossible to suppose a heterogenous structure, part of which is conducting and part absorptive. Also, since much of the thermal resistance is due to the point contact between grains, thought might be given to producing a carbon possessing a large-scale honeycomb structure that would not be broken up by the activation process. Activated carbon cloths are another class of materials that may have suitable adsorptive and thermal properties for refrigeration cycles.
7. CONCLUSIONS
Activated carbon adsorption cycles have great promise for small refrigeration systems, and in particular for solar-powered refrigerators for vaccine storage. The best refrigerants available are methanol and ammonia. Methanol should achieve cycle COPS of up to about S, and ammonia up to .4. However, other factors such as low system pressures and possible chemical instability do not make the use of methanol the obvious choice, and both need full evaluation. The porosity characteristics of available carbons give reasonable performance compared to the maximum that can be achieved with physical adsorption. Advances in heat and mass transfer to and from the carbon bed would allow the use of semicontinuous cycles that could improve the system performance by a factor of 2.
REFERENCES
1. R. Plank and J. Kuprianoff, In Die Kleinkiilternaschine. Springer-Verlag, BerliniG6ttingenlHeidelberg (1960). 2. World Health Organization, Solar Powered Refrigerators for Vaccine Storage and Icepack Freezing, Status Summary, June 1985, EPIICCIS185.4. 3. A. Harvey, Solar Icemaking in the Zambian Fi.rhing Economy, Development Technology Unit. Report
4. 5. 6. 7. 8.
No. 1. Engineering Department, Warwick University (1986). R. E. Critoph and R. Vogel, ht. J. Ambient Energy 7(4). Oct. 1986, 183-190. L. E. Haseler, J. M. Robertson, J. R. Henry. D. Chisolm, Absorption Cycle Heat Pumps for Domestic Heating, AERE-G 104R. AERE Harwell (1978). R. E. Critoph, Solar Energy 41, No. 1, 31-41 (1988). A. Michel, Reaslbierbarkeit eines monovalenten Zeolithl Wasser-Wiirmepumpen-Speicherheizgeriites, HLH Vol. 35, No. 9, Sept. 1984 pp. 425-431. M. Smisek and S. Cerny, In Active Carbon-Manufacture, Properties and Applications, Elsevier. Amsterdam-London-New York (1970).
70
R. E. CRITOPH
9. M. Pons and Ph. Grenier, Solar Ice Maker Working with an Activated Carbon-Methanol Adsorbent-Adsorbate Pair, INTERSOL 85, Proc. ISES Conf. Per-
gamon, New York (1985). 10. F. Meunier. J. J. Guilleminot, M. Gurgel, F. Meunier, G. Paye, M. Pons, Solar Powered Refrigeration Using Intermittent Solid Adsorption Cycles. Internal paper, Campus Universitaire, Orsay, France, (1986). 11. Brissonneau et Lotz Marine S. A., sales literature. Nancy, France (1987). 12. E. Passos, F. Meunier, and J. C. Gianola, J. Heat Recovery Systems 6. 259-264 (1986). 13. F. Meunier, J. Heat Recovery Systems 6,491-498 (1986). 14. J. J. Guilleminot, M. Gurgel. F. Meunier, G. Paye, M. Pons, Solid adsorption solar refrigerator system: A
15.
16. 17. 18.
condenser as part of solar reactor. Proc. ISES Solar World Congress, Hamburg, Pergamon, New York (1987). J. J. Guilleminot, F. Meunier, and J. Paklesa, An experimental and numerical (Two-dimensional) study of heat and mass transfers in a fixed bed of solid adsorbent in a transient state-establishment of a thermograviduct effect. Internal paper, Campus Universitaire, Orsay, France (1986). U.S. Patent 4,595,774. J. Davies, Sutcliffe Speakman, Private Communication. K. Sridhar, Studies on activated carbon-methanol pairs with relevance to ice-making, MSc. Thesis, AIT Bangkok, Thailand (1987).