10020

Ind. Eng. Chem. Res. 2008, 47, 1002010026

SEPARATIONS Separation of Ammonia/Water/Sodium Hydroxide Mixtures Using Reverse Osmosis Membranes for Low Temperature Driven Absorption Chillers
Simona Steiu, Joan Carles Bruno,*, Alberto Coronas, Ma Fresnedo San Roman, and Inmaculada Ortiz
UniVersitat RoVira i Virgili, CREVER-Grup dEnginyeria Te ` rmica Aplicada, AV. Pa sos Catalans, 26, 43007 Tarragona, Spain, and UniVersidad de Cantabria, ETSIIyT, Dpto de Ingenier a Qu mica y QI, AV. Castros s/n, 39005 Santander, Spain

The conventional working uids used in absorption chillers (water/lithium bromide and ammonia/water) present several disadvantages that limit their effective application. Recently, some works have reported the addition of NaOH to the ammonia/water working pair to improve the separation of ammonia in the generator, reducing the chiller driving temperature by taking advantage of the common ion effect. However, the presence of NaOH in the absorber has a negative impact on the absorption process. This study analyzes the technical viability of separating NaOH from ammonia/water/NaOH mixtures by using reverse osmosis membranes to incorporate this separation method into future chiller designs that work with these mixtures. The concentration range analyzed covers the solution concentration values of interest for absorption chiller applications (approximate 0.02-0.05 mass fraction of NaOH and 0.3 mass fraction of NH3). The results obtained show that, by using an in-series conguration of the modules, reverse osmosis technology is suitable for separating NaOH from the ternary mixtures studied.
1. Introduction and Objectives In recent years, the demand for refrigeration in buildings has increased signicantly mainly due to higher standards of indoor comfort conditions. This demand for refrigeration is usually covered by the electricity supply available, which produces a high peak load in the electricity distribution grid during the summer. To reduce this demand, absorption chillers activated with waste heat or thermal solar energy are an interesting option for the near future. Conventional working uids used in absorption chillers (water/LiBr and ammonia/water) have some drawbacks that limit the number of applications for absorption chillers: for example, limited solubility and corrosion problems in water/LiBr chillers and the need for rectication in the case of the mixture ammonia/water. Moreover, the relatively high driving temperatures for ammonia/water chillers are a problem for the application of these type of chillers in a eld that is attracting great interest, such as the solar assisted air-conditioning of buildings.1 In this case, the possibility of using ammonia/ water chillers to dissipate the absorption and condensation heat in the ambient air offers advantages in comparison with water/ LiBr chillers. The addition of other substances, such as hydroxides, can help to reduce these problems and improve the absorption cycle performance.2-5 In particular, the addition of NaOH (eq 1) to the ammonia/water mixture causes a shift in the solution equilibrium toward the ammonia gas form through the common ion effect, favoring a more effective separation of ammonia in the chiller generator with a lower energy requirement. NaOH f Na+ + OH(1)
NH3 + H2Or NH+ 4 + OH f

(2)

Existing concentration measurements of the vapor-liquid equilibrium of the mixtures ammonia/water/NaOH and ammonia/water/KOH conrm that the ammonia concentration in the liquid phase decreases signicantly as the addition of sodium or potassium hydroxide increases.6-8 As is expected and conrmed, comparing the equilibrium data from both ternary mixtures, the reduction of ammonia in the liquid phase is more signicant in the case of sodium hydroxide so it has been decided to conduct the study using this hydroxide. Other possible factors such as the corrosion potential have not been considered at this stage. The main benet from the addition of NaOH to the working uid is the reduction in the chiller driving temperature and consequently an increase in the cycle COP (ratio between the chiller refrigeration capacity and the required heat supplied). However, the presence of NaOH in the absorber produces a negative effect because it reduces the ammonia absorption capacity of the solution with lower ammonia content, the socalled weak solution. Thus, it is necessary to include some kind of separation system between the generator and absorber9,10 in the refrigeration cycle working with ammonia/water/NaOH mixtures. Up to now, several separation systems have been proposed for the separation of the hydroxyl ions from the ternary mixtures.5 In this study, we examine the analysis of the technical viability of the reverse osmosis system. As an example of the advantages of the addition of hydroxides to ammonia, Table 1 shows the COP values for an absorption
Table 1. COP Values as a Function of the NaOH Mass Fraction10 NaOH (mass fraction) 0 0.05 0.10 tgenerator (C) 107.1 90.8 75.5 COP 0.48 0.65 0.88

* To whom correspondence should be addressed. E-mail: juancarlos.bruno@urv.cat. Universitat Rovira i Virgili. Universidad de Cantabria.

10.1021/ie8004012 CCC: $40.75 2008 American Chemical Society Published on Web 11/14/2008

Ind. Eng. Chem. Res., Vol. 47, No. 24, 2008 10021

Figure 1. Absorption refrigeration cycle using ammonia/water/NaOH mixtures.

Figure 2. Reverse osmosis experimental setup conguration. Table 2. Technical Data of the FILMTEC-DOW Membranes

chiller of 135 kW (chilled water capacity) with the corresponding NaOH mass fractions in the ammonia/water mixture.10 Table 1 shows that the driving temperature goes down and COP values grow as more NaOH is added to the cycle. The evaporation temperature of the cycle for the case is shown in Table 1. The aim of this article is to study the technical viability of the separation of NaOH from ammonia/water/NaOH mixtures by using reverse osmosis technology in the concentration range useful for the development of new absorption chillers based on these mixtures to improve performance with respect to conventional ammonia/water chillers, that is, reducing the required driving temperature and increasing the cycle COP. When two solutions at different concentrations are placed on both sides of a semipermeable membrane, the solvent moves from the more diluted solution through the membrane to the other side, until the chemical potential on both sides is equal. This spontaneous phenomenon is known as direct osmosis. The osmotic pressure exerted on the membrane is a measurement of this concentration difference. The reverse osmosis process produces the opposite effect. An external pressure higher than the osmotic pressure is applied to a concentrated solution in contact with a semipermeable membrane, and the solvent permeates across the membrane. In this case, two streams are generated, one with the solvent that crosses the membrane (called permeate) and another one (rejection) more concentrated in solute than the feeding solution.11,12 Figure 1 shows a block diagram of the proposed absorption refrigeration cycle using a reverse osmosis unit to separate the ammonia/water/NaOH mixture. The poor solution coming from the generator with large sodium hydroxide content enters the reverse osmosis unit that may consist of several membrane modules. In this unit, the feed is separated into two solutions: one of them more concentrated in hydroxide (stream 1) that is recycled back to the generator together with the rich solution from the absorber, and the second one with a reduced content in hydroxide that is sent directed to the absorber (rich solution, stream 8). COP values of the cycle shown in Figure 1 are a function of the temperatures of the cycle (evaporator, condenser, absorber, and generator) and also depend on the concentration of NaOH and the efciency of the separation system. 2. Experimental Setup The experimental setup used to carry out the experiments consists of a laboratory scale stainless steel (316SS) reverse osmosis cell SEPA CF, prepared to withstand pressures of up to 69 bar. The main characteristics of this cell are: max.

membrane model membrane area material max. operating pressure operating pressure pH operating range max. operating temperature NaCl rejection (%) max. permeate ow (L/m2 h) permeability coefcient (L m2 h bar)

BW30 155 cm2 polyamide 41 bar 40 bar 2-11 45 C 99 120 3

SW30HR LE 155 cm2 polyamide 83 bar 40 bar 2-11 45 C 99.6 80 2

temperature, 177 C; max. pressure, 69 bar; and ooding volume, 70 mL. The membrane effective area is 155 m2. The cell is designed to simulate the membrane uid dynamic behavior of industrial modules, testing their performance in a wide range of operating conditions. The experimental setup conguration employed for the separation of NaOH from the mixture of ammonia/water is shown in Figure 2. As shown in Figure 2, the experimental setup consists of a cell and cell body, a hydraulic pump, a positive displacement circulation pump, a manometer, a concentrate control valve, and the necessary storage tanks for the feed, permeate, and concentrate when the plant is operated in a continuous mode. The experimental procedure was as follows: once the membrane had been placed in the cell body, hydraulic pressure was applied to produce the required hermetic sealing, thus ensuring a good pressurization. The circulation pump ensures the desired feeding solution ow rate from the tank to the cell body at the required operating pressure. The inlet solution is parallel to the membrane, and the ltration takes place tangentially producing the permeate ow. Cleaning of the system is necessary as soon as a decrease of permeate ow is perceived. The concentrate ow, that is, the ow that does not cross the membrane, is controlled using a stainless steel valve that produces the required trans-membrane pressure on the membrane surface. The continuous ow operation is simulated using three different tanks for the feeding solution, permeate, and concentrated solution. The objective pursued with the continuous operation mode is to simulate the continuous ow of solution in and out of the membrane and to avoid the increase in osmotic pressure if the system is operated in discontinuous mode. The reverse osmosis membranes used in the experiments conducted were BW30 and SW30HR LE, both supplied by Filmtec-Dow. These membranes come in tubular form, and it was necessary to adapt them to the shape corresponding to the membrane cell. The characteristics of these membranes are given in Table 2. The NaOH/water and ammonia/water/NaOH solutions used as feed were prepared by dissolving NaOH pellets and ammonia, with a purity of 28-30% and a density of 0.9 kg/L; both reagents were supplied by MERCK.

10022 Ind. Eng. Chem. Res., Vol. 47, No. 24, 2008

The experimental results were monitored by measuring the solution conductivity using a CRISON BASIC 30 conductimeter in the range of 0.01-199.9 mS/cm. For the analysis of the CHI was used ammonium ion, a K-355 distillation unit from BU with a detection limit of nitrogen g0.1 mg. The analytical method of this unit is based on the Kjeldahl method. Additionally, permeate samples were collected every 10 min, and NaOH concentration was determined. In the case of the experiments with solutions that also contained ammonia, only the composition of the nal product was analyzed because of the difculty in manipulating solutions with high ammonia content. The composition of NaOH and ammonia of the nal product was determined with the conductimeter BASIC 30 for NaOH and the distillation unit K-355 for ammonia. Once the experiment was completed, the nal volume of the permeate tank was used to calculate the permeate ux. 3. Results and Discussion The parameters commonly used to determine the separation efciency factor in reverse osmosis membranes are the permeate ux through the membrane and the rejection coefcient. The permeate ux or solvent ux (Jw) is given by the following equation: Jw ) K(P - ) (3)

Table 3. Experimental Planning with Water/NaOH Solutions (Task 1) membrane BW30 experiment EXP 1 EXP 2 EXP 3 EXP 4 feed (mass fraction) 0.007 0.013 0.019 0.042 membrane SW30HR LE experiment EXP 5 EXP 6 EXP 7 EXP 8 feed (mass fraction) 0.008 0.013 0.021 0.039

Table 4. Experimental Planning with Ammonia/Water/NaOH Ternary Solutions (Task 2) membrane BW30 feed (mass fraction) experiment EXP 9 EXP 10 EXP 11 EXP 12 EXP 13 EXP 14 NH3 0.0786 0.0632 0.1959 0.1922 0.3093 0.3028 NaOH 0.0172 0.0321 0.0146 0.0432 0.0154 0.0449 experiment EXP 15 EXP 16 EXP 17 EXP 18 EXP 19 EXP 20 membrane SW30HR LE feed (mass fraction) NH3 0.0755 0.0659 0.1984 0.1928 0.3028 0.3024 NaOH 0.0159 0.0290 0.0144 0.0400 0.0148 0.0418

Table 5. Calculated Osmotic Pressure as a Function of the NaOH Concentration feed solution (mass fraction) 0.012 0.020 0.028 0.050 osmotic pressure (bar) 4.6 8.4 11.8 24.0

where K is the permeability coefcient characteristic of each membrane, P is the transmembrane differential pressure, and is the osmotic pressure difference. The term (P - ) represents the net driving force that is responsible for the solvent permeation through the membrane. To calculate the osmotic pressure developed in the experiments, the following equation based on thermodynamic considerations13 was used: ) 0.08308(T + 273.15)

(4)

where is the osmotic coefcient (usually 0.93 for brackish water and 0.902 for seawater), T is temperature in C, and mi is molality of all of the solution, ionic, or nonionic components in mol/L that can be calculated by: mi ) Ci 1000MWi 106 - TDS 106 (5)

where Ci is the concentration of component i in the solution, in mg/L; MWi is the molecular weight of component i; and TDS is the total dissolved salt content in the solution, in mg/L. The retention or retention coefcient (R) is dened as: R) 1Cpermeate Cfeed (6)

This parameter expresses the extent to which a solute is retained by the membrane. In an ideal case, the retention coefcient could be 1 or 100%, that is, the membrane does not allow the solute to cross the permeate stream (Cpermeate ) 0). In practice, it is not possible to achieve a retention of 100%, but values between 95% and 99% are possible.14 Besides the permeate ux and the retention coefcient, the variables that inuence the reverse osmosis performance are temperature (inuences the ux through the membrane and the tendency to compaction), net pressure, ux density, solute concentration in the permeate and concentrate, and the membrane lifetime.

The experimental study is described below. First (task 1), the separation of binary solutions water/NaOH was tested to determine the technical viability of the separation of NaOH using reverse osmosis technology. In the second experimental step (task 2), the separation of ternary mixtures ammonia/water/ NaOH was analyzed to determine the inuence of the presence of ammonia in the NaOH separation and the permeate ow rate with respect to the results of task 1. For the purpose of this work, the concentrations of NH3 and NaOH used in the experiment are close to the ones that could be found in a refrigeration absorption cycle (approximate 0.01-0.05 mass fraction NaOH and 0.1-0.3 mass fraction NH3). Each experiment was conducted twice and with both membranes until steady state was reached. Table 3 shows the operating conditions of experiments 1-8 using the water/NaOH mixtures, and Table 4 provides similar information for experiments 9-20, using the ternary mixtures. In both cases, the transmembrane pressure was set to 39 bar, the operating maximum pressure for the BW30 membrane. This pressure is high enough to overcome the osmotic pressure of the most concentrated analyzed solutions and corresponds to the difference between the pressure of the concentrate set by the control valve (40 bar) and the pressure on the permeate stream side, atmospheric pressure. The theoretical osmotic pressure for each solution was calculated using eqs 4 and 5. The results are shown in Table 5. As seen in this table, as the NaOH concentration increases, the osmotic pressure also increases. That means that working with the same membrane and the same transmembrane pressure, the permeate ux decreases as the NaOH mass fraction increases. Thus, for the most concentrated NaOH solutions (0.05 mass fraction), it is necessary to pressurize the feed solution at pressures higher than 24 bar to obtain a signicant permeate ux. 3.1. NaOH/Water Binary Mixture. Table 6 gives the values of permeation ux and retention coefcients corresponding to experiments 1-8.

Ind. Eng. Chem. Res., Vol. 47, No. 24, 2008 10023
Table 6. Permeate Flux and Retention Percentage in Experiments EXP 1-8 ux (L/m2 h) experiment EXP EXP EXP EXP 1/EXP 2/EXP 3/EXP 4/EXP 5 6 7 8 BW30 (EXP 1-4) 49.3 ( 0.6 35.9 ( 0.7 24.7 ( 0.8 24.3 ( 0.6 SW30HR LE (EXP 5-8) 16.0 ( 1.1 11.9 ( 0.3 9.6 ( 0.7 8.4 ( 0.3 BW30 (EXP 1-4) 56.5 48.7 47.5 31.7 retention (%) SW30HR LE (EXP 5-8) 91.5 90.2 87.0 87.6

According to the results presented in Table 6, both the permeate ux and the retention coefcients decrease working with both membranes as the NaOH mass fraction increases, as is to be expected from the increase in the osmotic pressure. The permeate ux obtained at the lowest NaOH mass fraction (approximate 0.012) is between 2.5 (BW30) and 5 (SW30HR LE) times lower than the maximum ux shown in Table 3 when the membranes are used to separate water and NaCl. Regarding the results of the retention coefcient, working with both membranes, a better separation is achieved with the SW30HR LE than using the BW30 membrane, although the permeate ux obtained is lower for the former membrane (SW30HR LE). Figures 3 and 4 present the NaOH mass fraction in the permeate ux as a function of time for EXP 1-8. As shown in Figures 3 and 4, membrane SW30HR LE leads to a better separation for the same operating conditions than does the membrane BW30, in accordance with the results of the retention coefcient shown in Table 6. As can be see in Table 6, with one separation step, the retention values are lower than required for the future application of the separation process (it is necessary to maximize the separation of NaOH and to avoid the presence of hydroxyl ions in the absorber of the refrigeration cycle). To increase efciency in the retention of NaOH, a separation methodology is proposed to simulate the operation of a membrane process in-series. This methodology consists of performing an experiment with high concentration of NaOH and to use the permeate obtained as feed in a subsequent step, and so on, in such a way that a membrane conguration in-series is simulated. With this

procedure it is expected to achieve two objectives: (i) retention values of 99% and (ii) higher permeate uxes. This is also a suitable methodology for overcoming the osmotic pressure of highly concentrated solutions. The initial feed solution concentrations for both membranes (BW30 and SW30HRLE) were those corresponding to EXP 4 and EXP 8. With each membrane, a complete cycle of experiments was conducted, that is, successive separation steps until a retention value close to 99% was achieved. In the case of membrane BW30, the cycle of experiments was named SERIES_A, and for membrane SW30HRLE it was named SERIES_B. As a typical example of the results obtained, Figures 5 and 6 show the time evolution of the NaOH mass fraction in the permeate at each step of both cycles of experiments. As shown in Figures 5 and 6, to achieve a good separation of NaOH working with membrane BW30, four separation steps were necessary, whereas only two steps were needed with membrane SW30HRLE. Table 7 shows the values of permeate ux and retention coefcient obtained in experiments SERIES_A and SERIES_B. It is worth noting that the permeate ux values are higher for the BW30 membrane (28.2 L/m2 h) than for the SW30HRLE, as was also the case in the one-step experiments performed previously. The retention parameter calculated in the last step ranges between 86% (membrane BW30) and 94% (membrane SW30HRLE). In the case of the membranes connected in series, the total retention parameter was calculated as the ratio between the concentration in the permeate ow in the last separation step and the feed concentration in the rst step (Figure 7):

Figure 3. Time evolution of the NaOH mass fraction in the permeate solution for membrane BW30.

Figure 5. Time evolution of the NaOH mass fraction in the permeate. Experiments SERIES_A.

Figure 4. Time evolution of the NaOH mass fraction in the permeate solution for membrane SW30HR LE.

Figure 6. Time evolution of the NaOH mass fraction in the permeate. Experiments SERIES_B.

10024 Ind. Eng. Chem. Res., Vol. 47, No. 24, 2008

Figure 7. Schematic diagram of a reverse osmosis system connected in series.

Figure 8. Inuence of the NH3 concentration in the permeate ux (NaOH approximate feed concentration ) 0.02).

Rtotal ) total retention ) 1 -

Cpermeate n+1 Cfeed

(7)

Figure 9. Inuence of the NH3 concentration in the permeate ux (NaOH approximate feed concentration ) 0.05). Table 7. Flux Permeate and Retention Coefcient for the Cycle of Experiments SERIES_A and SERIES_Ba membrane BW30 (SERIES_A) step
1 2 3 4
a

For both membranes, the calculated total retention after four steps (membrane BW30) and two steps (membrane SW30HR LE) was 99%. 3.2. Ternary Mixture Ammonia/Water/NaOH. The results presented above show the technical viability of the separation of NaOH from aqueous solutions in the range of mass fractions between approximately 0.01 and 0.05 NaOH. As stated in Table 4, in the experiments corresponding to task 2 EXP 9-20, ammonia was added to the studied solutions at several concentrations (approximately 0.1, 0.2, and 0.3 ammonia mass fraction). The aim of this task was to analyze the effect of ammonia on the separation of NaOH from the solution and to determine the viability of the NaOH separation from the ammonia/water/ NaOH ternary mixture. As with task 1, the results were analyzed by measuring the nal permeate ux, the retention, and the mass fraction of OH- and NH3, in the permeate as well as in the nal concentrate stream (Figure 8, Figure 9, Table 8). The experimental results presented in Figures 8 and 9 show that the addition of ammonia leads to a slight decrease of the permeate ux in comparison with the values obtained with the NaOH aqueous solutions. This reduction in the permeate ux is justied by a slight increase in the osmotic pressure retentate solutions that also reduces the separation of NaOH (Table 8). Tables 9 and 10 show the inuence of the ammonia concentration in the permeate concentration. The results presented in Table 9 represent the NaOH concentration in the permeate solution for the last separation step corresponding to the water/NaOH binary mixtures and to the ammonia/water/ NaOH ternary mixtures with a concentration of NaOH in the feed stream of approximately 0.02. Likewise, Table 10 shows the NaOH approximate concentration in the permeate solution for the binary mixture water/NaOH and for the ternary mixtures ammonia/water/NaOH. In this case, the NaOH mass fraction in the feed solution is approximate 0.05. To calculate the mass

membrane SW30HR LE (SERIES_B) ux (L/m2 h)

8.4 ( 0.3 13.8 ( 0.6

ux (L/m2 h)
24.3 ( 0.6 25.6 ( 0.6 26.6 ( 0.7 28.2 ( 0.8

retention (%)
31.7 57.7 69.3 86.3

retention (%)
87.6 94.1

The retention was calculated with molar concentration of NaOH.

Table 8. Retention Percentage in Experiments EXP 9-20 retention (%) (calculated with mass fraction) experiment EXP EXP EXP EXP EXP EXP 9/EXP 15 10/EXP 16 11/EXP 17 12/EXP 18 13/EXP 19 14/EXP 20 BW30 (EXP 9-14) 39.2 20.1 38.9 21.5 37.5 21.1 SW30HR LE (EXP 15-20) 73.4 31.3 71.2 31.2 70.2 31.8

fraction of each component, the parameters densities presented in Salavera et al.6 were used. The presence of ammonia in the solution reduces the efciency of the NaOH separation using both membranes.According to the results presented in Tables 9 and 10, the NaOH concentration in the permeate is lower for the binary mixtures (EXP 2, 6, 4, and 8) with respect to the ternary mixtures (EXP 9-20), so the retention is greater for the binary mixtures. Furthermore, it is interesting to note that the ammonia concentration is almost the same in the permeate and in the concentrate solutions; that is, the ammonia is not rejected by either of the membranes and goes through them freely. The error standards in the measurements of the results, calculated for ammonia and NaOH mass fraction (in the feed,

Ind. Eng. Chem. Res., Vol. 47, No. 24, 2008 10025
Table 9. Inuence of the NH3 Concentration in NaOH Concentration in the Permeate (NaOH Feed Approximate Concentration ) 0.02) feed NH3 concentration (mass fraction) experiment EXP EXP EXP EXP 2/EXP 6 9/EXP 15 11/EXP 17 13/EXP 19 membrane BW30 0 0.0786 0.1959 0.3093 membrane SW30HR LE 0 0.0755 0.1984 0.3028 permeate NaOH concentration (Mass fraction) membrane BW30 0.0066 0.0104 0.0089 0.0097 membrane SW30HR LE 0.0012 0.0042 0.0041 0.0044

Table 10. Inuence of the NH3 Concentration in NaOH Concentration in the Permeate (NaOH Feed Approximate Concentration ) 0.05) feed NH3 concentration (mass fraction) experiment EXP EXP EXP EXP 4/EXP 8 10/EXP 16 12/EXP 18 14/EXP 20 membrane BW30 0 0.0632 0.1922 0.3028 membrane SW30HR LE 0 0.0659 0.1928 0.3024 permeate NaOH concentration (Mass fraction) membrane BW30 0.0286 0.0256 0.0339 0.0355 membrane SW30HR LE 0.005 0.0219 0.0272 0.0290

Table 11. Permeate Flux and Retention Values for Experiments of Ternary Mixtures SERIES_C and SERIES_Da membrane BW30 (SERIES_C) step
1 2 3 4 5 6 7 8
a

membrane SW30HR LE (SERIES_D) ux (L/m2 h)

7.3 ( 0.8 9.2 ( 0.8 12.3 ( 0.8 15.6 ( 0.7

ux (L/m2 h)
22.6 ( 0.6 23.9 ( 0.8 24.2 ( 0.6 25.4 ( 0.6 30.0 ( 0.7 32.3 ( 0.8 34.6 ( 2.0 37.2 ( 1.0

retention (%)
21.7 30.2 33.1 38.2 47.6 56.1 65.1 79.2

retention (%)
30.9 48.8 61.2 78.3

The retention was calculated with molar concentration of NaOH.

in the case of membrane SW30HR LE. In both cases, the number of necessary steps is higher than the required number for the binary mixtures that started with a feed of NaOH mass fraction approximately of 0.05. The permeate ux and retention values for both series of experiments are presented in Table 11. As shown in Table 11, membrane BW30 allows higher permeate ux than does membrane SW30HR LE. However, the retention percentages are higher for the latter membrane, following behavior similar to that observed in the experiments with the binary mixture NaOH/water. The total retention coefcient was 99% for both membranes. The values of permeate ow rate obtained in this work are still lower than the values required in the operation of an absorption chiller.10 This difference can be explained by considering the small membrane surface (0.155 m2). Membranes with a larger surface or modules in parallel could be used to increase the permeate ow rates. The experimental results will be used to perform simulations of the refrigeration cycle with ammonia using Aspen Plus simulation software to evaluate the separation process as compared to a conventional ammonia-water cycle. 4. Conclusions The addition of NaOH to ammonia/water absorption refrigeration cycles reduces the temperature required to drive the cycle and increases its efciency. To obtain a signicant improvement in the process performance by addition of NaOH to the cycle, and to avoid its presence in the absorption section of the cycle, it is necessary to incorporate a system that separates the hydroxyl ions from the solution sent to the absorber. The use of reverse osmosis technology is a suitable system for achieving this separation. The following conclusions can be drawn from the experimental results obtained from this work. The separation of NaOH from their aqueous solutions with a maximum mass fraction of 0.05 (5% weight) is viable using reverse osmosis with the tested membranes, BW30 and SW30HR LE, providing enough contact time of the solution with the experimental data of the membrane in series conguration. In this case, maximum retention values between 86% and 94% and average permeate uxes between 14 and 28 (L/m2 h) were obtained. The experiments carried out with the ternary mixtures ammonia/water/hydroxide reveal a slight inuence of the ammonia on the separation of NaOH with respect to the binary mixture NaOH/water. Comparing the results obtained with binary and ternary mixtures, it was observed that the permeate ux and the percentage of retention decrease slightly with the

Figure 10. Permeate NaOH mass fraction of experiments SERIES_C and SERIES_D.

permeate, and retentate streams), are 0.05% and between 0.001% and 0.02%, respectively, for the experiments with both types of membranes. Following the same methodology employed in task 1, two new sets of experiments were performed to simulate the conguration of membranes connected in series: SERIES_C (membrane BW30) and SERIES_D (membrane SW30HR LE). The initial feed solutions concentrations for both membranes were those corresponding to EXP 14 and EXP 20 of task 2 (Table 4). In this methodology, the permeate collected from each step was used as feed solution in the following step. Figure 10 shows the NaOH mass fraction in the permeate solution corresponding to experiments SERIES_C and SERIES_D. The rst step indicated in Figure 10 corresponds to the feed concentration (for the BW30 membrane is 0.0449 mass fraction and 0.0418 for the other membrane) and not to the rst permeate concentration. To maximize the separation of the NaOH from the studied ternary mixture, seven separation steps are necessary using membrane BW30, whereas only three steps are needed

10026 Ind. Eng. Chem. Res., Vol. 47, No. 24, 2008

ternary mixture. In the range of solution concentrations studied (approximate 0.1-0.3 ammonia mass fraction, 10-30% weight), the retention parameter reached values of 78-79% and the permeate ux comprised between 16-37 (L/m2 h) depending on the NaOH concentration. The ammonium ion is not retained; it permeates through both types of membranes, as almost the same concentration was measured in the permeate and concentrate solutions. These results favor the use of these membranes in absorption refrigeration cycles because they will not modify substantially the ammonia concentration at the absorber inlet. Acknowledgment We appreciate the funding support from the Spanish Science and Education Ministry through research projects DPI200304752 and CTQ2006-14360/PPQ. S. Steiu thanks Departament dUniversitats, Recerca i Societat de la Informacio de la Generalitat de Catalunya i del Fons Social Europeu, for a scholarship. Literature Cited
(1) Henning, H. M. Solar assisted air conditioning of buildings - an overview. Appl. Therm. Eng. 2007, 27, 1734. (2) Peters, R.; Greb, O.; Korinth, C.; Zimmermman, A. Vapor-liquidequlibria in the system NH3-H2O-LiBr, Part I: Measurements in the range T ) 303-423 K and p ) 0.1-1.5 MPa. J. Chem. Eng. Data 1994, 40, 769. (3) Peters, R.; Korinth, C.; Keller, J. U. Vapor-liquid equilibria in the system NH3-H2O-LiBr.2. Data correlation. J. Chem. Eng. Data 1995, 40, 775. (4) Reiner, R. H.; Zaltash, A. Densities and viscosities of ternary ammonia water uids. ASME Winter Annual Meeting, 1993. (5) Reiner, R. H.; Zaltash, A. Evaluation of ternary ammonia-water uids for GAX and regenerative absorption cycles. Report ORNL/CF-91/263, 1991.

(6) Salavera, D. Propiedades Termosicas de Nuevos Fluidos de Trabajo (H2O+LiBr+LiNO3+LiCl+LiI, NH3+H2O+NaOH y NH3+H2O+KOH) para Sistemas de Refrigeracio n por Absorcio n. Thesis Doctoral, Universidad Rovirai Virgili, Tarragona, Espan a, 2005. (7) Brass, M.; Pritzel, T.; Schulte, E.; Keller, J. U. Measurement of vapor-liquid equilibria in the systems NH3-H2O-NaOH and NH3-H2O-KOH al temperatures of 303 and 318 K and pressures 0.1 MPa < p < 1.3 MPa. Int. J. Thermophys. 2000, 21, 883. (8) Salavera, D.; Chaudhari, S. K.; Esteve, X.; Coronas, A. Vapor-liquid equilibria of ammonia + water + potassium hydroxide and ammonia + water + sodium hydroxide solutions at temperatures from (293.15 to 353.15) K. J. Chem. Eng. Data 2005, 50, 471. (9) Balamuru, V. G.; Ibrahim, O. M.; Barnett, S. M. Simulation of ternary ammonia-water-salt absorption refrigeration cycles. Int. J. Refrig. 2000, 23, 31. (10) Bruno, J. C.; Vidal, A.; Esteve, X.; Coronas, A. Modelling and thermal analysis of ammonia-water-sodium hydroxide absorption refrigeration cycles. International Sorption Heat Pump Conference, ISHPC, Colorado, EE.UU, 2005. (11) Senica, D.; Rzen, J.; Ustar, M.; Drnovsek, P.; Kogej, S. Reverse osmosis to concentrate lisinopril puried by means of liquid chromatography: From pilot-plant to industrial-scale unit. Ind. Eng. Chem. Res. 2005, 44, 18601867. (12) Alexiadis, A.; Bao, J.; Fletcher, D. F.; Wiley, D. E.; Clements, D. J. Analysis of the dynamic response of a reverse osmosis membrane to time-dependent transmembrane pressure variation. Ind. Eng. Chem. Res. 2005, 44, 78237834. smosis inVersa fundamentos, tecnolog (13) Farin as, I. M. O a y aplicaciones; McGraw-Hill: New York, 1999. (14) Pomerantz, N.; Ladizhansky, Y.; Waisman, M.; Daltrophe, N.; Gilron, J. Prevention of scaling of reverse osmosis membranes by zeroing the elapsed nucleation time. Part I. Calcium sulfate. Ind. Eng. Chem. Res. 2006, 45, 20082016.

ReceiVed for reView March 11, 2008 ReVised manuscript receiVed October 6, 2008 Accepted October 7, 2008 IE8004012