3 REGENERATION PROCESS BY REVERSE OSMOSIS (RO) AND VACUUM MEMBRANE DISTILLATION (VMD)

Transcription

1 27 3 REGENERATION PROCESS BY REVERSE OSMOSIS (RO) AND VACUUM MEMBRANE DISTILLATION (VMD) Introduction It is known that lithium bromide absorbtion refrigeration system (LBARS) differs from the more prevalent compression chillers in that its cooling performance is driven by heat energy, rather than by mechanical energy. Although the Coefficient of Performance (COP) in absorbtion refrigeration system is much lower than that in compression chillers (Vargas et al., 2009), LBARS tends to have higher energy-saving performance since it can utilize or callback waste heat. With energy shortage and price booming of fossil fuel, low grade energy such as solar energy, geothermal energy and low temperature waste heat energy has become popular in the application of LBARS. However, not all regenerative energy can be used as generator heater because the temperature of most regenerative energy is relatively lower than 85 C (Ma et al., 1998). Traditional single-effect LBARS can t operate effectively when driven by heat resource below 80 C because of low COP (Vargas et al., 2009), and aqueous lithium bromide solution must be heated to its boiling point so that water vapor can vaporize from the solution and be transported continually. Many different methods have been presented to improve COP of the absorbtion refrigeration system. Gu et al. (2006, 2008) proposed a solar pump-free lithium bromide absorbtion chiller system with a second generator, whose temperature scope was from 80 to 93 C. The average COP of the system was Wu et al. (2007) improved the absorbtion refrigeration system by adding a vapor compressor between the generator and the condenser. The vapor compressor could reduce the vapor pressure in generator when the driving heater temperature was too low, and the COP of the system could be adjusted from 0.65 to 0.75 according to the generator temperature. However, the vapor compressor seemed too complicated due to the larger specific volume of water vapor. Ma et al. (1998) and Sumathy et al. (2002) introduced a twostage lithium bromide absorbtion refrigeration system which could apply low temperature heater from 70 C to 85 C. However, the COP was only 0.39 and the system seemed more complicated and the initial cost would increase one third. Moreover, the large contact area needed to separate water vapor from aqueous lithium bromide solution makes traditional generator too bulky and heavy to be fitted into small scale device (Kim et al., 2008). Therefore, from the viewpoint of low grade thermal energy application and facility miniaturization, traditional generator in LBARS seems deficient to some extent. Membrane technology in absorbtion refrigeration system has been previously reported by several researchers although still focus on the membrane performance. Riffat and Su(1998) used centrifuge reverse osmosis (RO) membrane in a refrigeration system to reduce the utilization of high pressure pump. The research found that ratehigher than rpm at r=50 mm was required in order to obtain solution concentration of 64%. The disadvantage of this system was the uses of high velocity which corresponds to the increasing of mechanical energy used in the system. Another research was conducted by Wang et al., (2009) who used membrane distillation technology based on PVDF-hollow

2 28 fiber module for LiBr-H 2 O separation. In this research, several parameters i.e. feed flux, temperature in lumen side and vacuum pressure in shell lumen were observed. It was found that the increasing of feed temperature and feed flux will increase the water vapor permeateion flux. Ahmed and Peter(2009) conducted an experiment to analyze the effect of membrane characteristic towards the absorbtion process in absorber. A good absorbtion performance was obtained from membrane characteristic which has high permeability upon water vapor, uses high pressure hydrophobic membrane to avoid membrane pore wetness, and no water vapor capillary condensation to avoid membrane pore block. While the problem associates with low value of COP in absorbtion refrigeration system should analyze the effectiveness of energy utilization and exergy analysis. The objective of this research is to study the separation characteristic of LiBr-H 2 O solution using reverse osmosis (RO) membrane, study the characteristic of separation products during refrigeration process, study was to determine the important parameters in the process of separation-aqueous LiBr and optimal parameter VMD application on absorbtion refrigeration system LITERATURE REVIEW Reverse Osmosis Membrane One of membrane type can be used as a regenerator is the reverse osmosis membrane. A reverse osmosis membrane acts as a barrier to flow, allowing selective passage of a particular species (solvent) while other species (solutes) are retained partially or completely. Solute separation and permeate water (solvent) flux characteristics of membrane depend on the membrane material selection, the preparation procedure, and the structure of membrane barrier layer. Primary separation of solutes occurs at thin film (skin) barrier layer. Two situations may arise depending on the extent of solvent membrane affinity i.e. solvent preferential and solute preferential sorption. Osmosis is natural phenomenon in which water passes through a semipermeable (no solute flow) membrane from higher solute concentration region until reaches equilibrium at the both sides, solvent (water) chemical potential is restored. At equilibrium the pressure difference between the two sides of the membrane is equal to the osmotic pressure difference. To reverse the flow of water, a pressure difference which greater than the osmotic pressure difference is applied, as a result, separation of water from solution becomes possible. This phenomenon is termed as reverse osmosis or hyper filtration. Reverse osmosis processes are classified into three types: high pressure RO (5.6 to 10.5 MPa, such a seawater desalination), low pressure RO (1.4 to 4.2 MPa, such brackish water desalination) and nano filtration or loose RO ( 3 to 4 MPa, such a partial demineralization or 0 to 20 % NaCl rejection). High and low pressure RO is typically used for very high rejection of inorganics (95 to 99.9% NaCl rejection) and for moderate to high rejection of low molecular weight organics, respectively. Organics rejection depends on membrane polymer types and structures and membrane/solute interactions. The terminology used for pressure driven membrane process (such RO) was recently reviewed by Gekas (1988). The important operating variables for RO are

3 29 feed flow rate and concentrations of dissolved solutes, types of solutes, trans membrane pressure (Δp), temperature (T), ph and concentration of suspended solids (if any). Any membrane process produces two streams i.e. the permeate (portion of the feed passing through membrane) and the retentatee or concentrate (portion of the feed not passing through membrane). A schematic of RO process is shown in Figure 3-1. High Pressure Side Feed Δp-Δπ = net or effective transmembrane pressure difference Δp = transmembrane pressure difference c b = bulk solute consentration c w=c wall=solute consentration at membrane surface Membrane Retentate Or Consentrate Solute Consentration :c r or c c Flow rate : Qr or Qc Low Pressure Permeate Side Jw=water flux (m/s, cm/s, kg/m2 s, kmol/m2s, gal/ft2 day) C = permeate solute consentration Js = solute flux (kmol/m2s, kmol/cm2 s) Qp = permeate flow rate Figure 3.1 Schematic of reverse osmosis mechanism Vacuum membrane distillation (VMD) Vacuum membrane distillation (VMD) is one of method utilized for driving flux in membrane distillation. Membrane distillation (MD) itself defined as a thermally driven process, in which only vapor molecules are transported through porous hydrophobic membrane (El Bourawi et al, 2006). Because of hydrophobic membrane there is only one side of membrane that contact with liquid solutions, resulting a vapor liquid interface at each pore entrance (Figure 3.2). To make vapor go across the membrane, some known driving force are: direct contact with condensing fluid (DCMD), an air gap separated condensing surface (AGMD), a sweeping gas (SGMD), or a vacuum (VMD). Each of MD configuration has its own advantage and specific application field (El Bourawi et al., 2006). The summary of each MD system is presented at Table 3.1.

4 30 Figure 3.2 Vapour-liquid surface at each pore of membrane distillation VMD process is quite similar with pervaporation (PV) process. The main difference between both of them is that in VMD use porous membrane and thus, the transport is based on flow through pores; while PV use dense membrane so that the transport generally described by the solution diffusion mechanism. In this research, VMD is also preferable than nanofiltration since the liquid solution used in this experiment is LiCl which is monovalent ion. In nanofiltration (NF) process, monovalent ion still go through the membrane, thus if it happen at the generator of absorbtion refrigeration, it will decrease the purity of refrigerant and will be resulted in lower performance.

5 31 Table 3-1. Summary of membrane distillation configuration Membrane Configuration System configuration System description Application Direct contact membrane distillation (DCMD) An aquaeous solution colder than the feed is in direct contact with permeates side of the membrane. Drawback is heat lost by conduction. Desalination, concentration of aqueous solution in food industries, acids manufacturing Air Gap membrane distillation (AGMD) A stagnant air gas introduced between the membrane and condensation surface. The stagnant air is to reduce heat loss. Desalination, removing volatile compounds from aqueous soltion Sweeping gas membrane distillation (SGMD) Vacuum membrane distillation (VMD) A cold inert gas sweeps the permeate side and carry the vapour molecules to condensate it outside membrane module. The moving gas enhance mass transfer coefficient. Need large condenser Vacuum is applied in the permeate side by mean of vacuum pump. Heat loss by conduction is negligible Removing volatile compounds from aqueous soltion Separation of aqueous volatile solution 31

6 32 Both NF and VMD experiment had been conducted on many applications. For desalination purposes, NF usually placed as pretreatment (AM Hassan et al, 2000) for SWRO (sea water reverse osmosis) while VMD is used to reconcentrate SWRO output (Mericq, Laborie, and Cabassud, 2010). VMD coupling gave higher recovery (89%) than NF (70% and 80%). For treatment of dyed water, Bruggen et al. (2001) used NF while Criscuoli et al. (2008) use VMD. NF need pretreatment for its feed since NF flux was depending on the ion concentration of the dye. VMD in other hand do not need pretreatment and the flux was depending on the process condition. Both dye used in the experiment was almost similar in molecular weight, but MD process was depending a little on the molecular weight. Vacuum membrane distillation process condition Heat and mass transfer that happen in VMD process give consideration about the effect of process condition. The performance of VMD mainly focused on its. Temperature and concentration polarization effect resulted flux. Some process conditions that should be regarded are: 1. Feed Temperature Heat transfer across the boundary layers is often the rate limiting step for mass transfer in MD, because such a large quantity of heat must be supplied to the surface of the membrane to vaporize the liquid (Lawson and Lloyd, 1997.). The temperature polarization effect thus needs to be reduced. Al-Asheh et.al (2006) found that when the feed temperature increased, the temperature polarization effect will decreased. In another research, Banat et al. (2003) found that increasing the feed bulk temperature results in an exponential increase in the pure water flux, but the mass flux decreases. 2. Solute concentration From Al Asheh et al. (2006), it was found that the flux of water decreases slightly with increasing sucrose concentration. This is attributed to the fact that the addition of sucrose reduces the partial pressure of water and thereby reduces the driving force of VMD. Khayet and Matsuura (2011) concluded that in general, the total permeate flux increase with increasing of solute concentration in the feed due to the increase of driving force for the solute mass transfer, leading to the enrichment of solvent in the permeate. 3. Feed flow rate To reduce temperature and concentration polarization effects, the feed flow rate must be increased (Khayet and Matsuura, 2011). Wang (2009) showed that permeate flux of VMD increased with increasing of feed flux. Urtiaga et al. (2000) conducted VMD for chloroform removal and the result clearly showed that increasing the low rate of the feed resulted in a faster removal of chloroform from the feed phase thus indicating that the mass transfer is influenced by the diffusion of the organic compound through the liquid boundary layer near the membrane. Criscuoli et al. (2008) also found the same trend with the dye solutions. They mentioned that higher flow rates lead to higher fluxes because of the reduction of the heat transport resistance. Another experiment by Mohammadi and Safavi

7 33 (2009) gave different result on feed flow rate effect. The permeate flux increase until certain value and then decrease. 4. Downstream pressure Khayet and Matsuura (2011) stated that both permeate flux and trans membrane hydrostatic pressure difference increase with the decrease in the downstream pressure. But the risk of membrane pore wetting got higher. Bandini et al. (1997) concluded that the typical operation conditions for VMD consist of permeate pressures always smaller than the saturation pressure in order to prevent condensation, reducing the permeate pressures resulted in an increase in the overall permeate flux, however, there was a correspondingly remarkable decrease in the separation factor of the process. 5. Another process condition Criscuoli et al. (2008) worked with treatment of different dye solutions by VMD. Their result showed that the dye with lowest molecular weight has no highest permeates flux, while the dyes with bigger molecular weight have relatively high permeated flux. It indicated that the permeate flux is strongly related to the chemical properties of dyesand their interaction with the membrane. Based on process conditions described above, the experiment of regeneration of LiBr using VMD will take temperature, feed flow rate and concentration as its parameters. The vacuum pressure on this system is fixed in 725 mbar. From this parameter, by using statistical method and analysis, an optimal value could be achieved for specified system. This optimal value then could be crosschecked to ensure the system reliability. The mass transfer and heat transfer in the VMD process will also be discussed to have more understanding on VMD mechanism in regeneration of LiBr. METHOD The research was conducted from August 2012 untill March 2013 in Laboratory of Heat and Mass Transfer, Department of Mechanical Engineering and Biosystem, Faculty of Agricultural Engineering and Technology, Bogor Agricultural University (IPB). Experiment on VMD membrane is conducted from April untill July 2012 in Membrane Center, Department of Chemical Engineering, CYCU Taiwan. Experiment on Reverse Osmosis (RO) Membrane Figure 3.3 shows absorbtion refrigeration system scheme using reverse osmosis (RO) membrane that was applied in this research. The membrane was used as a tool to separate the refrigerant from absorbent which performs as regeneration process in conventional absorbtion system. As separation component, RO membrane employs work pressure to hold and pass certain component through the pores. RO membrane used in this research was made by Dow Filmtech TW

8 34 Membran RO Absorber 4 Evaporator Figure 3.3 Absorbtion refrigeration system scheme: 1. Pump; 2. RO membrane; 3. Evaporator; 4. Absorber. Experiment test was conducted under three treatments of LiBr-H 2 O concentration i.e. 20, 25 and 30 Brix. Solution was made by mixing LiBr with aquades under mass based (w/w %) until reach desired concentration ( o Brix). The TSS (total soluble solid) content of solution was then measured using refractometer. The solution concentration was calculated using equation (3.1). ( ) (3. 1) where m LiBr is LiBr mass (g), m H2O is H 2 O mass. The first step was regeneration process in absorbtion refrigeration system through separating weak solution to obtain the strong solution as retentatee and water as permeateusing RO membrane. In this experiment, the effect of operation pressure on permeateion flux and R rejection factor of solution that enters the membrane will be examined. Permeateion flux was calculated using equation (3.2). ( ) (3.2) whereas :J w is permeateion flux that pass the membrane; A is permeatepermeability coefficient; Δ is work pressure or different pressure occurs on two sides of membrane; Δπ is solution osmosis pressure on two sides of membrane. Membrane selectivity was expressed by R (rejection factor) as shown in equation (3.3). (3.3) Whereas c p is dissolved concentration in permeate (g.s -1.m -2 ), c f is dissolved concentration in retentatee (g.s -1.m -2 ). The next step was refrigeration process. This step was conducted by placing the permeateinto evaporator and retentatee into absorber. In this experiment, all valves which connecting system equipments were closed, while valve which connecting evaporator and absorber were opened. During this process, data associates with temperature changes in absorber and evaporator, water vapor mass transferred from evaporator into absorber were recorded. The changing magnitude

9 35 of transferred water vapor was measured using digital scale merck AND GF-300 with 0.01 g accuration. Three concentration levels of LiBr-H 2 O solution i.e. 20%, 25% and 30% (w/w) were used in this research. Several equipments were used in this research i.e. 1 unit of RO membrane, pump, valve, pressure gauge, hand refractometer ATAGO NI-K Fuji 13976, digital scale AND GF-300 with 0,01 g accuration, hybrid recorder, thermocoupletype CC. Membrane used in this research was reverse osmosis (RO), Merck of Dow Filmtech TW , spiral-wound module, length 298 mm, outer diameter (d o ) 44,5 mm and inner diameter (d i ) 17 mm. The data obtained from the first and second step were then used to analyze energy and exergy in absorbtion refrigeration system. These analyses were employed to calculate required energy at each component and also to analyze the ineffectiveness processof release and absorbtion energy. Experiment on Vacuum Membrane Distillation (VMD) The experimental device was showed in Figure 3.4. The central part is a commercial shell-and-lumen membrane module UMP-0047R, supplied by Mycroza is trademark of Asahi Kasei Corporation. Basically, it consists of a set of equal polyvinylidene fluoride (PVDF) porous hydrophobic capillaries, assembled and made into a shell-and-lumen module. Hot feed of aqueous lithium bromide solution flows into the lumen side while the shell chamber is keeping in a vacuum state by cooling water or by vacuum pump. Figure 3.4 (a and b), shows the hollow fiber membrane module, in which twenty one of PVDF capillaries are assembled in plastic body and sealed securely to both ends. The principal characteristics of the hollow fiber membrane, as specified by the manufacturer, are showed in Table 3.2 and Table 3.3. Table 3.2 Specification material constructions of membrane module Specifications material contructions : Membrane : Polyvinyldene fluoride (PVDF) Housing : Native clear polysulfone (P) or filled polysulfone (W) Potting Material : Epoxy Resin Gasket : Silicon (P) Bacteriostat : Glycerin 65%, Ethanol 2 %, Water balance Table 3.3 Part number and technical specification of membrane module Part Number and Technical Specification : Part Number : UMP-0047R Number of membrane : 21 Fiber ID/OD (mm) : 1.4/2.2 Nominal Length(mm) : 310 Membrane Area (m 2 ) : 0.02 Clean Water Flux : 19 (Lh -1 bar, 25 C) Crossflow Rate for 1 ms -1 : 116 Fluid velocity Maximum Pressure (barg) : 3/3/1.5 Feed/TMP/Permeate at 50 C Volume Feed : 23 Volume Permeate (ml) : 30

10 36 The temperature and pressure of the liquid feed were measured at the inlet and outlet of the membrane module. The temperature and pressure were measured continuously, in steady state. The temperature of the hot feed was measured at the inlet and outlet of the membrane module, while the temperature of water vapor was measured at the outlet of the permeateion side. The temperatures was measured with thermocouple which connected to a digital multimeter, with an accuracy ± 0.1 C and the pressure was measured with pressure control type GP- D-0040GAPX, with an accuracy of bar. The feed inlet temperature was controlled by a thermostat Water Bath G-20 DENG YNG, with a temperature fluctuation of less than ± 0.1 C. P T T P C P T C 3 Figure 3.4 Experimental VMD set-up: 1. water bath; 2. thermostat water bath; 3. pump; 4. solution flow meter; 5. VMD module; 6. condenser; 7. cooling water flow meter; 8. receiving tank; 9. cold trap; 10.vacuum pump (a) (b) Figure 3.5 PVDF hollow fiber membrane module used in this research

11 37 The liquid feed was circulated with a circulation pump FLOJET MOD: and the feed flow was measured with a flowmeter HJ D- S, withprecision ±2%. The volume of solution in water bath was determined in the beginning of experiment. A water circulating vacuum pump (GAST Model DAA-V503-EB) with a pressure adjuster was connected to the permeateion side of the membrane module to produce the vacuum condition before the test began, and then the vacuum pressure would maintain constant at the condensing temperature of cooling water during the test. A glass cold trap by cooling water was connected to the permeateion side to recover the water vapor. The flux of distilled water was calculated, in every case, by measuring of the permeate flux every ten minutes, during two hours, and by measuring the concentration before and after the process using refractometer index (RI). Permeateion flux means the quality of permeateion water per unit area in an hour. The permeateion flux is described as follow: ( ) (3.4) Where J is the permeateion flux, (kg.m -2.h -1 ); W is the quantity of water, kg; S is the membrane area, m 2 ; and t is the time, h. The operating parameter region of solution concentration, temperature and flow rate was showed in Table 3.4. Volume of the aqueous lithium bromide solution used in the experiment was 5 L, but effective internal membrane area of hollow fiber module was only 0.02 m 2. Table 3-4 shows the operating region and the levels of the variables in actual and coded values. The measured VMD distillate fluxes, the responses and the standard deviations are also presented in Table 4. The experimental design and analysis of data were done by using a commercial statistical package, JMP software version 7. By using the software, a Box Behnken design with 3 center points was employed with three factors and three levels. This design is not randomized. The Box Behnken design contains of 15 experiments with three center points. Table 3.4 Coded and actual designed variables used for experimental design Design variables Coded Actual values of coded levels (factor) Variables Concentration (%) x Temperature ( C) x Flow rate (L/min) x RESULT AND DISCUSSION Performance of Reverse Osmosis Membrane Permeateion Flux and Rejection Factor (R) The increasing of operation pressure in H 2 O separation process from LiBr- H 2 O solution using spiral wound module-based RO membrane will increase the permeateion flux (g.s -1.m -2 ). However, permeateion flux depends on the pressure

12 38 as increasing pressure will increase the flux rate. The increasing of flux rate will increase the feed flux as more particles on membrane surface that can be diffused. That condition is illustrated in Figures 3.6. The increased permeateion flux due to increased operation pressure can be described through Equation 2, whereas the increased of pressure difference at two sides of membrane ( )causes the increased of permeateion flux. Figures 3.6 also provide information on decreased R Rejection factor. Figure 3.6 Relationship among operation pressure, permeate concentration (%) and mass flux (g.s -1.m -2 ) Figure 3.7 Relationship among operation pressure, rejection factor (R) and mass flux (g.s -1.m -2 ) Based on the figure above, it can be seen that operation pressure on membrane decreased rejection factor. Rejection factor at concentration 25% (w/w) and operation pressure 4.6, 5.4 and 6 bar were 0.687, and 0.646,

13 39 respectively. Rejection factor expresses the availability of a membrane to reject certain compound or product from passing the pores. Moreover, based on the mentioned value, it can be gained an information that there were 31.3%-35.4% salt compound that pass the membrane which means that thepermeatestill contained LiBr salt. The Effect of Permeate Flux and Rejection Factor on Evaporator Temperature The experiment showed that operation pressure differences used during separation process affected the permeateion flux and rejection factor. Rejection factor showed that permeatestill contained LiBr component. In other words, the membrane could not reject all salt components. The mass flux increased along with the increased of permeateion flux. The increasing of permeateion flux and the decreasing of rejection factor and enhanced with the high retentatee concentrationwill decreaseevaporator temperature. Figure 3.8 illustrates the temperature changes in evaporator during absorbtion process in absorber. Figure 3.8 Evaporator temperature during absorbtion process at concentration 30%

14 40 Figure 3.9 Evaporator temperature during absorbtion process at concentration 25% Figure 3.10 Evaporator temperature during absorbtion process at concentration 20%. Figures 3.8, 3.9, and 3.10, show that produced concentration from separation process in absorber and impurity of permeate affected the achievable temperature and the length of cooling process. Based on Figure 3.8, it can be described that rejection factor at initial concentration 30% produced water vapor absorbtion for minutes and temperature decreased by C. At concentration 25%, the cooling process time was 20 minutes with average decreased temperature was C. At concentration 20%, the average cooling process was 5 minutes and the temperature that could be achieved was C. Moreover, the concentration amount of LiBr solution in absorber and permeate in evaporator were affected by water vapor amount in absorber.

15 41 Mass changes (Δ mass) of a solution in absorber during certain period of time are shown in Figures 3.11, 3.12 and Based on the figures, it can be seen that the highest water vapor absorbtion by strong solution in absorber occurred during the first five minutes. Along with the increased of time, the absorbing availability of a solution was continue to slower. This condition showed that the water vapor pressure difference in evaporator and absorber continually decreased. The amount of mass water vapor transfer was affected by water vapor pressure as shown in Table 3.4. Table 3.5 Pressure differences in absorber and evaporator Water vapor pressure Consentration (% w/w) (kpa) Initial State Absorber Evaporator Absorber Evaporator Figure 3.11Mass solution changes at C=30%

16 42 Figure 3.12 Mass solution changes at C=25%. Figure 3.13 Mass solution changes at C=20% The Optimum Operation of Vacuum Membrane Distillation (VMD) The response surface model (RS-model) with interaction terms was developed for the VMD distillate flux using equation(3.5) and the experimental data was summarized in the Table 3.6. (3.5)

17 43 Where y is the predicted response, x i is the coded variables, and b 0, b i, b ii, b ij are the regression coefficient. The values of the regression coefficients were determined using the ordinary least square method written as follows: ( ) (3.6) Where b is a column vector of the regression coefficients, x is the design matrix of the coded levels of the input variables and y is column vector of the response. Table 3.6 Box-Behnken design and experimental VMD No. Run Factor (Controllable input variable) Response Concentration Temperature Flow rate J (%) ( C) (L.min -1 ) w x 1 C x 2 T x 3 m (kg.m -2.s -1 ) The obtained RS-model is written in equation (3.7), as a function of the coded variables and permits to predict the distillate flux, J w (kg/m 2.s), as follows: (3.7) It should be pointed out that the RS-model includes only the significant term. The significance of the regression coefficients is expressed in equation 3.7. Figure 3.14 presents the plot of experimental and predicted value. Moreover, the statistical validation of the RS-model was performed by means of analysis of variance (ANOVA). The results were presented in Table 3.7.

18 44 Table 3.7 Analysis of variance (ANOVA) of the RSM model corresponding to the response: performance index (Y). Source DF 1 SS 2 MS 3 F Ratio R 2 R 2 adj Model Error C. Total Degree of freedom 2 Sum of squares 3 Mean of square Figure 3.14 Comparison between the experimental and predicted VMD performance index (Y) determined by the RSM model. The mathematical equation used to calculate the ANOVA estimators(i.e. SS, MS, F-value, R 2, R 2 adj) are detailed elsewhere. As can be seen in Table 3.7, the F- value is high, P- value is smaller than and R 2 value is about in agreement with the adjusted coefficient of determination R 2 adj. These indicate that the RS-model in equation(3.7) is statically valid and can be used for prediction of the VMD distillate flux. The significance of the regression coefficient of the models written as function of the coded variables could be tested with statistical Student t test. Table 3.8 shows the t test result for the experiment. From the test results, it shows that the feed initial temperature (x 2 ), concentration (x 1 ) and flow rate (x 3 ) are significance. Furthermore, its interaction effect with the feed initial concentration and temperature, concentration and feed flow, temperature and feed flow are still significant.

19 45 Table 3.8 Parameter estimates and t-test results Term Estimate Std Error t Ratio Prob> t x 2 ( C) <0.0001* x 2 ( C) * x 2 ( C) * x 2 ( C) * x 3 (L/min) * x 1 (%)(45,50) * x 3 (L/min)(1.1,1.9) * x 3 (L/min)* x 3 (L/min) * x 1 (%)* x 2 ( C) * x 1 (%)* x 3 (L/min) * x 1 (%)* x 1 (%) The response surface for the given factors and response is a saddle point one, which meant for the range given, there is no optimal value. This saddle point parameter is at the outside of the data range in which concentration is set at %, feed temperature at C and flow rate at L.min -1 while its predicted value is g.m -2 s -1. To have a more understanding on the effect of each parameter and their interaction, a prediction profile of the experiment is shownin Figure As can be seen, the feed inlet temperature shows an obvious increasing curvature since its significance. The initial feed concentration shows a slight decreasing curvature which shows its effect on the response. The feed flow shows a slight increasing curvature which shows its effect on the response. Figure 3.15 Prediction profiler of experiment result The effect of the VMD operating parameters on the distillate flux are plotted in Figures 3.16, 3.17 and 3.18 in 3-D and 2-D contour plots. For example, Figure 3-13 shows the effect of the feed inlet temperature (x 2 ) and initial feed concentration (x 1 ) on the permeate flux. It can be seen, the increase of initial feed concentration will leads to a decrease of the permeate flux while the increment of feed inlet temperature will leads to an increase of the response. At high

20 46 concentration, the effect of temperature is stronger than at low concentration. For example, at concentration between 45%-50%, temperature 60 C and feed flow 1.5 L.min -1 will produce permeate flux from g.m -2.s -1 to g.m -2.s -1. In contrast, at concentration 45 % and temperature between 60 C-80 C will produce permeate flux from g.m -2.s -1 to g.m -2.s -1.The graph also confirms the strong effect of the feed inlet temperature over the initial feed concentration. Figure 3.16 Response surface plot and contour profiler showing the VMD performance index (Y) as a function of Temperature ( C) and Concentration (%) for flow rate: 1.5 (L.min -1 ) Figure 3.17 Response surface plot and contour profile showing the VMD performance index (Y) as a function of flow rate (L.min -1 ) and Concentration (%) for temperature: 70 ( C).

21 47 Figure 3.17 shows the effect of flow rate (x 3 ) (L.min -1 ) and concentration (x 1 ) (%) on the response Y. The main effect of flow rate (x 3 ) (L.min -1 ) is greater than the main effect of concentration (x 1 ) %. The increase of both variable leads to the enhancement of Y. As it is mentioned before, due to the quadratic effect of x 1, the response Y increases up to a maximum and then decreases for high x 1 values. Interaction effect of x 3 and x 1 on the response Y is significant, although these interactions give negative values. Figure 3.18 Response surface plot and contour profile showing the VMD performance index (Y) as a function of flow rate (L.min -1 ) and temperature ( C) for concentration: 47.5 (%). Figure 3.18, shows a strong main effect of temperature (x 2 ) ( C) on the response Y. The increment of this variable causes an increase of Y and the main effect of flow rate (x 3 ) (L.min -1 ) is smaller compared to the main effect of temperature. However, the quadratic effect of x 2 is greater than the quadratic effect of x 3,and the interaction effect between temperature and flow rate is significant and gives a positive value. Conclusions According to those aforementioned results, it can be drawn several conclusions as listed below: 1. The optimum water vapor absorbtion in absorber using spiral wound modulebased RO membrane occurred at initial concentration 30%. 2. The increasing of operation pressure during separation process will increase permeateion flux but will decrease rejection factor. 3. High permeate concentration will affect the achievable cooling temperature and cooling period. 4. Higher concentration in retentatee will increase water vapor absorbtion rate.

22 48 5. Higher concentration in permeate will increase irreversibility value in evaporator. Conversely, in absorber, higher concentration in permeate will reduce irreversibility value in absorber. 6. Design of experiments and response surface methodology (RSM) were applied for desalination by VMD. H 2 O separation process from LiBr-H 2 O solution using VMD show that at concentration 47.5%, temperature 80 C and flow rate 1.9 L.min -1 gave the highest response permeate flux was (kg.m -2.s - 1 ).10-3.