Citation
Physical Mechanisms of Heat and Mass Transfer in Nanoengineered Membrane-Based Absorption Cooling System

Material Information

Title:
Physical Mechanisms of Heat and Mass Transfer in Nanoengineered Membrane-Based Absorption Cooling System
Creator:
Nasr Isfahani, Rasool
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (140 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Mechanical Engineering
Mechanical and Aerospace Engineering
Committee Chair:
MOGHADDAM,SAEED
Committee Co-Chair:
CHUNG,JACOB NAN-CHU
Committee Members:
SHERIF,SHERIF AHMED
YOON,YONG KYU
Graduation Date:
8/9/2014

Subjects

Subjects / Keywords:
Cooling ( jstor )
Desorption ( jstor )
Flow velocity ( jstor )
Heat exchangers ( jstor )
Inlet temperature ( jstor )
Microchannels ( jstor )
Pressure ( jstor )
Vapors ( jstor )
Water temperature ( jstor )
Water vapor ( jstor )
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
absorption -- desorption -- libr -- membrane -- nanofibrous
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Mechanical Engineering thesis, Ph.D.

Notes

Abstract:
Physics of water vapor absorption into and desorption from a lithium bromide (LiBr) solution flow through an array of microchannels capped by a porous membrane are studied. The membrane allows the vapor to exit the flow while it retains the liquid. As opposed to conventional falling film heat exchangers, in this configuration, the solution film thickness and velocity can be controlled independently. Parametric studies are conducted on both absorber and desorber processes. An absorption rate of approximately 0.0044 kg/m2s was measured at a LiBr solution channel thickness of 100 micron, a flow velocity of 5 mm/s, and a pressure potential of 600 Pa. It is demonstrated that decreasing the solution film thickness and increasing the solution velocity enhance the absorption rate. Next, micro surface structures are employed in microchannels to manipulate thermohydraulic characteristics of the Lithium Bromide (LiBr) solution flow. With micro structures, absorption rates as high as that of a 100 micron solution film is achieved into 500 micron thick solution microchannels, while the solution pressure drop is about two orders of magnitude lower. The absorption performance of ionic liquid [EMIM][MeSO3] as an alternative desiccant is also tested in the absorber with microstructures. An absorption rate of approximately 0.001 kg/m2s was measured at a flow velocity and pressure potential of 8 mm/s and 600 Pa, respectively. This absorption rate is about 4 times lower than that of LiBr solution in the absorber with similar surface microstructures. Further studies and new absorber designs are necessary to enhance the absorption rate In desorption studies two different mechanisms of desorption are analyzed. These mechanisms consisted of: (1) direct diffusion of water molecules out of the solution and their subsequent flow through the membrane and (2) formation of water vapor bubbles within the solution and their venting through the membrane. Direct diffusion is the dominant desorption mode at low surface temperatures and its magnitude is directly related to the vapor pressure, the solution concentration, and the heated wall temperature. Desorption at the boiling regime is predominantly controlled by the solution flow pressure and mass flux. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2014.
Local:
Adviser: MOGHADDAM,SAEED.
Local:
Co-adviser: CHUNG,JACOB NAN-CHU.
Statement of Responsibility:
by Rasool Nasr Isfahani.

Record Information

Source Institution:
UFRGP
Rights Management:
Copyright Nasr Isfahani, Rasool. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
969976735 ( OCLC )
Classification:
LD1780 2014 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

1 P HYSICAL MECHANISMS OF HEAT AND MASS TRA NSFER IN NANOENGINEERED MEMBRANEBASED ABSORPTION COOLING SYSTEM By RASOOL NASR ISFAHANI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFI LLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014

PAGE 2

2 2014 Rasool Nasr Isfahani

PAGE 3

3 To my mom, dad and brother

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my advisor, Professor Saeed Moghaddam, for giving me the opportunity to work in his group and for supporting my research. He was always available and provided insightful advice throughout my education. He always gave encouragement when I faced difficult ies in my research. He taught me that there is almost no problem without a solution and I have to just be patient and hardworking. I will always appreciate his wisdom and helpfulness . I thank my supervisory committee, Dr . Chung, Dr . Sherif and Dr. Yoon for their g uidance throughout my years of study. Without their support and advice, my work would not be in the current shape. I would like to mention all my colleagues : Abdy Fazeli, Sajjad Bigham, Mehdi Mortazavi, Devesh Chugh, Abhilash Paneri , Sai Tej, Han Lai, Qanit Takmeel, David Horner and Karthikeyan Sampath in Nanostructure Energy System (NES) lab for their enormous support . I always enjoyed discussing scientific matters with them which greatly enhanced the quality of my work and I learned a lot from t hem. Abdy Fazeli was always ready to help me in any matter . Sajjad Bigham was nice enough to help me with numerical simulations. Abhilash Paneri measured ionic liquid properties for my tests . Mehdi Mortazavi greatly helped me in my experiments. Karthikeyan Sampath helped me in testing the membranes used in my experimental setup. I had very useful scientific discussions with Devesh Chugh and he helped me a lot. I would like to mention all the undergrad and master students that worked with me and helped me a lot in my experiments: Andrew Piotti, Xing Wei and Zhiyuan Du.

PAGE 5

5 I thank my friends for their support during the last four years. I truly apprec iate the friendship of my dear friends: Nima Rahmatian, Guita Banan, Mohammad Malakooti, Ayyoub Mehdizadeh and Sh ahin Navardi . Last, but not the least, I express gratitude to my parents and my brother for their encouragement and support during all stages of my life. I thank my dad for being a kind father who always supports me without any hesitation. I especially th ank my mom for always caring for me and doing all she can for me. Without my family, I could not achieve anything in my life.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 8 LIST OF FIGURES .......................................................................................................... 9 LIST OF ABBREVIATIONS ........................................................................................... 14 NOMENCLATURE ........................................................................................................ 15 ABSTRACT ................................................................................................................... 16 CHAPTER 1 INTRODUCTION .................................................................................................... 18 Absorption Refrigeration Systems (ARSs) .............................................................. 18 Absorber Heat Exchanger ....................................................................................... 20 Conventional Absorbers ................................................................................... 20 Membranebased Absorbers ............................................................................ 22 Membranebased Absorber with Micromixing .................................................. 24 Desorber Heat Exchanger ...................................................................................... 26 Conventional Des orbers ................................................................................... 26 Membranebased Desorbers ............................................................................ 28 Ionic Liquids ............................................................................................................ 30 2 EXPERIM ENTAL SETUPS ..................................................................................... 34 Absorption and Desorption Test Setup ................................................................... 34 Membranebased Absorber Heat Exchanger ................................................... 37 Membranebased Absorber Heat Exchanger with Ridges ................................ 42 Membranebased Desorber Heat Exchanger ................................................... 44 Membranebased Evaporator Heat Exchanger ................................................ 48 Tube Heat Exchangers and Condenser ........................................................... 50 Vacuum Leak Tests .......................................................................................... 51 Brazing ............................................................................................................. 52 Insulation .......................................................................................................... 53 Charging the Loop ............................................................................................ 54 Working Fluids’ Properties ................................................................................ 57 Experimental Procedure ................................................................................... 62 Cycle Analysis and Test Parameters ................................................................ 62 Experimental Uncertainty ................................................................................. 74 Data Reduction ................................................................................................. 75 Membrane Test Setup ............................................................................................ 76

PAGE 7

7 Visualization Test setup .......................................................................................... 78 3 MEMBRANE EVALUATION ................................................................................... 80 Membrane Transpor t .............................................................................................. 80 Membrane Permeability .......................................................................................... 81 4 RESULTS AND DISCUSSIONS ............................................................................. 84 Absor ption Results .................................................................................................. 84 Thin Film Absorption ........................................................................................ 84 Effect of cooling water temperature ........................................................... 87 Effect of solution inlet temperature ............................................................. 92 Effect of solution velocity ........................................................................... 93 Absorption with Micromixing ............................................................................. 97 Absorption with Ionic Liquid [EMIM][MeSO3] .................................................. 105 Desorption Results ............................................................................................... 107 Thin Film Desorption ...................................................................................... 107 Effect of solution pressure........................................................................ 112 Effect of solution velocity ......................................................................... 117 Effect of solution inlet temperature ........................................................... 119 Analysis of Bubbles Discharge ....................................................................... 120 5 CONCLUSION ...................................................................................................... 129 LIST OF REFERENCES ............................................................................................. 132 BIOGRAPHICAL SKETCH .......................................................................................... 140

PAGE 8

8 LIST OF TABLES Table page 1 1 Comparison of absorption rate in conventional absorbers .................................. 22 2 1 Working fluid properties at room temperature (i.e. 25C) ................................... 58 2 2 Model inputs for the baseline case ..................................................................... 65 2 3 Cycle COP and heat transfer loads of each component of a 1 Ton system for the baseline case ................................................................................................ 66 2 4 Components’ conditions for the baseline case ................................................... 67 2 5 I nput values for parametric studies on thin film absorption ................................. 72 2 6 I nput values for parametric studi es on absorption with micromixing ................... 73 2 7 Input values for parametric studies on absorption with IL [EMIM][MeSO3] ......... 73 2 8 Input values for parametric studies on thin film desorption ................................. 74 2 9 Variable uncert ainties ......................................................................................... 75 4 1 Parameters of the numerical simulation ............................................................. 91

PAGE 9

9 LIST OF FIGURES Figure page 1 1 A schematic of a singleeffect LiBr water absorption refrigeration system ......... 20 1 2 A schematic depicting a falling LiBr solution over a horizontal tube ................... 21 1 3 Structure of membrane based absorber, (a) A schematic of a constrained thin film absorption process; (b) A design of a membranebased absorber (membrane is not shown on the water LiBr flow channels) ................................ 23 1 4 A 3D schematic of surface features at the floor of a microchannel (microchannel is not shown in the picture) and a induced fluid particle streamline ........................................................................................................... 26 1 5 Schematics showing desorption from a falling LiBr water solution film through (a) direct diffusion and (b) boiling mechanisms .................................................. 27 1 6 Schematic representation of membranebased desorption process showing (a) direct diffusion of water from the solution flow and (b) bubble formation and desorption through the membr ane during the boiling regime. ...................... 29 2 1 A Schematic diagram of the membranebased absorber and desorber test setup. .................................................................................................................. 35 2 2 A photograph of the assembled experimental setup ........................................... 37 2 3 A schematic representation of the absorber heat exchanger cross section. ....... 38 2 4 A photograph of the main absorber part (solution side). ..................................... 38 2 5 A photograph of the stainless steel perforated sheet. ......................................... 39 2 6 A p hotograph of the stainless steel frame. ......................................................... 39 2 7 A photograph of the membrane assembly fixture attached to the main absorber part. ..................................................................................................... 39 2 8 A photograph of the main absorber part (Cooling water side). ........................... 40 2 9 A photograph of the brass cover (water side). .................................................... 41 2 1 0 A photograph of the brass cover (solution side). ................................................ 42 2 11 A photograph of the assembled absorber heat exchanger. ................................ 42 2 12 A phot ograph of the absorber heat exchanger ................................................... 43

PAGE 10

10 2 13 Microchannels and micro structures dimensions ................................................ 43 2 14 A photograph of the absorber heat exchanger with the installed membrane ...... 44 2 15 Schematic diagram of the desorber heat exchanger. ......................................... 45 2 16 A photograph of the desorber solution microchannels. ....................................... 45 2 17 A photograph of the membrane assembly fixture attached to the main desorber part. ..................................................................................................... 46 2 18 A photograph of the installed thermocouples and flexible heater on the back side of the desorber ............................................................................................ 47 2 19 A photograph of the assembled desorber heat exchanger. ................................ 47 2 20 A front view of a thin film evaporator. ................................................................. 48 2 21 A photograph of the membrane assembly fixture attached to the thin film evaporator. ......................................................................................................... 49 2 22 A photograph of the assembled thin film evaporator. ......................................... 49 2 23 A photograph of the assembled tube heat exchanger. ....................................... 51 2 24 A temperature controller used with tube heat exchangers. ................................. 51 2 25 A photograph of the brazed joints on a Hastelloy part. ....................................... 53 2 26 Photographs of an absorber before and after insulation. .................................... 53 2 27 Photographs of fully insulated components. ....................................................... 54 2 28 A photograph of the fabricated charging components. ....................................... 55 2 29 A photograph of a lithium bromide solution reservoir. ......................................... 56 2 30 A photograph of a vacuum distillation set up. ..................................................... 57 2 31 Water vapor pressure as a function of temperature ............................................ 59 2 32 Vapor pressure as a function of temperature ..................................................... 60 2 33 Absorbent dynamic viscosity as a function of temperature ................................. 61 2 34 Schematic diagram of a doubleeffect absorption system .................................. 65 2 35 Effect of solution mass flow rate on cycle COP and refrigeration capacity ......... 68

PAGE 11

11 2 36 Effect of absorber cooling water temperature on cycle COP and refrigeration capacity .............................................................................................................. 69 2 37 Effect of chilled water inlet temperature on cycle COP and refrigeration capaci ty .............................................................................................................. 70 2 38 Variation of evaporator vapor pressure with chilled water inlet temperature ...... 71 2 39 Effect of heating medium inlet temperature on cycle COP and refrigeration capacity .............................................................................................................. 72 2 40 A photograph of the test setup for membrane permeability measurements. ...... 77 2 41 Cross sectional view of the test fixture. .............................................................. 77 2 42 A 3D schematic of the visualization test device. ................................................. 79 2 43 A diagram of the visualization test system. ......................................................... 79 3 1 SEM Micrographs of nanofibrous PTFE membrane with different nominal .......................................................................... 81 3 2 Pressure drop versus flow rate at 0.85 kPa and 10 kPa test pressures. ............ 82 4 1 Variation of absorption rate as a function of water vapor pressure at hr kg ma sol/ 6 . 0 and C Ta cw 25 . ........................................................................ 85 4 2 Schematics of the solution flow channel cross section (a) depicting heat release at the solution vapor interface and (b, c) illustrating the impact of change in solution flow channel thickness, at a constant flow rate, on the d). ...................................................... 86 4 3 Solution water vapor pressure as a function of temperature and concentrati on. ..................................................................................................... 86 4 4 Variation of absorption rate as a function of cooling water temperature at hr kg ma sol/ 6 . 0 and kPa Pa v1 . 1 . ....................................................................... 87 4 5 Variation of absorption rate as a function of water vapor pressure potential. ..... 89 4 6 Comparison between membrane and solution resistances at a solution film thickness of 160 ........................................................................................... 90 4 7 Temperature contours in a 4mm thick solution film at different pressure potentials. ........................................................................................................... 92 4 8 Absorption rate in a 4mmthick solution film as a function of pressure potential. ............................................................................................................. 92

PAGE 12

12 4 9 Variation of absorption rate as a function of solution inlet temperature at a hr kg ma sol/ 6 . 0 , C Ta cw 25 and kPa Pa v1 . 1 . ....................................................................................................... 93 4 10 Variation of absorption rate as a function of solution flow rate at a solution C Ta cw 25 and kPa Pa v1 . 1 . ...................................... 95 4 11 Variation of absorption coefficient as a function of solution flow rate. ................. 96 4 12 S olution pressure drop as a function of flow rate ................................................ 97 4 13 Variation of absorption rate as a function of water vapor pressure at hr kg ma sol/ 5 . 2 and C Ta cw 25 . ........................................................................ 98 4 14 Variation of absorption rate as a function of cooling water temperature at hr kg ma sol/ 5 . 2 and kPa Pa v1 . 1 . ....................................................................... 99 4 15 Variation of absorption rate as a function of water vapor pressure potential .... 101 4 16 LiBr concentration contours at different cross sections between x=0 to 100 mm (the white color in the pictures shows the cross section of ridges). ........... 102 4 17 Solution pressure drop as a function of flow rate .............................................. 103 4 18 Comparison betw een membrane and solution resistances .............................. 104 4 19 Variation of absorption rate as a function of solution inlet temperature at hr kg ma sol/ 5 . 2 , C Ta cw 25 and kPa Pa v1 . 1 .................................................. 105 4 20 Variation of absorption rate as a function of water vapor pressure hr kg ma sol/ 63 . 0 and C Ta cw 25 ...................................................................... 106 4 21 Performance of ionic liquid [EMIM][MeSO3] compared to that of LiBr solution 107 4 22 Effect of heated wall temperature on desorption rate at different vapor pressures at hr kg md sol/ 5 . 2 and kPa Pd s23 . ................................................. 108 4 23 Comparison of numerical results showing a lower concentration solution (or a higher solution water vapor pressure) in case B compared t o case A. A scale factor of 0.013 is used in the x direction t o show the concentration contours over the entire flow domain (a). Water pressures used in (b) are determined liquid interface. ................................... 112 4 24 Effect of soluti on pressure on desorption rate at hr kg md sol/ 5 . 2 and kPa Pd v10 . ...................................................................................................... 113

PAGE 13

13 4 25 Effect of solution pressure on the desorption rate in a boiling regime at hr kg md sol/ 5 . 2 and kPa Pd v10 . ...................................................................... 115 4 26 Comparison of the measured desorption rate with those of other studies. ....... 116 4 27 Schematic of the desorber configuration implemented in a prior membranebased study . ..................................................................................................... 117 4 28 Effect of solution flow rate on desorption rate at kPa Pd s23 and kPa Pd v10 . 118 4 29 Average concentration at different solution flow rate tests versus wall temperature. ..................................................................................................... 119 4 30 Effect of solution inlet tem perature to the desorber on the desorption rate at kPa Pd v10 , hr kg md sol/ 5 . 2 and kPa Pd s23 . ................................................. 120 4 31 Desorber heat loss as a function of its wall temperatur e. ................................. 121 4 32 Comparison of desorber heat loss and solution sensible heat to desorber input heat. ......................................................................................................... 122 4 33 Comparison of vapor des orption and generation rates at nominal conditions. . 123 4 34 Comparison of the vapor desorption and generation rates at different solution pressures. ......................................................................................................... 124 4 35 Comparison of vapor desorption and generation rates at different solution flow rates. ......................................................................................................... 125 4 36 Bubble extraction through the membrane with 13 kPa pressure differ ence across the membrane at three fluxes: (a) wm =12 kg/m2s (b) wm =41 kg/m2s (c d) wm =54 kg/m2s. ......................................................................................... 127

PAGE 14

14 LIST OF ABBREVIATIONS ARS Absorption refrigeration system CCHP C ombined cooling, heating, and power COP Coefficient of performance DLS D irec t laser scattering HX Heat exchanger IL Ionic liquid LBM Lattice Boltzmann method LIBR Lithium bromide solution PDMS P olydimethylsiloxane PTFE Polytetrafluoroethyl ene SEM Scanning electron microscope TEC Thermoelectric cooler VCS Vapor compression system

PAGE 15

15 NOME N CLATURE absm A bsorption rate ( kg / m2 s ) w s,P S olution water vapor pressure (Pa or kPa ) desm D esorption rate ( kg / m2 s ) R U niversal gas constant (J / mol K) pd M embrane mean pore diameters (m) T T emperature (C) mK A bsorption coefficient (kg/ m2 s Pa) cwT Co oling water inlet temperature (C ) solm S olution flow rate (kg / hr) supT W all superheat temperature (C) M M olecular weight (kg / mol) M embrane porosity N M olar flow rate (mol / s) P P ressure drop (Pa or kPa) P P ressure (Pa or kPa ) d C oncentration boundary layer thickness (m) V vapor pressure (Pa or kPa ) V F low velocity (m/s) X Concentration ( weight fraction % ) hD C hannel hydraulic diameter (m) D ensity (kg m3) m M ass flux ( kg/m2 s) V V oltage (volts) dC D rag coefficient h E nth alpy (J/kg /K) or heat transfer C oefficient (W/m2/K) or solution film T hickness (m) Subscripts A Area (m2) in I nlet I C urrent (amps) out O utlet lossQ D esorber heat loss (W) w W all desQ T otal heat input to the desorber (W) v V apor sensibleQ S olution sensible heat in desorber (W) i I nterface or substance i J P ermeate flux ( mol /m2 /s ) sol S olution L C oefficient of proportionally ( 1/ J m s) s S olution or l iquid C hemical potential (J/mol) sup S uperheat x P osition along the membrane c M olar concentration (mol/m3) Superscripts A ctivity coefficient v M olar volume a A bsorber GP P ermeability coefficient d D esorber vm V apor generation rate in the desorber ( kg / m 2 s ) o A t reference condition sP S olution pressure ( Pa o r k Pa) k t hermal conductivity (W/m/K)

PAGE 16

16 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy P HYSICAL MECHANISMS OF HEAT AND MASS TRANS FER IN NANOENGINEERED MEMBRANEBASED ABSORPTION COOLING SYSTEM By Rasool Nasr Isfahani August 2014 Chair: Saeed Moghaddam Major: Mechanical Engineering P hysics of water vapor absorption into and desorption from a lithium bromide (LiBr) solution flow through an array of microchannels capped by a porous membrane are studied . The membrane allows the vapor to exit the flow while it retains the liquid. As opposed to conventional falling film heat exchangers , in this configuration, the solution film thickness and velocity can be controlled independently. Parametric studies are conducted on both absorber and desorber processes . An absorption rate of approximately 0.0044 kg/m2s was measured at a LiBr solution channel thickness of 100 m, a flow velocity of 5 mm/s , and a pressure potential of 600 Pa. It is demonstrated that decreasing the solution film thickness and increasing the solution velocity enhance the absorption rate. Next, micro surface structures are employed in mi crochannels to manipulate thermohydraulic characteristics of the Lithium Bromide (LiBr) solution flow . With micro structures, a bsorption rates as high as that of a 100 micron solution film is achieved into 5 00 m thick solution microchannels, while the sol ution pressure drop is about two orders of magnitude lower .

PAGE 17

17 The absorption performance of ionic l iquid [EMIM][MeSO3] as an alternative desiccant is also tested in the absorber with microstructures. An absorption rate of approximately 0.001 kg/m2s was measured at a flow velocity and pressure potential of 8 mm/s and 600 Pa, respectively. This absorption rate is about 4 times lower than that of LiBr solution in the absorber with similar surface microstructures. Further studies and new absorber designs are necessary to enhance the absorption rate In desorption studies t wo different mechanisms of desorption are analyzed. These mechanisms consisted of: (1) direct diffusion of water molecules out of the solution and their subsequent flow through the membrane and (2) formation of water vapor bubbles within the solution and their venting through the membrane. Direct diffusion is the dominant desorption mode at low surface temperatures and its magnitude is directly related to the vapor pressure, the solution concentrat ion, and the heated wall temperature. Desorption at the boiling regime is predominantly controlled by the solution flow pressure and mass flux.

PAGE 18

18 CHAPTER 1 INTRODUCTION Absorption Refrigeration Systems (ARSs) Absorption refrigeration systems (ARSs) dominated the refrigeration industry for much of the 19th century. Increased access to electricity in the later part of that century triggered the gradual replacement of the ARSs with vapor compression systems (VCSs). The higher performance per unit cost, lower vo lume per unit cooling capacity, and favorable operational and maintenance characteristics of the VCSs fueled their vast market penetration, particularly in the residential air conditioning sector. Despite their great advantages over ARSs, VCSs consume significant electrical energy and use refrigerants that are not environment friendly. A significant increase in the demand for air conditioning in developing countries, the rise in fuel costs, and the environmental impacts of power production cycles have raised concerns about the longterm sustainability of the standard of living this technology has offered to the world population. Thus, development of alternative, more efficient technologies with less environmental impact is of significant interest. ARSs coul d play a larger role in the future cooling market if compact, inexpensive, high performance, and robust systems are developed. Such systems are particularly attractive in combined heating, cooling, and power (CCHP) systems in which the ARS is powered by waste heat. Recent advancements in solar thermal collectors have also enhanced the prospect of solar cooling using ARSs. Hybrid systems powered by solar energy and natural gas could conceivably provide cooling, space heating, and hot water to a building. Implementation of such a system could greatly reduce the electrical load during the peak demand for air conditioning.

PAGE 19

19 Of the different heat powered cooling systems (adsorption [1,2] , ammoniawater [2,3] , ejector refr igeration [4] and LiBr water [3,5] ), LiBr water systems deliver the highest performance using low quality heat (100 to 200 C) [2,6] . A doubleeffect LiBr water system can deliver a primary COP of about 1.21.3 [2,7] . T he primary COP factors in a multiplier for the primary energy consumption (a multiplier of 3.18 is used in the US corresponding to an average power production efficiency of 31.4%) [8] . Despite their great performance, LiBr water systems ar e not economical at small scales [9] . LiBr water ARSs consist of large heat exchangers responsible for their high cost and bulkiness. To build compact and inexpensive systems, alternative heat exchanger configurations and system architectures have been studied [10– 21] . The size of heat exchangers involved in absorption and desorption of water is impacted by the limited water mass transfer coefficient in the LiBr solution. Enhancement of the absorption and desorption transport processes and introduction of compact heat exchanger architectures facilitate development of small scale systems. Fig. 11 shows a schematic diagram of a LiBr water ARS. The concentrated LiBr solution entering the absorber absorbs the water (refrigerant) vapor generated in the system evaporator (i.e. the absorption process). The diluted LiBr solution is pumped to the generator (i.e. desorber) in which water is desorbed from the solution by adding heat (i.e. the desorption process). The water vapor passes into the condenser while the concentrated LiBr solution returns to the absorber. The condensed water after the condenser enters evaporator where it is vaporized and fed back to the absorber. The heat that liquid water absorbs in the evaporator to turn into vapor is the refrigeration product of the cycle.

PAGE 20

20 Figure 11 . A schematic of a singleeffect LiBr water absorption refrigeration system Absorber Heat Exchanger Conventional Absorbers A bsorber heat exchanger is widely recognized as a component that greatly impacts the system size, cost, and performance. In a conventional shell andtube absorber, the water vapor generated in the evaporator is absorbed into the LiBr solution sprayed over a tube bundle. The tube bundle removes the heat of absorption from the LiBr solution. Hydrodynamics of the falling films over the tubes dictates the formation of thick solution layers that impede heat and mass transfer. A horizontal tube (cf. Fig. 1 2 ) absorber is the most common design in the conventional systems. Condenser Evaporator Generator Absorber Expansion Valve Solution Pump

PAGE 21

21 Figure 12 . A schematic depicting a falling LiBr solution over a horizontal tube A number of numerical and experimental studies have been conducted to determine the heat and mass transfer performance during the absorption process in horizontal shell andtube [13,21– 25] and vertical wall or tube [26– 30] absorbers. Table 1 1 compares vapor absorption rates for different types of absorbers. In addition to the absorber geometry, the cooling water temperature, and the water vapor pressure are listed in the table. The reported data suggest a rather large discrepancy in the absorption rate at different conditions. However, more data points seem to suggest an absorption rate of around 0.0024 kg/m2.s at a vapor pressure of about 1.3 kPa and an inlet concentration of around 60%. Table 11 also lists test results on two proposed methods for enhancin g the absorption rate. These methods are solution film inversion on horizontal tubes [10] and the use of helical coil s [11] . It has been argued [10] that through assembling a wire mesh between the tubes of neighboring columns in a bundle, they could achieve periodic reversal of the solution film’s exposed surface. The exposed side of the solution film is more concentrated thus it absorbs water vapor at a higher rate. An absorption rate of 0.004 kg/m2s was measured at a vapor pressure of 2.08 kPa. However, the test was conducted at a water vapor driving potential approximately two times higher than that in a typical absorber (the average water vapor pressure in the Cooling fluid LiBr solution Water vapor

PAGE 22

22 solution is 400500 Pa). Using a helical coil [1 1] did not result in any increase in the absorption rate per surface area of the tubes. However, the authors argued that more surface area per volume could be packed into an absorber through implementation of the helical coil configuration. H. T. Horizontal Tube V. T.: Vertical Tube V. Wall: Vertical Wall Hel. T.: Helical Tube Membranebased Absorbers The present work is an att empt to study performance of a membranebased absorber. Fig. 1 3 a illustrates the basics of the process. In this approach, a thin LiBr solution film is mechanically constrained by a porous membrane on one side and a cooling surface on the other side. The m embrane is superhydrophobic and prevents the solution from entering into the membrane. The water vapor molecules pass through the membrane and diffuse into the LiBr solution. The implementation of a thin solution film Table 11. Comparison of absorption rate in conventional absorbers Reference Absorption rate (kg/m2s) Cooling water temp (C) Vapor pressure (kPa) Study Geometry Choudhury et al. [22] 0.0026 30 1.227 Num. H.T. Jeong et al. [13] 0.0023 32 1.387 Num. H. T. Sultana et al. [23] 0.0022 27 2.080 Num. H.T. Islam et al. [24] 0.0022 27 2.300 Exp. H. T. Yoon et al. [31] 0.0025 32 0.933 Exp. H. T. Kiyota et al. [25] 0.0015 30 0.666 Exp. H. T. Miller et al. [26] 0.0022 35 1.300 Exp. V. T. Matsuda et al. [32] 0.0027 1.300 Exp. V. T. Bo et al. [27] 0.00125 30 1.000 Num. V. Wall Patnaik et al. [28] 0.0023 29.4 1.200 Num. V. T. Medrano et al. [29] 0.0020 30 1.000 Exp. V. T. Karami et al. [30] 0.0022 32 1.000 Num. V. Wall Islam et al. [10] 0.0041 27 2.080 Exp. H.T. (Film Inversion) Yoon et al. [11] 0.0021 30 0.931 Exp. Hel. T.

PAGE 23

23 reduces the heat and mass transfer res istance. In addition, this configuration enables the fabrication of more compact absorbers (cf. Fig. 1 3 b). Figure 13 . Structure of membrane based absorber, (a) A schematic of a constrained thin film absorption pr ocess; (b) A design of a membranebased absorber (membrane is not shown on the water LiBr flow channels) Prior work on membranebased absorption is very limited. Ali and Schwerdt [16] and Ali [17] conducted studies on membranebased absorption process. They used a solution film thickness of 4 mm constrained by a membrane. They achieved an absorption rate approximately half that of the conventional absorbers (i.e. 0.0012 kg/m2s) at a differential water vapor pressure of more than twi ce the available pressure in a typical absorber. It should be noted that the reported thickness of the LiBr solution film over a tube bank in a conventional absorber varies from 0.1 to 1.0 mm [33] . Ali and Schwerdt [16] speculated that a significant mass transfer resistance through the membrane resulted in poor absorption rates. Their study does not provide details regardi ng the membrane structure nor does it present any data on measurement of the membrane pressure drop independent of the absorber. (b) (a) Saturated Water V apor Water LiBr S olution Heat of Absorption Membrane Water LiBr solution Cooling Fluid Water -LiBr solution

PAGE 24

24 Recently, Yu et al. [14] conducted a numerical study of the membranebased absorption process and determined that the benefits of the process in terms of increased absorption rate could only be realized if the solution channel thickness is less than a few hundred microns. The Yu et al. [14] study suggested that absorption rates as high as 0.012 kg/m2s can be achieved at a solution channel thickness of 50 m. An experimental study of the existing membrane technologies by [34] showed that nanofibrous membranes enable membranebased absorption because they have a reasonable pressure drop at an absorption rate a few times higher than that of the existing technology . In the absorption section of this study, a comprehensive experimental analysis of the membranebased absorption process is conducted [15,35,36] . The studies are performed on a model membranebased ARS extensively instrumented to measure parameters affecting the absorption rate. Tests are conducted at or near the working conditions of a typical ARS. Membranebased Absorber with Micromixing I n a recent theoretical study, Bigham et al. [37] have shown further enhancement of the absorption process in a membranebase absorber through utilizat ion of microstructures on the microchannel walls. Such structures have been readily implemented in mixing the laminar flow in microchannels [38– 42] . In a membranebased absorber, confinement of the LiBr solution flow has provided an opportunity to similarly manipulate the microscale transport events within the solution film. The objective is to move the concentrated solution away from the membranesolution interface, and carry the water rich solution from the middle and bottom of the flow channel t o the membranesolution interface. This approach also ease the manifolding burden of an

PAGE 25

25 absorber utilizing a thin solution flow channel and enhance the utility of the membranebased technology for absorbers with larger capacities. The reason is that thicker solution films can be utilized which reduces the solution flow pressure drop. Bigham et al. [37] employed the type of structures which Strook et al [41] proposed for mixing in mi crochannels . The method involves generating chaotic advection within the flow through stretching and folding the laminar streamlines. Fig. 14 shows a flow streamline over the ridges. As shown, the herringbone structures (hereafter called “ridges”) generate anisotropic resistance to the absorbent flow, which stretches and twists a portion of the absorbent flow volume. Bigham et al. [37] also conducted a parametric study to determine the optimal ridge geometry for a membranebased absorber. Through their simulation, they determined that the ridge height needs to be more that 50% of the main channel height to produce surface vortices with sufficient momentum to impact the main flow and continuously replenish the solution membrane interface with a concentrated solution. In the parametric range of their study, they achieved the maximum absorption rate at a channel width of 1mm, channel height ridge angle of 30 with respect to fluid direction. In this work, this geometry is experimentally examined and performance of microstructures in a membranebased a bsorber is assessed. The results are compared with those of prior membrane based absorbers and conventional falling film ones. The effects of important parameters such as vapor pressure and cooling water temperature are subsequently discussed.

PAGE 26

26 Figure 1 4 . A 3D schematic of surface features at the floor of a microchannel (microchannel is not shown in the picture) and a induced fluid particle streamline Desorber Heat Exchanger Conventional Desorbers Desorber heat exchanger is the other main heat exchanger in absorption refrigeration systems . Desorbers involving nucleate pool boiling [43– 45] and falling film over horizontal or vertical tubes [46– 48] are the common configurations studied in the literature. In the pool boiling configuration, as the name implies, water i s boiled off from a pool of LiBr solution. In a falling film desorber, the LiBr solution is sprayed over a tube bundle while the heating medium flows inside the tubes. The fallingfilm type desorbers are more suitable particularly with low temperature heat sources [49] since formation of thin solution films over the tubes (cf. Fig. 1 5 ) facilitates water desorption. At low surface temperatures, water directly diffuses out of the solution film as long as the solution temperature is high enough to sustain a solution water vapor pressure above the phase vapor pressure. Studies conducted by Charters et al. [43] and Yoshitomi et al. [50] suggest that a superheat temperature (the difference between the wall and the solution sa turation temperatures) of approximately 10 C is required for boiling inception. Flow direction Membra ne

PAGE 27

27 Figure 15 . Schematics showing desorption from a falling LiBr water solution film through (a) direct diffusion and (b) boiling mechanisms T he existing literature provides limited insight on the physics of LiBr solution boiling. The water ebullition process in a LiBr solution is fundamentally different from that which occurs in pure water. It is known that the water bubble growth rate is signi ficantly slow [51] because of the low water diffusion coefficient in the LiBr sol ution. In other words, bubble growth in the LiBr solution is limited by mass diffusion rather than by heat transfer, as in pure water. Consequently, a significant surface superheat temperature is required to grow the bubbles large enough so that they can depart from the heat transfer surface (the buoyancy force should overcome the surface tension force for departure). Slow growing bubbles at moderate surface temperatures impede the surface heat transfer [51] . Lee et al. [44] investigated the pool boiling of LiBr solution at saturation conditions on a heated vertical tube. A higher desorption rate was achieved as compared to a horizontal tube configuration. Lee et al. [44] argued that agitation of the LiBr solution near the surface is responsible for the higher performance. In addition, they reported a significant decrease in bubble size and an increase in surface heat flux (i.e. desorption rate) as the system pressure was increased. (b) Heating fluid Li Br water falling film Water vapor bubble Water vapor Heating fluid (a)

PAGE 28

28 Kim and Kim [48] stu died desorption from falling films on tubes tested at wall superheat temperatures of less than 10 C to avoid boiling. They observed enhancement in desorption rate with an increase in desorber pressure and argued that high solution temperatures at elevated desorber pressures lower the solution viscosity and thickness over the tubes. A thinner solution film was considered responsible for the observed increase in the desorption rate. Shi et al. [52] examined the heat transfer performance of a falling film desorber and reported a heat transfer coefficient more than four times higher, prior to boiling inception, than that of an immersed tube desorber. Membranebased Desorbers In an attempt to reduce the solution film thickness and enhance the desorption rate, Thorud et al. [53] mechanically constrai ned the LiBr solution flow between a solid wall and a heated porous membrane. The surface tension forces at the membranesolution interface prevented the LiBr solution from seeping through the pores. Thorud et al. (2006) conducted studies on devices with 170 m and 745 m thick solution channels and primarily at high superheat temperatures associated with the boiling regime. The desorption rate was higher at the 170 m thick solution channel and enhanced with an increase in pressure difference across the membrane. However, the overall desorption rate was significantly less than that of the falling film desorbers. Thorud et al. [53] do not discuss causes of the poor performance.

PAGE 29

29 Figure 16 . Schematic representation of membranebased desorption process showing (a) direct diffusion of water from the solution flow and ( b) bubble formation and desorption through the membrane during the boiling regime. The work of Thorud et al. [ 53] seems to be the sole published effort on the use of porous membranes for dewatering of the LiBr solution. However, studies on implementation of membranes for venting bubbles from a twophase water stream exist [54– 61] that can provide some ins ights on characteristi cs of the process. Meng et al. [54] showed that hydrophobic membranes could be utilized to successfully vent bubbles from a water stream. Zhu [58] demonstrated that the separation rate of the bubbles is directly proportional to the pressure difference applied across the membrane. Xu et al. [61] suggested a set of criteria that must be met for a bubble to be entirely removed from a two phase stream capped by a hydrophobic membrane. They argued that the b ubbles should stay in contact with the membrane at a velocity lower than a critical value. Otherwise, a stable liquid layer forms between the bubble and the membrane and prevents bubble extraction. Desorption study , in this work, is aimed at understanding the physics of the desorption process involved in a thin LiBr solution flow mechanically constrained by a (a) Water vapor LiBr solutio n Heating f luid Membrane Wall (b) Water vapor b ubbles

PAGE 30

30 nanofibrous polytetrafluoroethylene (PTFE) membrane. The experiments are conducted in both singleand twophase modes (cf. Fig. 16 ) to identify the parameters affecting the processes and to quantify their impact [62 – 64] . A numerical model developed in a prior study [14] is used to analyze the singlephase desorption process. To better understand the results of the twophase desorption mode, a micro scale visualization study is conducted in a separate test platform. Ionic Liquids LiBr solution has been successfully used in ARSs for decades, but it cannot operate in every condition. The applicability of LiBr solution in ARSs is limited because: first it is highly corrosive and second it tends to crystallize at high concentrations and low temperatures. In the last part of this study, the use of Ionic liquids (ILs) as an alternative absorbent in ARSs is investigated. Ionic liquids are organic salts which are liquid near ambient temperature, they have low vapor pressure, and are stable over a large range of operating temperature [65] . I onic liquids are composed of an organic cation, such as an imidazolium or pyridinium ring, and an organic or inorganic anion, such as tetrafluoroborate ([BF4]) or bis(trifluoromethylsulfonyl)imide ([Tf2N]) [66] . Recently Ionic liquids have drawn a great deal of attention due to their unique thermophysical properties . Their use is now being investigated is many applications such as C O2 capture technologies, separation processes and ARSs. Although corrosion inhibitors such as Molybdate are now widely being used in absorption chillers and they have been successful to some extent , the c orrosive nature of LiBr solution is still a challenge in ARSs . Corrosion is especially an issue in the desorber section of an ARS where the solution temperature is high. For instance, t he corrosion rate of copper in a LiBr solution with 55% weight fraction of LiBr is about

PAGE 31

31 fifteen times higher at a temperat ure of 90C compared to a temperature of 25C [67] . This requires the use of cos tly and corrosion resistant materials in the system construction. Corrosion in ARSs can significantly shorten the system life and increases system’s maintenance costs. Also the high corrosiveness of LiBr solution at high temperatures limits the application of multi effect systems cycles which require high driving temperatures (i.e . desorber temperature) . In contrast, IL s have the advantage of being less corrosive . Wasserscheid and Seiler [67] hav e compared the corrosion rates of LiBr solution, ionic liquid 1ethyl 3 methylimidazolium acetate and ionic liquid 1,3dimethylimidazolium di methylphosphate in contact with copper and steel. They have shown that theirs selected IL s have noticeably lower corrosion rates in the mentioned environment. For instance the corrosion rate of steel in ionic liquid 1 ethyl 3 methylimidazolium acetate is 6.7 times lower than that in a 55% LiBr solution at the same temperature. Another challenge in LiBr ARSs is the LiBr crystallization. LiBr solution tends to crystallize when the LiBr solution concentration exceeds the LiBr solubility limit (for instance about 63% at 25C). Crystallization significantly disturbs the operation of the system as crystals can inhibit any narrow passages in the system such as nozzles in falling film absorbers. To avoid crystallization, costly process control is required to monit or and control the cycle concentration. Also crystallization limits the use of air cooled heat exchangers (i.e. absorber and condenser) as they need to operate at higher concentrations to achieve high system capacity and performance. While IL s have very lo w melting temperature and lots of IL s are in liquid phase at ambient temperature. This is a n important opportunity for ARSs and it means that even if IL s lose significant

PAGE 32

32 portion of their water content in the desorber, there is still no danger of crystalli zation in the system. IL s have also their own limitations. Significantly higher viscosity of IL s compared to LiBr solution, considerable cost compared to LiBr solution and much lower diffusivity in water can be named as some of these limitations. High vis cosity would lead to formation of thick films, and low water diffusivity lowers the absorber and desorber performances. Among those limitations, the low water diffusion coefficient of IL s could be compensated to some extent with operation of the system at higher IL s concentrations. Increasing the concentration decreases the water vapor partial pressure and facilitates the absorption process. This cannot be easily done in a LiBr system due to the crystallization problem. The high cost of IL s is another main issue in conventional shell and tube ARSs since they need significant amount of liquid. Therefore, to facilitate the use of IL s in ARSs, introduction of new heat exchanger architectures such as our membranebased heat exchangers is essential. As mentioned, significant research efforts over the past decade have been directed towards studying IL s. Seiler et al. [68] has conducted a useful review on the use of IL s in ARSs. As Seiler et al. [68] argue early reseraches have focused more on measuring thermophisycal and thermodynamic properties of IL s. [65,69– 74] . More recentely , studies is directed towards the heat and mass transfer and cycle perormance of IL s [75– 82] . The number of these studies is limited and they are mostly theoritical . For expample, Zhang and Hu [83] conducted a cycle simulation with IL for a singleeffect absorption chiller. H2O [emim][(CH3)2PO4] was selected as their working pair. Their cycle analysis suggeted that about the same system performance (COP) as a LiBr

PAGE 33

33 system can be achived with selected working pair but at much higher concnetrations and a lower driving temperature compared to a conventional LiBr water system. Through cycle simulations, other researches have also confirmed that about the same system performance can be obtained compared to conventional systems but at a different operating condition. Experimental studies on IL s specially funamental heat and mass transfer studies are very limited. Those limited studies are also more like charging an exisiting absorption chiller with an IL and comparing the system performance and capacity with conventional working pairs. For instance, TU Ber lin and Evonik Industries [78,82] have tested different IL s in a typical smallscale singl e effect absorption chiller . Their findings also verified the theoritical results about getting close COP compared to LiBr water systems. However the system capacity was lower (i.e. less than two third of conventional working pairs ) . As Seiler et al. [68] suggests, more accurate results and evaluations of IL s is only possible if underlying heat and mass transfer phenomena is fundamentally studied. The last part of this study concentrates on evaluation of the heat and mass transfer phenomena of an IL in a membranebased absorber. Ionic liquid [EMIM][MeS O3] water (E thyl M ethylimidazolium M ethanesulfonatewater) is selected as the working pair since it has a relatively low vapor pressure, it is completely miscible with water and also its properties has been fully defined [65,66] . The absorption results are compared with those of conventional LiBr water working pair.

PAGE 34

34 CHAPTER 2 EXPERIMENTAL SETUPS A bsorption and Desorption Test S etup Fig. 2 1 shows a diagram of the experimental loop in which the membranebased desorber was tested. The loop consists of a LiBr solution line and a refrigerant (water) line. The solution line consists of an absorber, a desorber, a pump, a filter, a solution reservoir, a Coriolis mass flow meter, and two solution heat exchangers. The water line consists of an evaporator, a condenser, a Coriolis mass flow meter, and a water reservoir. In the solution line, a micro gear pump (HNP Mikrosysteme, Germany) drives the weak LiBr solution through a solution heat exchanger, where the solution is preheated to a desired temperature before entering the desorber. In the desorber, the weak LiBr solution is heated by a thin film heater (Ome ga Engineering, CT) to desorb water. The desorbed water vapor flows to a condenser, and the strong LiBr solution leaves the desorber and flows through a heat exchanger, where it is cooled to a preset temperature before entering a Coriolis mass flow meter ( Bronkhorst USA) and then the absorber. The condensed water leaves the condenser and flows through a Coriolis mass flow meter (Micro Motion, CO) to the evaporator, where it is vaporized and supplied back to the absorber. The strong solution flows through th e absorber and absorbs the water vapor generated in the evaporator. The weak solution leaving the absorber flows through a filter and is pumped back to the solution preheater and then to the desorber to complete the cycle.

PAGE 35

35 Figure 21 . A Schematic diagram of the membranebased absorber and desorber test setup. The solution heat exchangers (i.e. HX1 and HX2 in Fig. 21 ) that control the inlet temperature to the absorber and desorber are cooled or heated using thermoelectric cooling/heating (TEC) units. The heat exchangers are made of Inconel coil, fabricated through forming tubes, which are assembled within two aluminum plates with machined grooves to accommodate the coil. The TECs are attached to the heat exchanger assembly (i.e. on the aluminum surface) using a thermally conductive adhesive sheet. The TECs have air cooled heat sinks and fans. The outlet temperature of each solution heat exchanger is controlled using its corresponding TEC control panel. A similar tube heat exchanger with a temperature controller is used for the condenser. The desorber and evaporator heat exchangers are heated using flexible heaters. The applied power Mass Flow Meter Filter Condenser HX1 HX2 Mass F low Meter Solution Pump Generator Chiller Absorber Evaporator LiBr Water Water Vapor Water Vapor Strong Solution Weak Solution Strong Solution Weak Solution

PAGE 36

36 to the heaters is controlled manually. DC power supplies are used to provide power to the heaters and the TECs. The experimental setup is also equipped with two small reservoirs with sight glass to monitor the liquid in the solution and water lines. These reservoirs also serve as compensation chambers and assist in proper charging of the s ystem. Inconel tubing and Monel fittings that are highly corrosion resistant are used in the assembly of the solution line, and stainless steel tubing and fittings are used in the water line. Thermocouple probes (Omega Engineering, CT) with Inconel sheath are used to measure the solution temperature at the inlet and outlet of the absorber and desorber. The vapor temperature in the condenser and evaporator is measured by probes with stainless steel sheath. The solution and water mass flow rates and densities are measured using two Coriolis mass flow meters. Three pressure transducers with a range of 0100 kPa are installed to monitor desorber pressure conditions. Two of the transducers measure the LiBr solution flow pressure at the desorber inlet and outlet. The average desorber solution pressure ( d sP ) is calculated using the readings of these transducers. The third transducer measures the vapor pressure at the desorber vapor exit ( d vP ). Three pressure transducers with a range of 010 kPa screen the absorber pressure conditions. Two of transducers measure the LiBr solution flow pressure at the absorber inlet and outlet, while the third one measures the vapor pressure ( a vP ) at the absorber vapor inlet. The reading of two pressure transducers in the solution line is averaged to calculate the average absorber solution pressure ( a sP ). All the measured data are recorded by a data acquisition system. Fig. 22 shows a photograph of the exper imental loop.

PAGE 37

37 Figure 22 . A photograph of the assembled experimental setup Membraneb ased Absorber Heat Exchanger A schematic of the absorber heat exchanger is depicted in Fig. 23 . The overall dimensions of the absorber are 311 mm 117 mm with an active heat and mass transfer area of 203 mm 38 mm. Solution microchannels are machined in a C 22 Hastelloy plate (cf. Fig. 24 ). Hastelloy is a highly corrosionresistant alloy and is compatible with lithium bromide solution. The solution is constrained within the microchannels by three solid walls (a bottom and two side walls) and a hydrophobic nanofibrous membrane on the top. The membrane used in this test has a nominal pore size of 1 m and is 80% porous. The solution microchannels are 160 10 and 100 10 m deep (in two different tests), 1 mm wide and 38 mm long. Solution P re heater Solution P re cooler Mass Flow M eter Filter Evaporator Absorber Solution P ump Condenser Desorber Mass Flow M eter

PAGE 38

38 V Water v apor Solution i nlet Solution o utlet Cooling water o utlet Cooling water i nlet Cooling water microchannels Li Br Solution microchannels Membrane Thermocouple wires Figure 23 . A schematic representation of the absorber heat exchanger cross section. Figure 24 . A photogr aph of the main absorber part (s olution side) . A perforated stainless steel sheet ( cf. Fig. 2 5) was used to support and secure the membrane on the microchannels . The perforated plate has a 51% open area, a hole diameter of 1.2 mm, and a thickness of 0. 5 mm. T he support plate is then inserted in a slot machined in a stainless steel frame ( cf. Fig. 2 6 ) and is fastened to the main Hastelloy part by 10 stainless steel screws. A narrow crossbar in the middle of the frame provides additional structural integrity and prevents the support plate from buckling. The crossbar does not block the vapor path due to its small size and the staggered pattern of the perforated sheet. Fig. 27 shows a photograph of the membrane assembly fixture attached to the main absorber part . Inlet Outlet Inlet manifold Solution microchannels 100 &160 1 m m 100

PAGE 39

39 Figure 25 . A photograph of the stainless steel perforated sheet. Figure 26 . A photograph of the stainless steel frame. Figure 27 . A photograph of the membrane assembly fixture attached to the main absorber part . The water vapor generated in the evaporator flows through the membrane and gets absorbed by the strong lithium bromide solution. To cool the solution, m icrochannels are machined on the backside of the Hastelloy plate (cf. Fig. 28 ). The

PAGE 40

40 height and width of the water channels are 0.4 mm and 4 mm, respectively. The cooling water from a chiller is pumped through the channels. Figure 28 . A photograph of the main absorber part (Cooling water side). To measure the surface temperature in the solution side, several thermocouple slots were machined on the back side of the main Hastelloy part as shown in Fig. 28 . A silver filled epoxy was first used to secure the thermocouple tip inside the slots. After curing the silver filled epoxy overnight, it was covered by a BONDiT B45 epoxy to protect the thermocouples in the slot against fast flowing cooling water. Since, the thermocouple wire insulation is made of Teflon, common epoxies do not adhere well to the wire insulation. This epoxy is designed for hardto bond materials such as Teflon. After bonding the thermocouples to the Hastelloy part, all thermocouple wires were passed through small holes drilled in a brass cover (cf. Fig. 2 9 ) . To measure the cooling water temperature distribution along the channels, seven thermocouples were mounted on the cooling water side. The tips of these thermocouples should be in direct contact with the cooling water. Care was taken to ensure that the thermocouple tips were not in contact with the channel wall. To seal the thermocouple , BONDiT B45 Inlet manifold Cooling water microchannels Thermocouple slots

PAGE 41

41 was first applied followed by a metal to metal Epoxy Adhesive DP420 which offers high peel and shear strength. Two different brass parts were designed and manufactured to cover and seal the microchannels on the solution and water sides. Photographs of the fabricated parts are shown in Fig. 2 9 and Fig. 210. The cover for the solution side has a port at the center to supply water vapor from the evaporator to the solution microchannels in the absorber. The cover for the water side has two 1/4” (6.35 mm) inlet and outlet ports for cooling water and several small holes for thermocouple wires. Since the absorber operates at low sub atmospheric pressures, it must be vacuum sealed. Two oring grooves are machined on the brass parts for this pur pose (cf. Fig. 2 9 and Fig. 210) . Figure 2 1 1 shows a photograph of the assembled absorber where all the parts are fastened together by 16 stainless steel screws and nuts. Figure 29 . A photograph of the brass cover (water side). Thermocouple holes Cooling water inlet Cooling wa ter outlet

PAGE 42

42 Figure 210 . A photograph of the brass cover (solution side). Figure 211 . A photograph of the assembled absorber heat exchanger . Membraneb ased Absorber Heat Exchanger with Ridges A p hotograph of the absorber heat exchanger with added micro ridges is depicted in Fig. 212. The overall dimensions of the absorber are 311 mm 117 mm with an active heat and mass transfer area of 195 mm 38 mm. Solution microchannels are machined in a brass plate. The solution microchannels are 500 m deep, 1 mm wide and 195 mm long. At the floor of microchannels, as shown in Fig. 212, herringbone structures (hereafter called “ridges”) are machined. Fig. 213 shows the dimensions of microstructures. As mentioned earlier, the dimensions used are from the optimized geometry discussed in Bigham et al. [37] . Vapor inlet

PAGE 43

43 Figure 212 . A photograph of the absorber heat exchanger 300 30 30 1m m n=30 m=30 500 2 0 300 Figure 213 . Microchannels and micro structures dimensions The solution is constrained within the microchannels by three solid walls (a bottom and two side walls) and a hydrophobic nanofibrous membrane on the top. The membrane used in this test has a nominal pore size of 1 m and is 80% porous. A perforated stainless steel (SS) plate with a pore size o f 1.2 mm, a thickness of 0.5 mm,

PAGE 44

44 and an open area of 63% is used to support and secure the membrane on the microchannels. Fig. 2 14 shows the absorber heat exchanger with installed membrane which is secured by the perforated plate. Figure 214 . A photograph of the absorber heat exchanger with the installed membrane To cool the solution, microchannels are machined on the backside of the brass plate. The height and width of the water channels are 0.4 mm and 4 mm, respectively. The cooling water from a chiller is pumped through the channels . Membraneb ased Desorber Heat Exchanger A schematic of the desorber heat exchanger cross section is provided in Fig. 21 5 . The desorber consists of: (1) a corrosionresistant C 22 Hastelloy plat e in which the solution microchannels are machined (2) a brass enclosure with a sight glass. The overall size of the desorber is 16.8 16.5 cm2. The solution microchannels are machined over a 5.7 8.9 cm2 area (cf. Fig. 21 6 ). A nanofibrous polytetrafluoroethylene (PTFE) membrane with a pore size of 0.45 m and a thickness of 50 m is placed on the solution microchannels and secured in place by a perforated stainless steel sheet (cf. Fig. 2 1 7 ) .

PAGE 45

45 Water v apor Weak Solution Strong S ol ution Li Br Solution microchannels Membrane Flexible Heater + / Figure 215 . Schematic diagr am of the desorber heat exchanger. Figure 216 . A photograph of the desorber solution microchannels. 1 m m

PAGE 46

46 Figure 217 . A photograph of the membrane assembly fixture attached to the main desorber part. Twelve thermocouples are installed within three trenches machined on the backside of the Hastelloy plate to measure the wall temperature. The remaining space within the trenches is filled with a high temperature thermally conductive epoxy and a flexibl e thin film heater (Omega Engineering, CT) is subsequently assembled over the entire surface (cf. Fig. 2 1 8 ). Fig. 21 9 shows a photograph of the assembled desorber heat exchanger.

PAGE 47

47 Figure 218 A photograph of the installed t hermocouples and flexible heater on the back side of the desorber Figure 219 . A photograph of the assembled desorber heat exchanger .

PAGE 48

48 Membraneb ased Evaporator Heat Exchanger A thin film evaporator was designed and fabricated to generate and supply water vapor to the absorber for a wide range of operations conditions . T he overall size of the evaporator is 219 mm 187 mm. The thinfilm evaporator consists of a main copper part with machined microchannels (cf . Fig. 220) , a water supply port, a brass cover, and a membrane/support assembly (cf. Fig. 2 21) . The dimensions of the microchannels are 200 m high, 2 mm wide, and 102 mm long and they are machined over a 10.2 12.7 cm2 area. A flexible heater (Omega E ngineering, CT) is attached to the back side of the copper plate to provide heat for water evaporation. As shown in Fig. 21 8 , the scheme for the membrane assembly on the microchannels is the same as those used in the absorber and desorber heat exchangers . Figure 222 shows a photograph of the assembled thin film evaporator. Figure 220 . A front view of a thin film evaporator .

PAGE 49

49 Figure 221 . A photograph of the membrane assembly fixture attached t o the thin film evaporator . Figure 222 . A photograph of the assembled thin film evaporator .

PAGE 50

50 Tube Heat E xchangers and Condenser Efforts were made to find suitable commercially available heat exchangers that are compatible w ith the lithium bromide solution. However, commercially available heat exchangers are made from stainless steel or copper which are not compatible with the lithium bromide solution. Thus, resistant tube heat exchangers was designed and fabricated. To make the heat exchangers a 1/4” Inconel tube was bent into a serpentine shape and sandwiched between two machined aluminum plates and fastened by screws as shown in Fig. 223. The solution flows through the serpentine tube and is cooled or heated using a thermo electric cooler (TEC) module. One side of the TEC module is attached to the aluminum plate and the other side is attached to a fancooled aluminum heat sink using a thermal conductive adhesive sheet. A similar configuration is used for the heat exchanger installed before the desorber. In this case, the TEC module works as a heat pump to heat the solution. The TEC modules and fans are powered by external DC power supplies. To control the desired inlet temperatures to the absorber and desorber heat exchangers, two thermoelectric temperature controllers ( cf. Fig. 22 4 ) capable of controlling up to 50 volts and 20 amps are used to adjust the power input to the TEC module via pulse width modulation and onboard power transistors. An important advantage of this controller is that it provides an analog proportional output signal, which allows continuous power control. A thermistor mounted on the aluminum plate continuously sends an input signal to the controller. A similar tube heat exchanger with a temperature controller is used as the condenser.

PAGE 51

51 Figure 223 . A photograph of the assembled tube heat exchanger . Figure 224 . A temperature controller used with tube heat exchangers . Vacuum Leak Tests Prior to vacuum leak tests, each individual component was first cleaned with acetone, ethanol and distilled water to remove any grease stain or dirt to minimize degassing problems. Each component was then vacuumed and its pressure was monitored by a highly accurate vacuum pressure transducer with a resolution of 0.2 Pa. All heat exchangers and instruments including mass flow meters, pressure transducers , pump, and thermocouple probes were tested separately before adding them to the loop. The assembly approach was to add one component at the time and after each step the system was vacuumed and leakage tested again. Although great efforts were made in

PAGE 52

52 selecting and fabricating airtight components, vacuum sealing proved to be quite challenging and the vacuum leak troubleshooting took a lot of time . Unlike conventional systems, this loop has a very small volume and a large surface area. Thus, a very small leak c ould become a significant source of pressure increase in the loop. Brazing To minimize potential leakage points, a 0.25 inch outside diameter , 2 inch long tubing was brazed to all heat exchanger ports. Fig. 22 5 shows a photograph of the brazed joints on a Hastelloy part. Brazing is a metal joining process where a filler material flows by capillary action between two closefitting parts with the application of heat. Due to material compatibility issues, nickel based filler materials was selected and used in the brazing. The brazing can be done using a torch or in a vacuum furnace. Since the torch brazing can produce thermal stresses which can cause distortion in the parts, the brazing was performed in a vacuum furnace. The vacuum furnace brazing provides a controlled heat cycle which l eads to a strong metallurgical bond. A 0.25 inch inconel tubing was used to braze parts in the solution side while stainless steel tubing was used to braze parts in the water line. Hastelloy brazing was performed at a shop with extensive expertise and exp erience with Hastelloy brazing which requires short heating and cooling cycles. In addition, care was taken when brazing brass parts since zinc can be released in the process due to the fact that the melting point of zinc (420C) is lower than that of bras s (900C). Loss of zinc can destroy the proportions of the metals in brass and also may causes pinholes or porosity in the joint, resulting in reduced strength.

PAGE 53

53 Figure 225 . A photograph of the brazed joints on a Hastelloy part . Insulation Since the evaporator, absorber operate at temperatures lower than the ambient temperature, they were well insulated to prevent heat gains from the surroundings. In addition, the desorber and solution heat exchanger operate at temperatures m uch greater than the ambient and need to be insulated. Silicone rubber foam sheets with a thickness of 3/8 inch were used to insulate these components. Each component was first wrapped with the silicone rubber foam and secured with a duct tape. Photographs of the absorber before and after insulation are shown in Fig. 2 2 6 . Each component was then encased in an aluminum U channel for a better stability. Figure 2 2 7 shows photographs of the final encased components. a) Absorber before insulation b ) Abso rber after insulation Figure 226 . Photographs of an absorber before and after insulation.

PAGE 54

54 a) Absorber b) Desorber c) Evaporator d) Tube heat exchanger Figure 227 . Photographs of fully insulated components . Charging the Loop The following procedures wer e developed to charge the water and LiBr solution lines. Since the membrane is delicate, the loop cannot be directly charged from the atmospheric pressure. If the charging line pressure exceeds the breakthrough pressure limit, the liquid is forced through the membrane. To prevent this, water or LiBr solution is first charged from a container at atmospheric pressure to a chamber at an intermediate pressure of ~10 kPa. The loop was then charged from the intermediate chamber pressure. A photograph of the changing setup is shown in Fig. 2 2 8 . A 55%

PAGE 55

5 5 Lithium Bromide solution with Lithium Molybdate inhibited (LevertonClarke Ltd, UK) is used to charge the loop. Before charging, the intermediate chamber and the main loop were vacuumed. Then, a valve connecting the LiBr solution container to the intermediate chamber was opened until the intermediate chamber was filled to a certain level. This valve then was closed and a valve connecting the chamber and the main loop was opened. A LiBr reservoir with a sight glass (cf . Fig. 22 9 ) located at the highest elevation of the loop was used to ensure that sufficient LiBr was charged into the loop. The reservoir was machined in a Naval Brass alloy (Alloy 464), which offers a good corrosion resistance to the salt solution. Figure 228 . A photograph of the fabricated charging components.

PAGE 56

56 Figure 229 . A photograph of a lithium bromide solution reservoir. This procedure was repeated for charging the water line. A sim ilar reservoir with sight glass (cf. Fig. 229 ) was used in the loop to monitor the water level. The water line was charged with degassed, deionized water. A vacuum distillation setup was used to remove dissolved non condensable gasses in distilled water. A photograph of the setup is shown in Fig. 2 30. This setup consists of a boiling flask on the right, a receiving flask on the left, and a water cooled condenser in between. The water was brought to a boil in the boiling flask using a hot plate. The water vapor flows to the condenser and condenses on the surface of the cooling coil. The liquid water was accumulated in the receiving flask and the noncondensable gases exited through a purge port. The whole charging process was repeated for testing ionic liquid [EMIM] [MeSO3] . Sight glass Charging port

PAGE 57

57 Figure 230 . A photograph of a vacuum distillation set up. Working Fluids’ P roperties As mentioned in the previous sections, two working pairs (i.e LiBr water and [EMIM][MeSO3] water) are tested in the absorption experimental loop. In both cases, water is used as the refrigerant. In this section the properties of these fluids are presented. The properties that are involved in our analysis include thermal conductivity, specific heat, density , vapor press ure, viscosity, water diffusion coefficient and enthalpy . Table 21 lists specific heat, density and thermal conductivity of the three working fluids at room temperature. 55% weight fraction is selected for LiBr solution which is about the average working concentration of LiBr solution in ARSs.

PAGE 58

58 Table 2 1 . Working fluid properties at room temperature ( i.e. 25C) Variable value Water LiBr solution (55%) Pure [ EMIM][ MeSO3] Specific heat (cp) 4.183 k J/kgK (a) 2 . 057 k J/kgK (b) 1.6 k J/kgK (d) 997 kg/m3 (a) 1592 kg/m3 (a) 1241.52 (e) T hermal conductivity (k) 0.59 W/mK (a) 0.44 W/mK (c) 0.21 W/mK (f) (a) Data are obtained from educational version of Engineering Equation Solver (EES) Software (b) Data are obtained from ASHRAE [84] (c) Data are obtained from Florides et al. [85] (d) Data are obtained from Ficke [66] (e) Data are obtained from Hasse et al. [86] (f) Data are obtained from Tenney et al. [87] Water vapor pressure is plotted in Fig. 231 as a function of temperature. As expected, vapor pressure increases with the tem perature. Fig. 232 compares the vapor pressure of LiBr solution and ionic liquid [EMIM] [MeSO3] at two temperatures of 25C and 40C as a function of weight fraction (or concentration) . Vapor pressure beyond 3 kPa is not plotted since they are not useful in absorption studies . In fact, the absorbent vapor pressure has to be below the evaporator pressure condition (i.e. ~1kPa) to be able to absorb water vapor. For LiBr solution, values beyond the solution concentration of 65% are not plotted either because LiBr would crystallize beyond that point. Further discussion about crystallization of LiBr will be provided in next sections.

PAGE 59

59 T (oC ) V a por P r e s s ur e ( kP a ) 0 1 0 2 0 3 0 4 0 5 0 0 2 4 6 8 Figure 231 . Water vapor pressure as a function of temperature ( values are obtained using educati onal version of Engineering Equation Solver (EES) Software, McGraw Hill, Inc.) . Fig. 232 suggests that , at the temperature of 25 C, ionic liquid [EMIM][MeSO3] would be able to absorb water vapor at evaporator vapor pressure condition (i.e. ~1kPa) only if its concentration is beyond 85%. While this value for LiBr solution is much lower (i.e. ~ 48% ) which explains the popularity of LiBr solution in ARSs.

PAGE 60

60 X % ( w e i ght f r a c t i on) V a por pr e s s ur e ( kP a ) 4 0 5 0 6 0 7 0 8 0 9 0 1 0 0 0 0 . 5 1 1 . 5 2 2 . 5 3 3 . 5 [ E M I M ] [ M e S O3] w a t e r L i B r s o l u t i o n T = 2 5oC T = 4 0oC T = 2 5oC T = 4 0oC Figure 232 . Vapor pressure as a function of temperature ( values for LiBr solution are obtained using educational version of Engineering Equation Solver (EES) Software , McGraw Hill, Inc. and values for [EMIM](MeSO3] are taken from Ficke [66] ) . Another property that significantly affects the pressure drop in the heat exchangers and consequent ly the system required pump power is the viscosity of the desiccant. The pressure drop in the refrigerant (i.e. water) line does not affect the performance of the cycle. The viscosity of water at 25 C is about 0.0009 kg/ms. Fig. 2 33 compares the viscosi ty of typical 55% LiBr solution and pure ionic liquid [EMIM] [MeSO3] as a function of temperature at different concentrations. As observed, the viscosity of ionic liquid [EMIM] [MeSO3] is more than one order of magnitude (i.e. 10 to 30 times) higher than a 55% LiBr solution. This directly translates to higher required pumping power. The high pressure drop in the loop, aside from higher electricity

PAGE 61

61 consumption, increases the pressure on the membrane which requires employment of thicker and stronger membranes in the loop. Thicker membranes, as it will be discussed later, increase the absorption resistance and decrease the absorption rate. It should be mentioned that the reported viscosity for ionic liquid [EMIM] [MeSO3] is for the pure IL ; however in actual c ondition a mixture of IL and water is running in the loop. IL water mixture is expected to have a lower viscosity but still it is much higher than that of the LiBr solution. T (oC ) V i s c os i t y ( kg/ m s ) 2 0 3 0 4 0 5 0 6 0 7 0 1 0-3 1 0-2 1 0-1 1 00[ E M I M ] [ M e S O3] w a t e r L i B r s o l u t i o n ( 5 5 %) Figure 233 . Absorbent dynamic viscosity as a func tion of temperature ( values for LiBr solution are obtained using educational version of Engineering Equation Solver (EES) Software , McGraw Hill, Inc. and values for [EMIM](MeSO3] are taken from Hasse et al. [86] ) . Enthalpies of LiBr solution and ionic liquid [EMIM] [MeSO3] are obtained from EES software and Ficke [66] , respectively; however they are not reported here. Diffus ion coefficient is another desiccant’s properties which drastically impacts the

PAGE 62

62 absorption and desorption processes. The reported values of water diffusion coefficient for LiBr solution are not consistent; however an averaged value of about 2x109 m2/s can be assumed with reasonable accuracy for a 55 % LiBr solution [88] . Diffu sion coefficient measurement using d ynamic light s cattering (DLS) was also conducted in our group and, for a 55% LiBr solution at room temperature, the diffusion coefficient was found to be 2.2x109 m2/s. The diffusion coeffi ci ent of [EMIM][ MeSO3] in water is much lower and it has been reported to be about 1.64x1010 m2/s for a 98% ionic liquid [EMIM](MeSO3] (i.e. mass fraction=98%) [89] . Since mass transfer mechanism in laminar flows is diffusion process, much lower absorption rates are expected for desiccant [EMIM](MeSO3] compared to LiBr solution. Experimental Procedure After charging the loop, each experimental run began by operating the solution pump and setting the flow rate to a desired value. The water chiller was then turned on and the temperature of the absorber cooling water was set. A valve between the evaporator and absorber was then The system was assumed to have reached steady state when variations in the absorber pressure and the solution density were within 10 Pa and 5 kg/m3, respectively, for at least 30 min utes . T he absorption/desorption rate was then directly measured by the water line mass flow meter . Cycle A nalysis and Test P arameters To determine working conditions (temperature, pressure, and mass flow rate) of absorber and desorber heat exchangers, a thermodynamic analysis of a 1ton doubleeffect water/LiBr absorption system was performed. Countercurrent flow is considered in all heat exchangers. Fig. 23 4 shows a schematic diagram of a doubleeffect

PAGE 63

63 water/LiBr absorption system. The cycle includes a high desorber, a high condenser, a low desorber, a low condenser, an absorber, an evaporator, two solution heat exchangers, pumps, and expansion valves. A double eff ect system has two pressure stages to regenerate the refrigerant and one pressure stage to absorb the refrigerant. There could be two basic configurations for a double effect system; in series or parallel. In a parallel system, like the one selected for this work, the LiBr solution flow, after the absorber, is pumped to the medium pressure level of the system and then is branched into two streams (cf. Fig. 2 3 4 ). Each stream enters a different desorber. The One that flows into the high desorber is pumped to the high pressure level of the loop beforehand. The LiBr solution streams after losing some of their water content in desorbers combine and flow back to the absorber. Expansion valves exist in the return line to break the pressure to the low pressure level of the system. In this configuration, the high pressure and temperature regenerated water vapor in the high desorber is employed to heat up the low desorber and produce more water vapor. The produced water vapor in the low desorber flows into a condenser where it condenses and mixes with the other water stream that originated from condensation in the low desorber (cf. Fig. 23 4 ). The water stream finally enters the evaporator where it is vaporized and fed to the absorber. The cycle was modeled by apply ing mass and energy balances on all of these components. The external heat transfer interactions are represented as heat transfer loops. Heat transfer rates for each component were expressed as functions of U, A, and LMTD where U is the overall heat trans fer coefficient, A is the heat transfer surface

PAGE 64

64 area, and LMTD is the logmean temperature difference. Eq. (1 ) shows sample mass and energy balances equations for the absorber heat exchanger. ) ( ) ( ) ( ) ( / ) (23 1 24 6 23 1 24 6 23 24 23 1 1 6 6 10 10T T T T Ln T T T T LMTD LMTA Q UA T T C m Q h m Q h m h mabs abs abs abs p abs abs ( 2 1 ) The numbers in the indices of Eq. (2 1 ) shows the location of the thermophysical property in the cycle (cf. Fig. 2 3 4 ). The same type of equations was written for each heat exchanger in the cycle. In addition, heat transfer models were included by imposing effectiveness for solution heat exchangers. The UA values, solution heat exchanger effectiveness, the mass flow rate of the working fluid in the external loops, and the inlet temperature of the fluid to each external loop are inputs to the model. The pressure drops between the evaporator and absorber, and between the desorber and condenser are neglected. The input values for the bas eline case are listed in Table 22 . An Engineering Equation Solver (EES) Program with built in thermophysical property functions was used to solve the governi ng equations. The model calculates coefficient of performance, heat transfer rates, state point temperatures, pressures, and mass fractions. Table 2 3 depicts heat transfer loads of various components and COP (i.e. coefficient of performance) of the whole system for the baseline case. It should be noted that the heat load of the evaporator is essentially the system’s capacity of refrigeration. The coefficient of performance for absorption cycles in a double effect absorption refrigeration system is defined as evaporator heat load or system refrigeration capacity

PAGE 65

65 divided by heat input to the high desorbers. Table 24 lists the flow rate, partial pressure, temperature and concentration of each point for the baseline case. Strong solution HX1 HX2 Evaporator Strong solution Weak solution Strong solution Weak solution Weak solution Water vapor Water vapor Condensed water Condensed water Water vapor Condenser Low desorber High desorber Absorber 2 1 3 16 14 15 13 12 11 4 17 18 19 7 8 9 10 21 22 23 24 26 25 27 28 Figure 234 . Schematic diagram of a double effect absorption system Table 22 . Model inputs for the baseline case Parameter Value Parameter Value Paramete r Value 1m 0.01 kg/s 21T 140 C Absorber UA 0.5 kW/K 21m 0.1 kg/s 23T 25 C Evaporator UA 0.85 kW/K 23m 0.2 kg/s 25T 32 C Condenser UA 0.65 kW/K 25m 0.2 kg/s 27T 12 C Low desorber UA 0.65 kW/K 27m 0.2 kg/s HX s effectiveness 0.8 High desorber UA 0.25 kW/K

PAGE 66

66 Table 23 . Cycle COP and heat transfer loads of each component of a 1 Ton system for the baseline case Parameter Value Absorber heat transfer load ( absQ ) 4.33 kW Evaporator heat transfer load ( evapQ ) 3.45 kW Condenser heat transfer load ( cndQ ) 1.72 kW Low desorber heat transfer load ( des lQ ) 1.96 kW High desorber heat transfer load ( des hQ ) 2.61 kW 1st solution heat exchanger eat transfer load (1 HXQ ) 0.43 kW 2nd solution heat exchanger heat transfer load (2 HXQ ) 0.33 kW Cycle Cop ( COP ) 1.32 To obtain an insight into the effect of various parameters on the cycle performance, inputs to the model were varied over a wide range by runni ng more simulations. It should be noted that the experimental test system is much smaller than a 1 ton (i.e.3.5 kW) system. The goal was to build a 100 W system. Therefore any optimized extensive parameter in this section has to be divided by 35 to obtain the experiment’s parameter. Figure 23 5 shows the effect of solution flow rate on the cycle performance and system’s capacity of refrigeration . The solution flow rate can affec t both the absorption and the desorption process. The results indicate that the COP decreases as the solution flow rate increases. The reason is that by increasing the solution flow rate more sensible heat is required in the high desorber to heat the solut ion to a certain temperature. The results also indicate that cycle refrigeration capacity initially increases with increasing solution flow rate, and then it reaches a maximum at a solution flow rate of about 0.15 kg/s. Increasing the flow rate beyond this point does not change the refrigeration capacity significantly. Fig. 23 5 suggests that to achieve high coefficient of

PAGE 67

67 performance and high refrigeration capacity, the solution flow rate should be around 0.012 kg/s. Table 2 4 . Components’ conditions for the baseline cas e Point ) / ( s kg m ) (kPa Pi T(C) X(%) 1 0.01 0.892 28.89 52.082 2 0.01 5.95 28.89 52.082 3 0.01 5.95 48.89 52.082 4 0.009 5.95 81.62 60.996 5 0.009 5.95 55.26 60.996 6 0.009 0.892 46.47 60.996 7 0.001 5.95 63.33 8 0.001 5.95 36.02 9 0.001 0.892 5.32 10 0.001 0.892 5.32 11 0.006 5.95 63.33 52.082 12 0.006 50.714 63.34 52.082 13 0.006 50.714 90.44 52.082 14 0.005 50.714 135.13 60.996 15 0.005 50.714 99.24 60.996 16 0.005 5.95 83.19 60.996 17 0.001 50.714 114.65 18 0.001 50.714 81.69 19 0.001 5.95 36.02 21 0.1 140 22 0.1 133.79 23 0.2 25 24 0.2 30.16 25 0.2 32 26 0.2 34.05 27 0.2 12 28 0.2 7.9

PAGE 68

68 ms o l( kg/ s ) C O P Qe v a p( kW ) 0 0. 01 0. 02 0. 03 0. 04 0 0. 4 0. 8 1. 2 1. 6 2 1 2 3 4 5 6C O P Qe v a p . . . Figure 235 . Effect of s olution mass f low r ate on cycle COP and refrigeration c apacity The effect of absorber cooling water temperature (i.e. T2 3), which significantly affect the performance of the absorber, on the cycle performance and system’s capacity of refrigeration is shown in Figure 23 6 . The other cycle input parameters were kept constant as listed in table 22 . As expected increasing the cooling water temperature decreases the system capacity and the COP. However, the cycle capacity is more sensitive to the cooling water temperature. Fig. 23 6 shows that with an absorber cooling water temperature of 25C , a relatively high COP and System capacity can be achieve d. Designing a system with a cooling water temperature below this point (i.e.25C) is not very practical since costly cooling tower technologies are required.

PAGE 69

69 T2 3(oC ) C O P Qe v a p( kW ) 20 25 30 35 40 0 0. 4 0. 8 1. 2 1. 6 2 1 2 3 4 5 6C O P Qe v a p . . Figure 236 . Effect of absorber cooling water temperature on cycle COP and refrigeration c apacity One other factor that significantly affects the absorption performance is the evaporator vapor pressure. It is known that increasing the evaporator vapor pressure increases the absorption rate. However it adversely affects t he system performance and capacity. In cycle analysis, the change in the evaporator vapor pressure (i.e.P10) can be achieved by varying the chilled water inlet temperature (i.e. T27). Fig. 23 7 shows the variation of system performance and capacity with the chilled water inlet temperature. Fig. 23 8 shows the corresponding evaporator vapor pressure. As shown in Fig. 23 7 , increasing the chilled water inlet temperature (or evaporator vapor pressure) increases the system performance and capacity. The trade of f here is that the chilled water outlet temperature also increases with the increase of the chilled water inlet temperature. This vastly limits the application of the system. For some industrial applications, the returning

PAGE 70

70 chilled water enters the evaporat or at a temperature of about 12C. For this case, the cycle analysis indicates an evaporator vapor pressure of 890 Pa (c.f. Fig. 2 3 8 ) and a chille d water outlet temperature of 7.9C. T2 7(oC ) C O P Qe v a p( kW ) 5 10 15 20 0 0. 4 0. 8 1. 2 1. 6 2 1 2 3 4 5 6C O P Qe v a p . . Figure 237 . Effect of chilled water inlet temperature on cycle COP and refrigeration capacity

PAGE 71

71 T2 7(oC ) P1 0 5 10 15 20 0 0. 4 0. 8 1. 2 Figure 238 . Variation of evaporator vapor pressure with chilled water inlet temperature It is expected that increasing the temperature of heating medium (i.e. T21) in th e high desorber increases the desorption rate. But it also changes the system capacity and performance. The effect of heating medium temperature on the system capacity and performance is investigated in Fig. 239. The heating medium temperature was changed from 120C to 160 C. As shown, increasing the desorber temperature increases the system capacity and it slightly decreases the system performance. Therefore higher temperatures ar e desired; however it is not accessible everywhere. Also working at higher temperature significantly accelerates corrosion in the system.

PAGE 72

72 T2 1(oC ) C O P Qe v a p( k W ) 110 120 130 140 150 160 170 0 .8 1 .2 1 .6 1 2 3 4 5 6C O P Qe v a p . . Figure 239 . Effect of heating medium inlet temperature on cycle COP and refrigeration capacity After understanding the cycle reaction to the main cycle parameters, it is time to decide about the range of input values for experimental studies. The studies on the thin film absorption process were conducted on effects of water vapor pressure, cooling water temperature, solution inlet temperature, solution film thick ness and velocity. Table 2 5 summarizes the tests conditions for absorption studies . Table 25 . Input values for parametric studies on thin film absorption Parameter Nominal Range LiBr-water solution flow rate (a solm ) 0.6 kg/hr 0.6-2.1 kg/hr Cooling water inlet temperature (a cwT) 25 C 25-35 C LiBr-water solution temperature at absorber inlet ( a inT ) 25 C 25-35 C Vapor pressure ( a vP ) 1.1 kPa 0.8-1 .8 kPa LiBr-water solution concentration at the absorber inlet (a inX) 60 1.5% NA

PAGE 73

73 To characterize absorption process with micromixing , the investigations were conducted on effects of water vapor pressure, cooling water temperature, sol ution inlet temperature, and velocity. Table 26 summarizes the tests conditions for absorption with micromixing studies . Table 26 . Input values for parametric studies on absorption with micromixing Parameter Nominal Range L iBr-water solution flow rate ( a solm ) 2.5 kg/hr 0.8-2.9 kg/hr Cooling water inlet temperature (a cwT) 25 C 25-35 C LiBr-water solution temperature at absorber inlet ( a inT ) 25 C 31-43 C Vapor pres sure (a vP) 1.1 kPa 0.8-1.6 kPa LiBr-water solution concentration at the absorber inlet (a inX) 60 1.5% NA To evaluate absorption process with [EMIM][MeSO3] water pair, tests were conducted at the test conditions l isted in Table 28. The effort was to choose the test parameters close to those of LiBr solutionwater pair (i.e.Table 22). Due to high viscosity of ionic liquid [EMIM][MeSO3] and consequently high desiccant pressure drop inside heat exchangers, the desic cant flow rate was kept low for the safe operation of membranes. It should be noted that ionic liquid [EMIM][MeSO3] was tested in the membranebased absorber with micro ridges. Table 27 . Input values for parametric studies on absorption with IL [ EMIM][MeSO3] Parameter Nominal Range IL-water flow rate (a solm ) 0.63 kg/hr Cooling water inlet temperature (a cwT) 25 C IL-water temperature at absorber inlet ( a inT ) 25 C Vapor pressure (a vP) 1.1 kPa 0.9-1.6 kPa IL-water concentration at the absorber inlet (a inX) 96.5 0.5%

PAGE 74

74 Studies on the desorption process were conducted on the effects of wall temperature, vapor pressure, soluti on pressure, and solution velocity on the desorption rate. T able 27 summarizes the tests’ conditions for desorption tests . Note that pressure in a typical desorber heat exchanger of a single effect ARS is approximately 10 kPa. Efforts were made to maintai n the solution concentration (i.e. the LiBr weight fraction) in the desorber close to that of an actual system. The solution flow rate and the heater power were adjusted to achieve this objective. These parameters were carefully controlled to avoid crystal lization of the solution that occurs if the solution concentration at the desorber exit exceeds the LiBr solubility limit (cf. Fig. 2 40) . Table 2 8 . Input values for parametric studies on thin film desorption Parameter Nominal Range LiBr-water solution flow rate (d solm ) 2.5 kg/hr 0.75-3.25 kg/hr LiBr-water solution pressure (d sP) 23 kPa 13-30 kPa LiBr-water solution temperature at desorber inlet ( d inT ) 70 C NA Vapor pressure ( d vP ) 10 kPa 6 -18 kPa Average wall temperature ( T w ) NA 50 125 C Experimental Uncertainty Table 29 lists uncertainty in measurement of pressure, concentration, solution flow rate, and temperature. The accuracy of the water mass flow meter that directly measures the water desorption rate at the condenser exit is 1%. However due to the unsteady nature of the condensate flow, a fluctuation of up to 5% was recorded during the experiment. The reported desorption rates are the average of the measured values over a period of time, after the system reached a steady state. The concentration uncertainty is calculated using the following equation.

PAGE 75

75 2 2 T T X X X ( 2 2 ) where X and T are the solution concentration (i.e. mass fraction) and temperature, respectively, and is the solution density. Table 2 9 . Variable uncertainties Variable Uncertainty Pressure (P) 0.5% Concentration (X) 0.25 Solution flow rate ( solm ) 0.2% Temperature ( T ) 0.3 C Da ta Reduction The solution mass flow rate and density are measured by the Coriolis mass flow meter installed in the solution line. The absorption/ desorption rate is measured by the mass flow meter on the water line. The desorber solution outlet concentrati on ( d outX ) which is equal to the absorber solution inlet concentration (a inX ) is calculated using the measured density and temperature of the solution at the exit of the desorber . The desorber solution inlet concentrati on (d inX) which is equal to the absorber solution outlet concentration ( a outX ) is then calculated using a mass balance on the desorber or absorber . Notably, it is assumed that the loop has reached the steady state condi tion. d out des sol sol d inX m m m X (2 3 ) where d inX and d outX are the LiBr solution concentrations at the desorber inlet and outlet, respectively, is the solution mass flow rate measured at the desorber outlet, and desm is the desorption rate. The heat input to the desorber is calculated using:

PAGE 76

76 VI Qdes ( 2 4 ) where V and I are the applied voltage and current to the desorber heater, respectively. This heat input is the sum of heat loss, sensible heat and latent heat of vaporization. The energy balance in the desorber can be written as: loss d in sol v sol d out sol sol d v v desQ h m m h m h m Q , ,) ( ( 2 5 ) where hv is the enthalpy of water vapor leaving the desorber, d in solh, and d out solh, are the solution enthalpies at the desorber inlet and outlet, respectively, is the vapor generation rate in the desorber , and lossQ is the desorber heat loss. Using Eq. ( 2 5 ), vm can be calculated as follows ) /( ) ) ( (, , , d in sol d v loss d out sol d in sol sol vh h Q h h m VI m ( 2 6 ) Note that is different from the d esorption rate (desm ) measured by the water line flow meter. The vm and desm would have been the same, if no vapor exited the desorber through the solution line. The uncertainty in vm calculation is 0.3% and it is due to uncertainty in temperature, solution flow rate, and voltage and current measurements. Membrane Test Setup An experimental setup was fabricated to test the pressure drop of the membranes. A photograph of the set up is shown in Fig. 241. The test setup consists of a membrane test fixture (with an open area of 10 cm2), a mass flow controller (MKS instruments), a vacuum pump, absolute and differential pressure transducers, and needle valves. The vacuum pump was inst alled at the downstream of the test fixture to draw the test fluid through the lines and to perform subatmospheric tests.

PAGE 77

77 A schematic of the membrane test fixture is shown in Fig. 242. The test fixture is made of Stainless Steel and consists of an upper chamber, a lower chamber, and a sight glass. Doublesided adhesive tape (3M ) was used to secure the membrane on the lower chamber (cf. Fig. 242). The test fluid enters the upper chamber, flows across the membrane into the lower chamber, and exits the low er chamber through an outlet port. Figure 240 . A photograph of the test setup for membrane permeability measurements . Figure 241 . Cross sectional view of the test fixture. Mass Flow Controller Membrane T est F ixture Vacuum P ump Differe ntial Pressure Transducer Absolute P ressure Transducer Inlet Outlet Double Sided Tape Membrane

PAGE 78

78 The mass flow controller was calibrated by the manufacturer with an accuracy of 1% of reading. The pressure drop across the membrane was measured using two differential pressure transducers (Omega Engineering), one with a range of 0250 Pa and an accuracy of 0.25% of full scale and the other with a range of 017 kPa and an accuracy of 0.08%. The absolute pressure in the test chamber was measured using an absolute pressure transducer (Omega Engineering) with a range of 0108 kPa and an accuracy of 0.08%. The experimental uncertainty in measuring the pressure drop across the membrane is approximately 1%. Visualization Test setup Figure 2 43 shows a schematic of the device used to visualize bubble extraction process through the nanofibrous membrane utilized in t he desorber. The device consists of a polydimethylsiloxane (PDMS) chip on which flow channels are built. PDMS is transparent and enables visualization of the bubbles during the venting process through the membrane. Water and air are injected into the devic e at various flow rates through a Tjunction to create a twophase flow with a controlled quality. The water and air supply channels are 500 and 100 m wide, respectively. The main flow channel, where the bubble extraction process takes place, is 1000 m w ide and 200 m deep. The PDMS device is fabricated through a standard soft lithography technique [90] . The PTFE membrane was subsequently bonded on the device, as shown in Fig. 2 43. Figure 2 44 shows a diagram of the visualization test system. Two syringe pumps (Fisher Scientific Inc., PA) are used to deliver water and air to the test device. PX 26 pressure transducers (Omega Engineering, CT) with a range of 05 psi are used for measurements. Images of the bubbles during the extraction process are captured using a high speed Gazelle camera (Point Grey Research Inc., Canada, BC). The uncertainty

PAGE 79

79 in the visualization tests results from 1 ml/ hr inaccuracy in pumping rate of the syringe pump and 1% inaccuracy of the pressure sensors . Figure 242 . A 3D schematic of the visualization test device. Mhl Si P Figure 243 . A diagram of the vi sualization test system. Membrane PDMS device Gas Inlet Liquid Inlet T junction P P Membrane High speed camera Gas tight Syringe Water reservoir Water syringe pump Air syringe pump PDMS chip

PAGE 80

80 CHAPTER 3 MEMBRANE EVALUA TION Membrane Transport Membranes can separate species when subjected to a driving potential. The driving potential could be the gradient of temperature, pressure, concentration, or the electrical potential. Overall, the rate of transport through a membrane can be expressed based on the gradient of the chemical potential: dx d L Ji i i ( 3 1 ) where Ji is the permeate flux, Li is the coefficient of proportionally of permeate i , x is position along the membrane thickness, and i is the chemical potential defined in Eq. ( 3 2 ) for incompressible phases and in Eq. ( 3 3 ) for compressible gases [91] : ) ( ) (o i i i i o i iP P v c Ln RT (3 2 ) o i i i o i iP P Ln RT c Ln RT ) ( ( 3 3 ) where o i is the chemic al potential of component i at a reference pressure of o iP , ic is the molar concentration of component i , i is the activity coefficient, P is pressure, iv is the molar volume of component i , T is temperature, and R is the gas constant. Calculations of the permeate flux using Eq. ( 3 1 ) require further assumptions about variations of pressure and concentration (cf. Eq. ( 3 2) and (33 )). Among different transport models, solutiondiffusion model [92] is widely utilized for transport through membranes. This model assumes that the chemical potential only changes due to concentration variations across the membrane. Using this model, for the case in which the membrane confines a multi component liquid and only allows permeation of species

PAGE 81

81 i , a simple expression for permeate flux as a function of partial pressures is determined [91] : ) (, , i v i s G i iP P P J ( 3 4 ) where G iP is the permeability coefficient, i sP, is the partial vapor pressure of component i in the liquid, and i vP, is the partial pressure of component i in the vapor side. This equation suggests that the driving potential is the difference in the partial pressure of the permeate between the two sides of the membrane [16,17] . Membrane Permeability Preliminary studies showed that nanofibrous membranes have the highest porosity and permeability compare to other membranes. Fig. 3 1 provides scanning electron microscope (SEM) images of two different nanofibrous membranes. Depending on their pore size and manufacturing method, they can have porosities as high as 85%. The PTFE membranes selected for this application have good thermal and chemical stability. (a) (b) Figure 31 . SEM Micrographs of nanofibrous PTFE membr ane with different nominal

PAGE 82

82 Tests on the permeability of nanofibrous membranes were conducted at a flow rate of up to 0.01 kg m2 s1 and at two different absolute pressures 0.85 kPa and 10 kPa [34] . Fig. 3 2 provides the test results. The following correlation suggested by Knudsen [93] was used to fit the pressure drop versus flow rate data, since the flow is mainly in the Knudsen regi me. P RTM d Np 2 1 3 4 ( 3 5 ) As the results shown in Fig. 32 suggest, the estimated pressure drop values are in a good agreement with the experimental values. Water vapor pressure drop was then estimated using this correlation and plotted in Fig. 32 . The results suggest that a drop at a typical absorption rate of 0.002 kg/ m2s. m ( kg m-2s-1) P ( P a ) 0 0. 00 25 0. 005 0. 0 075 0 . 01 0. 0125 0 30 0 60 0 90 0 120 0 E xp e r i m e n t a l ( P = 0. 85 0 kP a ) E xp e r i m e n t a l ( P = 10 kP a ) P r e di c t e d f or N i t r o ge n g a s P r e di c t e d f or w a t e r va po r 5 m 0. 2 m 0. 4 5 m 1 m. Figure 32 . Pressure drop versus flow rate at 0.85 kPa and 10 kPa test pressures .

PAGE 83

83 and tested over the entire test period. Desorber made using membr harsher working environment (local pressure fluctuation caused by bubbles nucleation used for the desorber. As the desorption data discussed in the later sections shows, the membrane pressure drop is insignificant compare to the pressure difference applied across the membrane.

PAGE 84

84 CHAPTER 4 RESULTS AND DI SCUSSIONS Absorption Results Thin Fil m A bsorption The driving force in the absorption process is the difference in water pressure between the two phases (i.e. the vapor and the LiBr solution). Thus, a net increase in the pressure potential should increase the absorption rate regardless of the phase in which the water pressure is changed. A set of tests was conducted to verify this effect. First, the water vapor pressure was increased while the other test parameters were kept constant (at the nominal test values listed in Table 25 ). Fig. 4 1 shows the effect of water vapor pressure on the absorption rate at two different solution film thicknesses. The results clearly show that increasing the water vapor pressure (a vP) linearly increases the absorption rate. It should be noted that the absorption rate of 0.0057 kg/m2 approximately two and half times the rate reported in falling film studies (see Table 11 ). Furthermore, the results show that decreasing the solution film thickness increases the absorption rate by an average of approximately 35%. An increase in the absorption rate was expected in light of the findings of a preliminary numerical study reported in Yu et al. [14] . As depicted in Fig. 42 , reduction of the solution channel thickness impact s both the thermal and concentration fields. The thermal resistance between the solutionvapor interface, at which the water vapor heat of phas e change is released (cf. Fig. 4 2 a), and the cooling surface is a function of h/k ( h is the solution film thickn ess and k is the solution thermal conductivity). Thus, reducing the solution film thickness results in better cooling of the solutionvapor interface and a lower water

PAGE 85

85 pressure (cf. Fig. 4 3 ) in the solution phase. The lower water pressure increases the pr essure potential and the absorption rate. In addition, decreasing the solution film thickness, at a constant solution flow rate, reduces the concentration ( X ) boundary layer thickness (cf. Fig. 4 2 b and c ) and mass transfer resistance through the solution film. Further analysis of the impact of these two factors on the absorption rate will be provided later in this work when additional tests are discussed. Pa v( kP a ) ma b s( kg / m2s ) 0. 6 0. 9 1 . 2 1. 5 1. 8 2. 1 0 0. 00 2 0. 00 4 0. 00 6 0. 00 8 0 . 0 1 h= 10 0 m h= 16 0 m . Figure 41 . Variation of absorption rate as a function of water vapor pressure at hr kg ma sol/ 6 . 0 and C Ta cw 25 .

PAGE 86

86 Figure 42 . Schematics of the solution flow channel cross section (a) depicting heat release at the solution vapor interface and (b, c) illust rating the impact of change in solution fl ow channel thickness, at a constant flow rate, on the d) . T (oC ) Ps , w( kP a ) 2 4 2 7 3 0 3 3 3 6 0 0 . 3 0 . 6 0 . 9 1 . 2 X =5 4 % X =5 6 % X =5 8 % X =6 0 % Figure 43 . Solution water vapor pressure as a function of temperature and concentration (values are obtained using educationa l version of Engineering Equation Solver (EES) Software , McGraw Hill, Inc.) . 1 X i (c) (a) 2, , V X min sol (b) Water vapor heat of phase change h LiBr solution Cooling surface X i X i n 2 Xin 1, , V X min sol

PAGE 87

87 Effect of c ooling w ater t emperature In another test, to demonstrate that variations of the solution water vapor pressure (a w s,P ) impact the absorption rate, the solution water pressure was changed while the vapor pressure was kept constant. The solution water pressure varies with the solution temperature and concentration as shown in Fig. 43 . In this test, the change in the solution pressure was achieved by vary ing the solution channel wall temperature from 25 to 35 C in 2.5 C increments. All other test conditions were kept constant at the nominal conditions listed in Table 25 . The results (cf. Fig. 44 ) showed a linear decrease in absorption rate with an incr ease in solution temperature (i.e. increase in the solution water vapor pressure). Tc w(oC ) ma b s( kg / m2s ) 24 2 7 30 33 36 0 0. 00 2 0. 00 4 0. 00 6 h= 10 0 m h= 16 0 m . Figure 44 . Variation of absorption rate as a function of cooling water temperature at hr kg ma sol/ 6 . 0 and kPa Pa v1 . 1 .

PAGE 88

88 All abso rption rates reported in Figs. 41 and 44 are normalized with respect to the pressure potential ( aw s, a vP P ) and plotted in Fig. 45 . The solution water pressure used in the calculation is the average of the absorber inlet and exit. The results suggest that the absorption rate linearly increased with the pressure potential regardless of the source of change in pressure. Furthermore, the results suggest that the average difference in the absorption rate between the two solution film thi cknesses is approximately 40% when the absorption rate of the two cases is compared at the same water pressure potentials. This difference (40% versus 35%) more accurately represents the impact of change in solution film thickness on the absorption rate, b ecause a higher absorption rate in the case of the 100thick solution film resulted in a higher water concentration in the solution and a l ower pressure potential. Fig. 45 also compares the results with those of Ali and Schwerdt [16] . Their results showed almost no change in the absorption rate when the vapor pressure potential was increased. Ali and Schwerdt [16] argued that their membrane mass transport resistance could have dominated the overall mass transfer process. In the absence of any data on the pressure drop or the construct of their membrane, it is hard to evaluate their conclusion.

PAGE 89

89 Pa vPa s , w( k P a ) ma b s( k g / m2s ) 0. 3 0.6 0.9 1.2 1.5 0 0.00 2 0.00 4 0.00 6 0.00 8 0.0 1 E xpe r i m e n t a l D a t a ( h = 100 m ) E xpe r i m e n t a l D a t a ( h = 160 m ) A l i a n d S h w e r dt ( 20 09) . Fi gure 45 . Variation of absorption rate as a function of water vapor pressure potential . To determine the significance of the membrane resistance and its contribution to the overall mass transfer resistance, the membrane pressur e drop is separately measured using a test setup discussed earlier. Fig. 46 compares the membrane pressure drop, a i a vP P (a iPis pressure at the membranesolution interface), with the overall pressure potential (a w s, a vP P comparison suggests that membrane mass transfer resistance is not dominant (only 10 to 15% of the total resistance) in the arrangement implemented in this study. The dominant resistance here is mass transfer through the solution ( a w s, a iP P ).

PAGE 90

90 ma b s( kg / m2s ) P ( kP a ) 0 0. 002 0. 004 0. 006 0. 008 0 0. 3 0. 6 0. 9 1. 2 1. 5 Pa vPa s , wPa vPa iPa iPa s , w . Figure 46 . Comparison between membrane and solution resistances at a solution film . In light of the Yu et al. [14] studies in which solution films thicker than 200 m were found to be too t hick to enhance the absorption rate beyond that of the existing technology, we have considered the thick solution film used in the Ali and Schwerdt [16] studies as a major factor in the measured low absorption rate. Yu et al. [14] showed that thin solution films are cooled quite effectively and that the solution vapor interface temperature closely follows that of the bulk. They demonstrated that reducing the solution film thermal resistance allows for the removal of a high rat e of heat released at the solution vapor interface thus enabling significant enhancement of the absorption rate, as realized in this study. To better explain this effect, the absorption process in a thick solution film is numerically modeled. Yu et al. [14] provides the numerical scheme used in this sol ution. The numerical model simulates the absorption process in the 4-

PAGE 91

91 mmthick solution film implemented in the Ali and Schwerdt [16] study. Table 41 lists the model input values obtained from Ali and Schwerdt [16] . Table 4 1 . Parameters of the numerical simulation Parameters Values Film thickness 4 mm Membrane thickness Membrane pore size Membrane porosity 75% Solution inlet temperature 27 C Solution inlet concentration 54.3% Maximum channel length 7.6 mm Estimated solution velocity 0.006 m/s Figure 47 shows the temperature contours at absorption pressure potentials of 0.5 and 1.8 kPa. The simulation results show that in both cases the solution channel height is larger than the thermal boundary layer formed at the solutionvapor interface. In other words, the solution film is not effectively cooled and consequently the high vapor pressure of the hot solution at the interface reduces the absorption rate. The absorption rates determined by the numerical simulation are compared with Ali and Schwerdt’s [16] data in Fig. 48 . The results suggest that increasing the pressure potential has a moderate impact on the absorption rate at low pressure potentials (i.e. low heat release rate at the solutionvapor interface). The increase in absorption rate reaches a plateau at high pressure potentials. Interestingly, the numerical results match Ali and Schwerdt’s [16] data reasonably well at high pressure potentials. Ali and Schwerdt’s [16] data do not seem to be accurate at low pressure potentials, as absorption should fully diminish at zero pressure potential.

PAGE 92

92 Figure 47 . Temperature contours in a 4mm thick solution film at different pressure potentials . Pa vPa s , w( kP a ) ma b s( kg / m2s ) 0. 5 1 1. 5 2 2. 5 0 0. 00 1 0. 00 2 0. 00 3 0. 00 4 0. 00 5 A l i a nd S h w e r dt [ 16] N um e r i c a l s i m ul a t i on . Figure 48 . Absorption rate in a 4mmthick solution film as a function of pressure potential . Effect of s olution i nlet t emperature To further characterize performance of our membranebased absorber, the impact of solution inlet temperature on the absorption rate was studied at a solution film results are provided in Fig. 49 . The absorption rate slightly declined when the solution inlet temperature was increased. This decline in absorption a w s, a vP P =0.5 kPa 32 31 30 29 28 27 26 25 a w s, a vP P =1.8 kPa

PAGE 93

93 rate is due to an increase in the solution water pressure. The high heat transfer coefficient associated with the microchannel flow results in the rapid cool ing of the solution flow shortly after it enters the absorber channels. Ta i n(oC ) ma b s( kg / m2s ) 24 27 30 33 36 0 0. 00 1 0. 00 2 0. 00 3 0. 00 4 0. 00 5. Figure 49 . Variation of absorption rate as a function of solution inlet temperature at a , hr kg ma sol/ 6 . 0 , C Ta cw 25 and kPa Pa v1 . 1 . Effect of s olution v elocity As discussed earlier, increase in the solution flow velocity resulted from decreasing the solution film thickness was partially responsible for the observed enhancement in the absorption rate. In the last set of tests, the solution velocity was The solution flow rate is known to impact the absorption rate in conventional shell andtube absorbers [29,31,32] . For example, Yoon et al. [31] measured a 14% increase in the absorption rate by doubling

PAGE 94

94 the solution flow rate. The impact of the solution flow rate on the dynamics of falling film and thus on the absorption rate is quite complicated and certainly much different than in a membranebased absorber. The objective of this test was simply to characterize this aspect of the membranebased absorption process rather than comparing the two technologies. Unlike in the falling film absorption process, the velocity and thickness of a solution film can be independently changed when constrained by a membrane. For a give n solution flow condition (rate, temperature, and concentration) and a required total absorption, a membranebased absorber can be designed in a different width and length. The absorber width determines the flow cross section and thus the solution velocity . The first test was conducted at the nominal conditions listed in Table 25 and at a kg/hr in several steps. As the results shown in Fig. 410 suggest, the absorption rate increased moderately at first but then the rate of increase greatly dec lined at higher flow rates. Increasing the vapor pressure to 1.6 kPa significantly enhanced the rate of increase even at high flow rates.

PAGE 95

95 ma s o l( kg/ hr ) ma b s( kg / m 2 s ) 0. 4 0. 8 1. 2 1. 6 2 2. 4 0. 00 3 0. 00 4 0. 00 5 0. 00 6 0. 00 7 0. 00 8 Pa v= 1. 1 kP a Pa v= 1. 4 kP a Pa v= 1. 6 kP a . . Figure 410 . Variation of absorption rate as a function of solution flow rate at a sol u tion , C Ta cw 25 and kPa Pa v1 . 1 . It can be safely concluded that the fact that the absorption rate did not proportionally increase with the solution flow rate indicates that the water pressure of the exiti ng solution declined with the increase in flow rate. This results in a net increase in pressure potential, which is partially responsible for the enhancement of the absorption rate. To subtract this effect, the pressure potential for each data point was ca lculated by subtracting the average solution water pressure between the inlet and exit flow from the water vapor pressure. Using the pressure potential and the absorption rate, the absorption coefficient was calculated. The results in Fig. 411 clearly show that the absorption coefficient (i.e. the normalized absorption rate) is quite close at different test pressures. More importantly, the absorption coefficient at different pressures increased linearly with the solution flow rate (i.e. velocity) at an alm ost similar rate.

PAGE 96

96 Using this data, it can be argued that enhancement in the absorption rate due to the increase in solution velocity and better cooling of the solutionvapor interface are both responsible for the overall enhancement observed when the solut ion film thickness was reduced from ms o l( k g/ hr ) Kmx 106( kg / m2s P a ) 0. 4 0. 8 1 . 2 1. 6 2 2 . 4 0 2 4 6 8 Pv= 1. 1 k P a Pv= 1. 4 k P a Pv= 1. 6 k P a A ve r a ge . Figure 411 . Variation of absorption coefficient as a function of solution flow rate. The results discussed so far clearly showed that reducing the solution channel thickness and increasing the solution flow velocity result in increase in the absorption rate. However, these parameters cannot be varied without considering their impact on the pressure drop. As mentioned earlier, the absorber inlet and exit pressures are measured using t wo pressure transducers. Fig. 412 shows the pressure drop results as a function of solution flow rate at two different film thicknesses. As expected in a laminar flow, the pressure drop linearly increased with the solution velocity. The laminar fl ow theory also supports the significantly higher pressure drop measured in the case of 100-

PAGE 97

97 thick channel, since ( is solution velocity and is the channel hydraulic diameter). The theory predicts an approximately 4fold pressure drop difference between the two cases. ms o l( kg / h r ) Ps o l( kP a ) 0. 4 0. 8 1. 2 1. 6 2 2. 4 0 1 2 3 4 5 h= 1 00 m h= 1 60 m . Figure 412 . Solution pressure drop as a function of flow rate Absorption with Micromixing In the first test, the effect of water vapor pressure on the absorption rate is investigated. Increasing the water vapor pressure increases the pressure potential between the water vapor and LiBr solution and consequently increases the mass driving potential for the absorption process. The water vapor press ur e was increased from 800 to 1600 Pa. The other test conditions are the nominal values in Table 26 . Fig. 413 shows increasing of the absorption rate with the vapor pressure.

PAGE 98

98 Pa v( kP a ) ma b s( kg / m2s ) 0. 6 0. 9 1. 2 1. 5 1. 8 0 0. 00 2 0. 00 4 0. 00 6 0. 00 8. Figure 413 . Variation of absorption rate as a function of water vapor pressure at hr kg ma sol/ 5 . 2 and C Ta cw 25 . In the second test, the cooling water temperature was changed from 25 to 35C while the other test conditions were kept constant at the nominal conditions listed in T abl e 2 2 . Changing the cooling water temperature changes the solution temperature and consequently the solution water vapor pressure. This variation will ultimately lead to a change in the pressure potential and abs orption rate. As shown in Fig. 414, the abs orption rate decreases linearly with the cooling water temperature.

PAGE 99

99 Tc w(oC ) ma b s( kg / m2s ) 24 2 7 30 33 36 0 0. 00 2 0. 00 4 0. 00 6. Figure 414 . Variation of absorption rate as a function of cooling water temperature at hr kg ma sol/ 5 . 2 and kPa Pa v1 . 1 . As mentioned earlier, the driving force for the absorption process in the pressure potential between the vapor and LiBr solution. Therefore any comparison of the absorption rates should be conducted at the same pressure potential . Fig. 415 shows the absorption rates pr esented in Fig. 413 and Fig. 414 against the corresponding pressure potential (a w s, a vP P) . The solution water vapor pressure used in the calculation is the average of the absorber inlet and exit. Fig. 415 also provides the results obtained in falling film absorption studies and previous membranebased studies. To calculate the pressure potential for the test results reported by Medrano et al. [29] , Miller and Keyhani [26] and Yoon et al. [11] , the solution exit temperature was assumed, since it was not reported, to be equal to the cooling water exit temperature. Interestingly, the absorption rates achieved for the 500micron thick solution film with herringbone ridges

PAGE 100

100 is as high as absorption rates for a 100mic ron thin solution film at the absorber operating condition (i.e. kPa 6 . 0 ~ 5 . 0 P Pa w s, a v which corresponds to the evaporator temperature of about 4C and solution average concentration of 57%). The high achieved absorption rate stems from the mixing effect of vortices generated in the flow field. The mixing mechanism can be visualized by the generated concentration contours within the flow field. Fig. 4 16 shows the concentration contours in a single microchannel at different cross sections along the channel . The first top three pictures are at the entrance of the absorber and in the first set of micro ridges. As shown, the highly concentrated solution enters the channel at a uniform concentration. The water vapor absorption at the vapor/liquid interface decr eases the solution concentration on the top which is shown in blue color. At the same time, the imposed vortex in the channel attempts to pull the water rich solution down from the vapor liquid interface to the bottom of the channel (cf. the blue region gr owing downward on the left and right sides of the channel). The next three pictures show the concentrated contours in the second set of ridges where the ridges direction changes as shown in Fig. 2 12 and Fig. 2 13. The change in the direction is aimed to m ix the mass of the solution at the center of the vortex with the rest of the solution. The mechanism is evident at x=57mm, x=67mm where the water rich solution which has reached to the bottom of the channel starts to rise from the middle of the channel (i. e. the center of the previous vortex). In the last three pictures, the direction of the ridges again changes. This change reinforces the original vortex which drags the water rich solution down from top to the bottom of the channel. As it can be seen after 100 mm, the solution is well mixed as the solution bulk concentration is uniform. While it has been shown that in the absence of

PAGE 101

101 microstructures, after 200 mm, more than two third of the solution flow film at the bottom would leave the channels at about t he original solution concentration which means that a significant portion of the solution flow has not participated much in the absorption process [37] . Overall, micro ridges implementation along with direction alteration results in vortices which continuously replenish the concentrated solution at the membranesolut ion interface with the water rich solution at the bottom of the channel and mix them subsequently with the bulk of solution. ma b s( k g / m2s ) 0. 3 0 .6 0 .9 1 .2 1 .5 1 .8 0 0.00 2 0.00 4 0.00 6 0.00 8 0.0 1A l i a n d S h w e r dt ( 200 9) M e dr a n o e t a l .( 2002) M i l l e r a n d K e y h a n i ( 2001) I s l a m ( 2008) * Pa vPa s , w( k P a )ma b s( k g / m2s )0. 3 0 .6 0 .9 1 .2 1 .5 1 .8 0 0.00 2 0.00 4 0.00 6 0.00 8 0.0 1Y o o n e t a l .( 2 008) 160 m f i l m 100 m f i l m 500 m f i l m + m i c r o m i xi n g ++*. Figure 415 . Variation of absorption rate as a function of water vapor pressure potential Absorber pressure operating condition

PAGE 102

102 X= 0 mm X=13 mm X=33 mm X= 42 mm X= 57 mm X=67 mm X=70 mm X= 7 3 mm X= 100 mm Figure 4 16 . LiBr concentration contours at different cross sections between x=0 to 100 mm (the white color in the pictures shows the cross section of ridges). As mentioned earlier, absorption rates as high as that of a 100 micron s olution film were achieved at the absorber pressure potential condition. The importance is better recognized when the solution pressure drops are compared. Fig. 4 17 shows the solution pressure drop of the 500micron thick solution film with micro mixing a nd that of a 100micron thick solution film. The results suggest about two orders of magnitude the pressure drop decreases in the new approach compared to that of a 100micorn solution film which is consistent with the laminar flow theory. In a laminar flo w 3 h solD m P ( solm is solution flow rate and hD is the channel hydraulic diameter). The theory predicts an approximately 125fold pressure drop difference between the two cases at the same solution flow rate. It should be noted that the pressure drop is a key

PAGE 103

103 parameter to evaluate the energy consumption of the solution pump and eventually the operational costs of the system. The relatively low pressure drop of the new absorber is promising to design large scale membranebased absorbers. mc h a n n e l( kg/ hr ) P / L ( kP a / m ) 0 0 . 01 0. 02 0 . 0 3 0. 04 0. 0 5 0. 06 0. 0 7 0. 08 0 2 0 4 0 6 0 8 0 10 0 12 0100 m i c r on f i l m 160 m i c r on f i l m 500 m i c r on f i l m + m i c r om i xi ng . Figure 417 . Solution pressure drop as a function of flow rate To further characterize our absorber, the significance of the membrane resistance and its contribution to the overall mass transfer resistance is investigated in Fig. 4 18. The membrane pressure drop is separately measured using the membr ane test setup discussed earlier . Fig. 4 18 compares the membrane pressure drop, a i a vP P , with the overall pressure potential (a w s, a vP P) . The comparison suggests that membrane mass transfer resistance is not dominant (only 10 to 15% of the total resistance) in the arrangement implemented in this study. The dominant resistance here is mass transfer through the solution (a w s, a iP P) .

PAGE 104

104 ma b s( kg / m2s ) P ( kP a ) 0 0. 002 0. 004 0. 006 0. 008 0 0. 3 0. 6 0. 9 1. 2 1. 5 Pa vPa s , wPa vPa iPa iPa s , w . Figure 418 . Comparison between membrane and solution resistances Finally, the impact of solution inlet temperature on the absorption rate was studied. The results are provided in Fig. 4 19. The absorpt ion rate slightly declined when the solution inlet temperature was increased. This decline in absorption rate is due to an increase in the solution water pressure. The high heat transfer coefficient associated with the microchannel flow results in the rapi d cooling of the solution flow shortly after it enters the absorber channels.

PAGE 105

105 Ta i n(oC ) ma b s( kg / m2s ) 30 33 36 39 4 2 45 0 0. 00 2 0. 00 4 0. 00 6. Figure 419 . Variation of absorption rate as a function of solution inlet temperature at hr kg ma sol/ 5 . 2 , C Ta cw 25 and kPa Pa v1 . 1 Absorption with Ionic Liquid [EMIM][MeSO3] To test the absorption performance of ionic liquid [EMIM][MeSO3], absorption tests were conducted at different water vapor pressures. The other test parameters are kept constant (i.e. nominal test conditions listed Table 27 ). Fig. 4 20 shows the results. As elucidated before, the absorption rate is expected to increase with water vapor pressure due to a higher absorption driving potential.

PAGE 106

106 Pa v( kP a ) ma b s( kg / m2s ) 0. 6 0. 9 1. 2 1. 5 1. 8 0 0. 00 1 0. 00 2 0. 00 3. Figure 420 . Variation of absorption rate as a function of water vapor pressure hr kg ma sol/ 63 . 0 and C Ta cw 25 To evaluate the absorption performance of ionic liquid [EMIM][MeSO3], the results are compared with those of LiBr solution in the same absorber in Fig. 42 1 . Fig. 4 2 1 also provides the absorption data for conventional absorption technologies with LiBr solution as the desiccant for the comparison. To the best knowledge of the author, there is no absorption data for ionic liquid [EMIM][MeSO3] or any other ILs to compare with our results. Therefore these absorption results could be a starting point for future absorption studies on IL s. As illustrated in Fig. 4 21, the absorption r ates are much lower (i.e. less than one third) than those of LiBr solutio n in the same absorber configuration (cf. 500 micron film with micromixing). The reason for the lower absorption performance of IL could be: 1) much lower water diffusivity of [EMIM][MeSO3] compared to LiBr solution and 2) lower desiccant flow rate test condition in the case of

PAGE 107

107 [EMIM][MeSO3]. As mentioned before, the desiccant flow rate was kept low for IL case due to the high viscosity and therefore high pressure drop inside desiccant heat exchangers. Higher absorption rates are expected if the ridges desi gn is optimized for IL s and also higher desiccant flow rates could be managed. ma b s( k g / m2s ) 0. 3 0 .6 0 .9 1 .2 1 .5 1 .8 0 0.00 2 0.00 4 0.00 6 0.00 8 0.0 1A l i a n d S h w e r dt ( 200 9) M e dr a n o e t a l .( 2002) M i l l e r a n d K e y h a n i ( 2001) I s l a m ( 2008) [ E M I M ] [ M e S O3] * Pa vPa s , w( k P a )ma b s( k g / m2s )0. 3 0 .6 0 .9 1 .2 1 .5 1 .8 0 0.00 2 0.00 4 0.00 6 0.00 8 0.0 1Y o o n e t a l .( 2 008) 160 m f i l m 100 m f i l m 500 m f i l m + m i c r o m i xi n g ++*. Figure 421 . Performance of ionic liquid [EMIM][MeSO3] compared to that of LiBr solution Desorption Results Thin Film D esorption In the first se t of desorption tests, the effects of heated wall temperature on the desorpt ion process were studied. Fig. 422 shows the data sets and their corresponding interpolations. The wall temperature, Tw, is the average reading of the twelve thermocouples imbedded within the heated wall. The first test was conducted at 6 kPa vapor pressure. The other test parameters were kept constant at the nomi nal conditions

PAGE 108

108 listed in Table 28 . The surface temperature was increased in approximately 5C increments until desorpti on started. The first nonzero desorption rate was measured at a surface temperature of approximately 60 C. The desorption rate then steadily increased with surface temperature at a moderate pace until it started to significantly rise at a surface temperature of 95 to 100 C, signifying a change to the desorption regime. This change was associated with some fluctuations in solution flow rate and pressure readings due to instabilities associated with boiling of the solution flow. Tw(oC ) md e s( kg / m2s ) 45 6 0 75 90 10 5 120 0 0. 00 2 0. 00 4 0. 00 6 0. 00 8 0 . 0 1 0. 01 2 Pd v= 6 kP a Pd v= 6 kP a ( n um e r i c a l c a s e A ) Pd v= 6 kP a ( n um e r i c a l c a s e B ) Pd v= 10 k P a Pd v= 18 k P a . Figure 422 . Effect of heated wall temperature on desorption rate at different vapor pressures at hr kg md sol/ 5 . 2 and kPa Pd s23 . Using the thermodynamic properties of the LiBr solution and concentration at the desorber inlet, it was determined that the solution water vapor pressure ( d w s,P ) exceeded 6 kPa (i.e. the desorber vapor pressure) at a solution temperature of higher than 56 C. The positive pressure (or chemical) potential (i.e. d v d w s,P P >0) across the membrane Saturation temperature Direct diffusio n only Boiling

PAGE 109

109 drove the desorption process with an onset of desorption at about 60 C surface temperature (cf. Eq. ( 3 4 )). The driving pressure potential, and consequently the desorption rate, further enhanced by increasing the surface temperature. Hereafter, this desorption mechanism is called “direct diffusion” mode of desorption, since water molecules directly diffuse out of the thin solution film and subsequently flow through the membrane. Desorption rate through this mechanism diminished when the vapor pr essure was increased (cf. Fig. 4 22), due to decrease in pressure potential. The fact that the solution pressure (d sP= 23 kPa) was always higher than the vapor pressure clearly suggests that desorption through this mechanism i s not driven by the solution pressure. The solution pressure, as mentioned earlier, is the applied pressure on the liquid side of the membrane and is different from the solution water vapor pressure that is a thermodynamic property and a function of soluti on temperature and concentration. In essence, it is the difference between the solution water and vapor pressures that drives the water molecules through the solution film and the membrane. As the results in Fig. 422 suggest, desorption through the direct diffusion mode increased linearly with temperature at low desorption rates. This is consistent with the theory because as the solution water pressure increases with temperature, the pressure potential driving the process increases. However, the rate of increase significantly declined at higher desorption rates. To understand the cause of this behavior, a numerical model of the desorber was prepared. The numerical model simulates the desorber geometry explained earlier and uses the nominal exper im ental values listed in Table 28 as the model input, except that d vP =6 kPa. Briefly, the model is a 89mm long and 200m deep channel capped by a membrane with 0.45 m pore size and a

PAGE 110

110 thickness of 50 m supported by a metal sheet with 3.2 mm openings that are space 2.1 mm apart. The numerical method used to solve the heat and mass transfer field is discussed in Yu et al. [14] . The numerical method is based on a continuum based approach to model heat and mass transfer inside the solution and the Dusty Gas model [94] for simulation of the vapor flux through the membrane. The heat and mass transfer equations are solved using the Lattice Boltzmann Method (LBM). A comparison of the numerical and experiment al results is provided in Fig. 422. The most important observation is that the numeri cal simulation (case A in Fig. 422) closely captures the rate of change in the desorption rate and, as it will be discussed shortly, it was later used to successfully determine the factor responsible for the observed decline in the rate of increase in the desorption rate. The absolute difference between the two results could be mostly due to the difference between the solution properties and the equationof state used for the LiBr water solution (note that the solution used in the system contains additives that could impact its properties) as well as the surface temperature and the liquid film thickness. Using the numerical model, it was determined that small variations in the solution concentration entering the desorber are responsible for the observed behavior. Although efforts were made during the tests to maintain this parameter constant at 50%, the system stabi lized at a slightly different concentration in each test. This resulted in an overall variation in concentration of 3% (change from 48% to 51%). Fig. 4 22 also provides numerical results at a hypothetical constant solution inlet concentration of 48%. A com parison of numerical results between cases A and B (cf. Fig. 4 22) at a wall temperature of 83 C and a concentration differ ence of 3% is provided in Fig. 423. The comparison shows that a more water rich

PAGE 111

111 (i.e. less concentrated) solution in case B provides a higher solution water vapor pressure and consequently a higher desorption pressure potential compared to case A. As mentioned earlier, increasing the temperature a few degrees above the solution saturation temperature (87 to 90 C) significantly enhanc ed the rate of increase in desorption (cf. Fig. 422), due to boiling inception. Desorption through the boiling mode superimposed itself on the direct diffusion desorption mode and gradually dominated as the surface superheat temperature (i.e. the difference between the surface temperature and the solution saturation temperature) was increased. This is evidenced by the fact that the effect of vapor pressure on the desorption rate gradually declined to the extent that at a high superheat temperature, there i s hardly a difference between the desorption rates at different vapor pressures. This is due to the fact that formation of bubbles generates a significant vapor solution interface area within the microchannels resulting in a shorter diffusion path for the water molecules. The bubbles then vent through the membrane. The independence of bubbles venting rate form the pressure potential across the membrane suggests that the membrane mass transfer resistance is insignificant.

PAGE 112

112 .4 7 . 4 8 . 4 9 . 5 . 5 1 . 5 2 . 5 3 . 5 4 . 5 5 . 5 6 X x ( m m ) Ps , w , c a s e BPs , w , c a s e A( P a ) 0 15 30 45 60 75 90 20 0 40 0 60 0 80 0 100 0 120 0 d d Figure 423 . Comparison of numerical results showing a lower concentration solution (or a higher solution water vapor pressure) in case B compared to case A . A scale factor of 0.013 is used in the x direction to show the concentration contours over the entire flow domain (a). Water pressures used in (b) are liquid interface. Effect of s olution p ressure In the second set of tests, the solution pressure was changed from 13 k Pa to 30 kPa while the other test parameters were kept constant at the nominal cond itions. The Case A Case B (a) (b)

PAGE 113

113 results (cf. Fig. 4 24) reaffirmed that the effect of solution pressure on desorption through direct diffusion mechanism is insignificant. By contrast, changing the solution pressure significantly affects desorption through the boiling process. Increasing the solution pressure delays the transition to the boiling desorption mode. This is due to an increase in solution saturation temperature. It should be noted that the solution saturation temperature is a function of the solution pressure and concentration. Unlike a falling film desorber, wherein the solution saturation temperature remains unchanged at a constant vapor pressure, in a membranebased desorber, the s olution pressure can be controlled independently of the vapor pressure. Tw(oC ) md e s( kg / m2s ) 6 0 75 90 10 5 1 20 0 0. 00 2 0. 00 4 0. 00 6 0. 00 8 0 . 0 1 0. 01 2 Pd s= 13 k P a Pd s= 16 k P a Pd s= 23 k P a Pd s= 30 k P a . Figure 424 . Effect of solution pressure on desorption rate at hr kg md sol/ 5 . 2 and kPa Pd v10 . To filter out the effect of change i n saturation temperature, the boiling desorption data were plotted versus the wall superheat temperature (cf. Fig. 4 25). The results

PAGE 114

114 suggest that increasing the solution pressure enhances the desorption rate through the boiling mode. An increase in pressure is known to enhance the heat transfer coefficient and heat flux in pool boiling [95,96] and boiling in micro/minichannels [97,98] . Kocamustafaogullar i and Ishii [99] have shown that the boiling heat transfer coefficient in not only a function of the superheat temperature, but also a function of the active nucleation site density. An increase in pressure has been found t o enhance the size range of activate cavities [95,100] . Similar observations have been made in shell andtube desorbers. Lee et al. [44] investigated the effect of solution pressure on pool boiling characteristics of a lithium bromide solution on a vertical tube. They observed that increasing the solution pressure and s urface temperature significantly increases the number of active nucleation sites, reduces the bubble size, and enhances the desorption rate.

PAGE 115

115 Ts u p(oC ) md e s( kg / m2s ) 0 5 10 15 20 25 30 0 0. 00 2 0. 00 4 0. 00 6 0. 00 8 0 . 0 1 0. 01 2 Pd s= 13 k P a Pd s= 16 k P a Pd s= 23 k P a Pd s= 30 k P a . Figure 425 . Effect of solution pressure on the desorption rate in a boiling regime at hr kg md sol/ 5 . 2 and kPa Pd v10 . The results presented above are compared with those of Thorud et al. [53] on a device with 170 m thic k solution channels (cf. Fig. 4 26). Thorud et al. [53] repo rted a desorption rate an order of magnitude lower than that measured in this study. However, they did not discuss reasons for the lower desorption rate achieved even compared to falling film desorber results reported by Kim and Kim [48] , as shown in Fig. 426, which typically involves solution films thicker than 1 70 m [101] . Furthermore, it is not clear why their desorption rate versus superheat temperature graph does not exhibit characteristics associated with the boil ing desorption mode (i.e. rapid increase in desorption rate with a small increase in surface superheat temperature). The device configuration and mass transfer resistance of the membrane used in their study could be responsible for the low performance. As the schematic of Fig. 4 27 shows, in their

PAGE 116

116 assembly the solution film is heated on the same side that the generated vapor exits. It could be argued that since the vapor generated within the solution tends to accumulate at the membrane surface (because the membrane surface is hydrophobic), the thermal resistance between the heated surface and the solution increases when more vapor is generated. This could lead to a self limiting process in which the heat supply to the solution, and hence the desorption rate, could only be enhanced by a proportional increase in temperature potential (i.e. the surface superheat temperature). Thorud et al. [53] does not provide the membrane properties or a parametric study that identifies the membrane mass transport characteristics. Thus, it is not possible to attribute the membrane contribution to the poor performance. Ts u p(oC ) md e s( kg / m2s ) 0 5 10 15 20 25 30 0 0. 00 2 0. 00 4 0. 00 6 0. 00 8 0 . 0 1 0. 01 2 0. 01 4T h o r u d e t a l . ( 2 0 0 6 ) ( Pd s= 3 3 . 5 k P a / M e m b a s e d ) T h i s s t u d y ( Pd s= 3 0 k P a ) K i m a n d K i m ( 1 9 9 9 ) ( Pd s= 1 0 k P a / F a l l i n g f i l m ) T h i s s t u d y ( Pd s= 1 3 k P a ) . Figure 426 . Comparison of the measured desorption rate with those of other studies.

PAGE 117

117 Vapor Heat Figure 427 . Schematic of the desorber configuration implemented in a prior membranebased study Effect of s olution v elocity To study the effect of flow velocity on the desorption rate, the solution flow rate was changed from 0.75 kg/hr to 3.25 kg/hr. The results (cf. Fig. 4 28) show almost no effect on the direct diffusion desorption mode, because desorption through this mode is limited by the rate of water molecules diffusion through the solution film, which remains unchanged as long as the flow regime is laminar. However, in the boiling mode, the desorption rate increased by as much as about 50%, since the supply of water rich solution to the microchannels is enhanced. Since changes in the solution pressure and concentration result in change in saturation temperature, and consequently the desorption rate, the results can represent the effect of velocity alone only if the desorber’s average concentration and solution pressure remain constant. The average solution pressure was kept constant at the nominal condition. The average solution concentration at different test conditions is plotted in Fig. 42 9 . It can be seen that the average concentration ( Xavg) varies by nearly 1%. Therefore, the results presented in Fig. 42 9 represent the effect of velocity. LiBr solution Porous a luminum Membrane Solid wall

PAGE 118

118 Tw(oC ) md e s( kg / m2s ) 60 75 90 105 120 135 0 0. 00 2 0. 00 4 0. 00 6 0. 00 8 0 . 0 1 0. 01 2 md s o l= 0. 75 kg / h r md s o l= 1. 5 kg/ hr md s o l= 2. 5 kg/ hr md s o l= 3. 25 kg / h r . . . .. Figure 428 . Effect of solution flow rate on desorption rate at kPa Pd s23 and kPa Pd v10 .

PAGE 119

119 Tw(oC ) Xa v g( % ) 100 105 110 115 120 125 4 4 4 8 5 2 5 6 6 0 6 4 md s o l= 0. 75 k g/ hr md s o l= 2. 5 kg / h r md s o l= 1. 5 kg / h r md s o l= 3. 25 k g/ hr . . . . Figure 429 . Average concentration at different solution flow rate tests versus wall temperature. Effect of solution inlet t emperature The final test was conducted on the effect of solution inlet temper ature on the desorption rate. The solution inlet temperature was decreased from 70C to 50C while the other test conditions were kept constant at the nominal conditions. The r esults (cf. Fig. 430) showed that the effect of solution inlet temperature on desorption rate is minimal and most of the desorber surface area was still engaged in desorption. The high heat transfer coefficient within the microchannels and the absorption of the desorber vapor, which is at a higher pressure than the solution water pressure at the desorber inlet, result in rapid heating of the solution flow shortly after it enters the microchannels.

PAGE 120

120 Tw(oC ) md e s( kg / m2s ) 45 60 75 90 105 120 135 0 0. 00 2 0. 00 4 0. 00 6 0. 00 8 0 . 0 1 0. 01 2 Td i n= 50oC Td i n= 60oC Td i n= 70oC . Figure 430 . Effect of solution inlet temperature to the desorber on the desorption rate at kPa Pd v10 , hr kg md sol/ 5 . 2 and kPa Pd s23 . Analysis of B ubbles D ischarge As discussed earlier, in the boiling regime, bubbles form and subsequently exit the flow through the membrane. However, there is always a chance that some bubbles e xit the desorber along with the solution flow (i.e. bubbles do not get extracted through the membrane). This is especially true for bubbles generated near the end of microchannels. This effect negatively impacts the efficacy of the desorber and the system. The fraction of bubbles exiting the solution flow can be determined using the desorber energy balance. To do so, the desorber heat loss to the ambient was measured as a function of desorber temperature. To measure the heat loss, all valves connected to the desorber were first closed and then the desorber heater was energized to hold the desorber at different temperatures. The energy supplied to the

PAGE 121

121 desorber to maintain it at any temperature is considered to be its heat loss. The results of this test are pr ovided in Fig. 431. Tw(oC ) Ql o s s( w a t t s ) 60 75 90 105 120 135 0 8 1 6 2 4 3 2. Figure 431 . Desorber heat loss as a function of its wall temperature. Another fraction of the supplied heat to the desorber increases the solution temperature (i.e. turns into sensible heat). The sensible heat is calculated using the difference in enthalpies of the inlet and outlet solution flows. Fig. 432 compares the heat loss and sensible heat with the total heat supplied to the desorber at the nominal test conditions. The results suggested that the two terms account for a significant portion of the heat supplied to the desorber.

PAGE 122

122 Tw(oC ) Q ( W a t t s ) 75 90 105 120 0 5 0 10 0 15 0 20 0 25 0 30 0 Qd e sQs e n s i b l eQl o s s ... . Figure 432 . Comparison of desorber heat loss and solution sensible heat to desorber input heat. The heat consumed in water vapor generation i s equal to the total heat supplied to the desorber minus the heat loss and the sensible heat. This balance is reflected in Eq. ( 2 6 ), which is used to calculate the vapor generation rate. Fig. 433 compares the generated vapor rate and the vapor flow rate through the membrane (measured by the water line flow meter) at the nominal test conditions. Calculations were similarly conducted for tests at different solution pressures (cf. Fig.4 24). The results are provided in Fig. 434. As can be seen in Fi gs. 4 33 and 434, the vapor generation rate closely matches the vapor flow through the membrane at the direct diffusion desorption mode. However, the difference between the two reaches a maximum of 15% in the boiling regime. This means that less than 15% of the g enerated vapor were not extracted through the membrane (i.e. exited the desorber through the solution line). The

PAGE 123

123 results also suggest that the vapor flow through the solution line exit is not a function of the vapor flux through the membrane. This implies that the membrane mass transport resistance is not responsible for this phenomenon. Tw(oC ) m ( kg / m2s ) 75 90 105 120 0 0. 00 2 0. 00 4 0. 00 6 0. 00 8 0 . 0 1 md e smv ... Figure 433 . Comparison of vapor desorption and generation rates at nominal conditions.

PAGE 124

124 Tw(oC ) m ( kg / m2s ) 6 0 75 90 10 5 1 20 0 0. 00 2 0. 00 4 0. 00 6 0. 00 8 0 . 0 1 0. 01 2mde s(Pd s= 13 kP a ) mde s(Pd s= 16 kP a ) mde s(Pd s= 23 kP a ) mde s(Pd s= 30 kP a ) mv(Pd s= 13 kP a ) mv(Pd s= 16 kP a ) mv(Pd s= 23 kP a ) mv(Pd s= 30 kP a ) ... . . . . . . Figure 434 . Compari son of the vapor desorption and generation rates at different solution pressures. The vapor generation rate was also calculated for the results in Fig. 428 where the solution flow rate was changed. Fig. 435 provides the difference between the vapor gener ation and desorption rates (averaged for all tests) as a function of the solution flow rate. The corresponding solution mass flux ( solm ) was also calculated and is shown in the graph. The results suggest that increasing the solution flow rate enhances the vapor exit rate through the solution flow line. At a flow rate of 0.75 kg/hr, the vapor generation and extraction rates were almost equal. Increasing the flow rate from 2.5 kg/hr to 3.25 kg/hr almost doubled the bubble escape rate through the solution line (from about 0.0005 kg/m2s to 0.0011 kg/m2s).

PAGE 125

125 md s o l( k g/ h r ) mvmd e s( k g / m2s ) 0 1 2 3 4 0 0. 000 5 0. 00 1 0. 001 5"..lms o l( kg / m2s ) 100 5 0 75 0.2 5. Figure 435 . Comparison of vapor desorption and generation rates at different solution flow rates. To understand the physics of bubbles escape through the solut ion line, further studies were conducted using the visualization test setup discussed earlier. The differential pressure across the membrane was kept constant in all tests at 13 kPa to simulat e the test conditions in Fig. 435. Tests were conducted at diff erent mass flux. Fig. 436 shows the bubbles images during the venting process at different conditions. The x and t shown on the figures indicate distance from the device inlet and time. Fig. 436a shows test results at a water mass flux ( wm ) of 12 kg/m2s. At this condition, the bubbles fully vent through the membrane before moving much along the flow channel (about 8 mm). Increasing the flux to 41 kg/m2s only slightly delayed the bubble full extract ion, as it is evident in Fig. 436b. H owever, signs of a transition to a different

PAGE 126

126 regime appeared in which increased drag force on the bubble significantly changed its shape. Further increase of the flux to 54 kg/m2s indicated that while some bubbles still completely vent through the membrane, albeit farther from the channel inlet (cf. Fig 4 36c), a portion of the bubble snaps off from the main bubble (cf. Fig 436d) and continues and flows through the channel without venting through the membrane. This indicates that the bubble is no longer in contact with the membrane. Xu et al. [61] showed that at a critical velocity, a liquid film forms between the bubble and the membrane and prevents bubbles extraction through the membrane. The frequency of this event (i.e. fraction of bubble escape) was approximately 30% at this flow rate. Increasing the flux to 83 kg/m2s enhanced the frequency of the event to 50%. Finally, at a flux of 110 kg/m2s, almost all bubbles experienced the same phenomenon. The overall physics and trends of the observed phenomena apply well for flow of vapor in a LiBr solution, with small variations due to differences in fluid properties. Of particular interest is the bubble snapping events that occur when drag forces on a bubble exceed the surface tension forces. The surface tension of the LiBr solution at the boiling test conditions discussed earlier ranges from 0.0068 to 0.0076 N/m [102] that is shape drag force on a bubble is A C m A C V Fd d D 2 22 1 2 1 ( 4 1 ) where is liquid density, V is the flow velocity, Cd is the drag coefficient which is mainly a function of the bubble shape, A is the cross sectional area of the bubble and m is the mass flux. The LiBr solution density varies from 1450 to 1650 kg/m3 in our tests. Eq. (4 1 ) implies that a bubble experience a similar drag forces in LiBr solution and water

PAGE 127

127 flows as long as the LiBr solution flux is approximately 1.2 to 1.3 times higher than the water flux. (a) x=8mm t=0ms x=8mm t=112ms x=8mm t=120ms x=8mm t=139ms x=8mm t=154ms x=8mm t=163ms x=8mm t=196ms x=8mm t=263ms x=8mm t=422ms x=8mm t=700ms (b) x=10mm t=0ms x=10mm t=10ms x=10mm t=26ms x=10 mm t=29ms x=10mm t=40ms x=10mm t=43ms x=10mm t=45ms x=10mm t=76ms x=10mm t=682ms x=10mm t=900ms (c) X=12mm t=0ms x=12mm t=29ms x=12mm t=34ms x=12mm t=36ms x=12mm t=41ms x=12mm t=72ms x=12mm t=88ms x=12mm t =129ms x=12mm t=521ms x=12mm t=800ms (d) x=12mm t=0ms x=12mm t=10ms x=12mm t=17ms x=12mm t=52ms x=12mm t=55ms x=12mm t=57ms x=12mm t=62ms x=12mm t=79ms x=12mm t=174ms x=12mm t=500ms x=17mm t=154ms x=25mm t=272m s Figure 436 . Bubble extraction through the membrane with 13 kPa pressure difference across the membrane at three fluxes: (a) wm =12 kg/m2s (b) wm =41 kg/m2s (c d) wm =54 kg/m2s. Overall, the results suggest that the bubbles could be fully vented from the flow as long as they stay in contact with the membrane. To do so, the bubbles should experience a sufficient residence time within the channels. Obviously, the residence x

PAGE 128

128 time is the shortest for bubbles generated at the end of c hannels. Perhaps, the addition of a short adiabatic section at the desorber exit could resolve this issue. Finally, it is noteworthy that the existing boiling literature provides significant insights on dynamics of flow boiling and bubble growth in microc hannels. The existing knowledge on the flow boiling process and the insight provided here on characteristics of the membranebased flow boiling desorption process may be utilized to analyze the impact of microchannels geometrical parameters on dynamics of the LiBr flow boiling and bubble growth.

PAGE 129

129 CHAPTER 5 CONCLUSION The efficacy of highly porous nanofibrous membranes for application in an absorption heat pump was studied. The membranes were successfully implemented in an absorber and a desorber heat exchanger. It was determined that the absorption rate can be significantly enhanced when a solution film thickness on the order of 100 m and highly porous nanofibrous membranes are implemented. The studies were conducted at two solution film thicknesses o velocities. It was demonstrated that reducing the solution film thickness and increasing its velocity enhance the absorption rate. The absorption rate achieved was significantly higher than in a previous study and approximately 2.5 times that of the existing falling film absorption technology. The absorption rate linearly increased with the pressure potential (i.e. the difference between the water pressure in the vapor and solution phases). The membrane mass tr ansfer resistance was not dominant at the absorption rates measured in this study. However, further enhancement of the absorption rate to and beyond 0.01 kg/m2.s at low pressure potentials could result in a significant increase in contribution of the membr ane mass transfer resistance. This study also investigated the effect of micro surface structures on the absorption characteristics of a membranebased absorber. The microstructures should have enough height to generate surface vortices with sufficient momentum to impact the main flow and continuously replenish the solutionmembrane interface with a concentrated solution. It was determined that the absorption rate can be significantly enhanced and absorption rates are comparable with those of thin solution film .

PAGE 130

130 Implementation of surface microstructures enables increasing the solution channel height and consequently decreasing the solution pressure drop significantly . Overall, the successful achievement of high absorption rates with relatively low solution pressure drop in this study suggests that large membranebased absorber heat exchangers could be developed. The absorption performance of Ionic liquid [EMIM][MeSO3] was also reported in a membrane based absorber with micro surface struct ures . To the best knowledge of the author, results are the first absorption data reported for ionic liquids. As expected, the absorption rates are much lower than those of LiBr solution due to the much lower water diffusivity of IL s compared to LiBr solution. However, the application of IL s as an alternative desiccant for LiBr solution in ARSs is interesting because: 1) IL s do not crystallize and 2) IL s are less corrosive. The crystallization and corrosion are big challenges in LiBr absorption chillers. Furthermore, a par ametric study was conducted to understand characteristics of the water desorption process from a thin LiBr solution flow constrained by a porous hydrophobic membrane. Two modes of desorption consisting of: (1) direct diffusion of water molecules out of the solution film and their subsequent flow through the membrane and (2) formation of water vapor bubbles within the solution flow and their exit through the membrane were observed and analyzed. The vapor pressure determined the onset of direct diffusion deso rption while the solution pressure determined the onset of the boiling desorption mechanism. The desorption rate increased moderately with temperature in the direct desorption mode and exponentially in the boiling desorption mode. Lowering the vapor pressure or elevating the water

PAGE 131

131 pressure inside the solution enhanced desorption through the direct diffusion mechanism while the effect of solution velocity was negligible on the same. In the boiling desorption mode, increasing the solution pressure and velocit y enhanced desorption at a fixed wall superheat temperature. Comparison of the vapor mass flow rate through the membrane and the solution exit line showed that the ratio of the two is independent of the desorption rate implying that the membrane mass transfer resistance did not limit desorption through the membrane. However, the solution flow velocity directly affected the rate of bubble exit through the solution line. The bubbles exit rate was near zero at 0.75 kg/hr flow rate and increased to a maximum of 20% of the total vapor generation at 3.25 kg/m2s mass flux. Visualization studies suggested that beyond a critical mass flux (i.e. flow velocity) some bubbles cease to extract through the membrane. To avoid this phenomenon, a membranebased desorber can be designed to a width and length, for a given solution flow condition, to avoid the critical mass flux. Below the critical condition, adding an adiabatic section at the end of the microchannels is expected to enable full bubble extraction, as this approac h increases the residence time of bubbles generated towards the end of the heated section. The successful demonstration of a membranebased absorption and desorption process es in the set of tests presented in this study suggests that compact membranebased absorber and desorber heat exchangers in a plate and frame configuration could be developed. This heat exchanger configuration is inherently more compact than the shell andtube heat exchangers and lends itself to small scale, low capacity ARS d esigns.

PAGE 132

132 LIST OF REFERENCES [1] A. Sakoda, M. Suzuki, Fundamental study on solar powered adsorption cooling system, J. Chem. Eng. Japan. 17 (1984) 52– 57. [2] J. Deng, R.Z. Wang, G.Y. Han, A review of thermally activated cooling technologies for combined cooling, heating and power systems, Prog. Energy Combust. Sci. 37 (2011) 172– 203. [3] P. Srikhirin, S. Aphornratana, S. Chungpaibulpatana, A review of absorption refrigeration technologies, Renew. Sustain. Energy Rev. 5 (2001) 343– 372. [4] T. Hendricks, V.H. Johnson, M.A. Keyser, Heat Generated cooling opportunities, (2004) 13– 15. [5] H.T.C. X. Wang, Absorption cooling: a review of lithium bromide– water chiller technologies, Recent Patents Mech. Eng. 2 (2009) 193– 213. [6] R.A. Zogg, M.Y. Feng, D. Westphalen, Guide to developing air cooled LiBr absorption for combined heat and power applications, 2005. [7] E. Herold, R. Radermacher, S.A. Klein, Absorption chillers and heat pumps, CRC Press, Boca Raton, FL, 1996. [8] DOEWas te Heat Recovery: Technology and Opportunities in U.S. Industr, 2008. [9] K.D. Rafferty, Absorption refrigeration, Klamath Falls, OR, 1998. [10] M. Raisul Islam, N.E. Wijeysundera, J.C. Ho, Performance study of a fallingfilm absorber with a film inverting configuration, Int. J. Refrig. 26 (2003) 909– 917. [11] J. I. Yoon, O. K. Kwon, P.K. Bansal, C. G. Moon, H. S. Lee, Heat and mass transfer characteristics of a small helical absorber, Appl. Therm. Eng. 26 (2006) 186– 192. [12] M.D. Determan, S. Garimella, A mmonia– water desorption heat and mass transfer in microchannel devices, Int. J. Refrig. 34 (2011) 1197– 1208. [13] S. Jeong, S. Garimella, Fallingfilm and droplet mode heat and mass transfer in a horizontal tube LiBr/water absorber, Int. J. Heat Mass Trans f. 45 (2002) 1445– 1458. [14] D. Yu, J. Chung, S. Moghaddam, Parametric study of water vapor absorption into a constrained thin film of lithium bromide solution, Int. J. Heat Mass Transf. 55 (2012) 5687– 5695.

PAGE 133

133 [15] R. Nasr Isfahani, S. Moghaddam, Absorption characteristics of lithium bromide (LiBr) solution constrained by superhydrophobic nanofibrous structures, Int. J. Heat Mass Transf. 63 (2013) 82– 90. [16] A.H.H. Ali, P. Schwerdt, Characteristics of the membrane utilized in a compact absorber for lithium bromide– water absorption chillers, Int. J. Refrig. 32 (2009) 1886– 1896. [17] A.H.H. Ali, Design of a compact absorber with a hydrophobic membrane contactor at the liquid – vapor interface for lithium bromide– water absorption chillers, Appl. Energy. 87 (2010) 1112– 1121. [18] Y. Tae Kang, A. Akisawa, T. Kashiwagi, Analytical investigation of two different absorption modes – 443. [19] G.S. Herbine, H. Perez Blanco, Model of an Ammonia– Water Bubble Absor ber, ASHRAE Trans. 101 (1995) 1324 – 1332. [20] E. Palacios, M. Izquierdo, J.D. Marcos, R. Lizarte, Evaluation of mass absorption in LiBr flat fan sheets, Appl. Energy. 86 (2009) 2574– 2582. [21] J. I. Yoon, T. T. Phan, C. G. Moon, P. Bansal, Numerical study on heat and mass transfer characteristic of plate absorber, Appl. Therm. Eng. 25 (2005) 2219– 2235. [22] S.K. Choudhury, D. Hisajima, T. Ohuchi, A. Nishiguchi, T. Fukushima, S. Sakaguchi, Absorption of vapors into liquid films flowing over cooled horizontal tubes, ASHRAE Trans. Res. 99 (1993) 81– 89. [23] P. Sultana, N.E. Wijeysundera, J.C. Ho, C. Yap, Modeling of horizontal tubebundle absorbers of absorption cooling systems, Int. J. Refrig. 30 (2007) 709– 723. [24] M.R. Islam, Absorption process of a falling film on a tubular absorber: An experimental and numerical study, Appl. Therm. Eng. 28 (2008) 1386– 1394. [25] M. Kiyota, I. Morioka, K. Asahara, Steam absorption into films of lithium bromide solution falling over horizontal pipes arranged in a vertical column, Heat Transf. Res. 30 (2001) 451– 462. [26] W.A. Miller, M. Keyhani, The correlation of simultaneous heat and mass transfer experimental data for aqueous lithium bromide vertical falling film absorption, J. Sol. Energy Eng. 123 (2001) 30– 42. [27] S. Bo , X. Ma, Z. Lan, J. Chen, H. Chen, Numerical simulation on the falling film absorption process in a counter flow absorber, Chem. Eng. J. 156 (2010) 607– 612.

PAGE 134

134 [28] V. Patnaik, H. Perez Blanco, W.A. Ryan, A simple analytical model for the design of vertical t ube absorbers, ASHRAE Trans. 99 (1993) 69– 80. [29] M. Medrano, M. Bourouis, A. Coronas, Absorption of water vapour in the falling film of water – lithium bromide inside a vertical tube at air cooling thermal conditions, Int. J. Therm. Sci. 41 (2002) 891– 898. [30] S. Karami, B. Farhanieh, A numerical study on the absorption of water vapor into a film of aqueous LiBr falling along a vertical plate, Heat Mass Transf. 46 (2009) 197– 207. [31] J. I. Yoon, T.T. Phan, C. G. Moon, H.S. Lee, S. K. Jeong, Heat and mass transfer characteristics of a horizontal tube falling film absorber with small diameter tubes, Heat Mass Transf. 44 (2008) 437– 444. [32] A. Matsuda, K.H. Choi, K. Hada, T. Kawamura, Effect of pressure and concentration on performance of a vertical falli ngfilm type of absorber and generator using bromide aqueous solutions, Int. J. Refrig. 17 (1994) 538– 542. [33] V. Subramaniam, S. Garimella, From measurements of hydrodynamics to computation of species transport in falling films, Int. J. Refrig. 32 (2009) 607– 626. [34] R. Nasr Isfahani, K. Sampath, S. Moghaddam, Nanofibrous membranebased absorption refrigeration system, Int. J. Refrig. 36 (2013) 2297– 2307. [35] R. Nasr Isfahani, S. Moghaddam, Experimental study of water vapor absorption into Lithium Bromi de (LiBr) solution constrained by superhydrophobic porous membranes, in: ASME 2013 Heat Transf. Summer Conf., Minneapolis, Minnesota, USA, 2013. [36] R. Nasr Isfahani, S. Moghaddam, Absorption Characteristics of Thin Lithium Bromide (LiBr) Solution Film Co nstrained by a Porous Hydrophobic Membrane, in: ASME 2013 11th Int. Conf. Nanochannels, Microchannels, Minichannels, Sapporo, Japan, 2013. [37] S. Bigham, D. Yu, D. Chugh, S. Moghaddam, Beyond the limits of mass transfer in liquid absorbent microfilms through implementation of surfaceinduced vortices, Energy. 65 (2014) 621– 630. [38] A.P. Sudarsan, V.M. Ugaz, Multivortex micromixing., Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 7228– 33. [39] J.M. Ottino, S. Wiggins, Designing optimal micromixers., Science. 305 (2004) 485– 6.

PAGE 135

135 [40] C. Xi, D.L. Marks, D.S. Parikh, L. Raskin, S. a Boppart, Structural and functional imaging of 3D microfluidic mixers using optical coherence tomography., Proc. Natl. Acad. Sci. U. S. A. 101 (2004) 7516– 21. [41] A.D. Stroock, S.K.W. De rtinger, A. Ajdari, I. Mezic, H. a Stone, G.M. Whitesides, Chaotic mixer for microchannels., Science. 295 (2002) 647– 51. [42] S. Bigham, R. Nasr Isfahani, S. Moghaddam, Direct molecular diffusion and micro mixing for rapid dewatering of LiBr solution, Appl . Therm. Eng. 64 (2014) 371– 375. [43] W.W.S. Charters, V.R. Megler, W.D. Chen, Y.F. Wang, Atmospheric and subatmospheric boiling of H2O and LiBr/H2O solutions, Int. J. Refrig. 5 (1982) 107– 114. [44] C.C. Lee, Y.K. Chuah, D.C. Lu, H.Y. Chao, Experimental i nvestigation of pool boiling of lithium bromide solution on a vertical tube under subatmospheric pressurs, Int. Commun. Heat Mass Transf. 18 (1991) 309– 320. [45] H.K. Varma, R.K. Mehrotra, K.N. Agrawal, U.P. Roorkee, S. Singh, Heat transfer during pool boi ling of LiBr water solutions at subatmospheric pressures, Int. Commun. Heat Mass Transf. 21 (1994) 539– 548. [46] A. Matsuda, K. Hada, T. Kawamura, A vertical type of absorber and generator for lirhium bromide aqueous solutions, Rerigeration. 7 (1990) 47– 56. [47] A. Matsuda, T. Kawamura, K. Hada, Evaporation for lithium bromide aqueous soluion in a vertical falling film type of generator under reduced pressure, Refrigeration. 7 (1990) 35– 45. [48] D. Kim, M. Kim, Heat transfer enhancement characteristics for falling film evaporation on horizental enhanced tubes with aqueous LiBr solution, Enhanc. Heat Transf. 6 (1999) 61– 69. [49] T. Fujita, Falling liquid films in absorption machines, Int. J. Refrig. 16 (1993) 282– 294. [50] H. Yoshitomi, H., Tajima, O., Takeuc hi, T., Hara , Refrigeration. 56 (1981) 271. [51] V.Y. Nakoryakov, N.I. Grigorveva, S.I. Lezhnin, P.L. V, Combined heat and mass transfer in film absorption and bubble desorption, Heat Transf. Res. 29 (1998) 333– 338. [52] C. Shi, Q. Chen, T. C. Jen, W. Yang, Heat transfer performance of lithium bromide solution in falling film generator, Int. J. Heat Mass Transf. 53 (2010) 3372– 3376.

PAGE 136

136 [53] J.D. Thorud, J.A. Liburdy, D. V. Pence, Microchannel membrane separation applied to confined thin film desorption, Exp. Therm. Fluid Sci. 30 (2006) 713– 723. [54] D.D. Meng, J. Kim, C. J. Kim, A degassing plate with hydrophobic bubble capture and distributed venting for microfluidic devices, J. Micromechanics Microengineering. 16 (2006) 419– 424. [55] M. Johnson, G. Liddiard, M. Eddings, B. Gale, Bubble inclusion and removal using PDMS membranebased gas permeation for applications in pumping, valving and mixing in microfluidic devices, J. Micromechanics Microengineering. 19 (2009) 95011. [56] C. Lochovsky, S. Yasotharan, A. Gnther, Bubbles no more: inplane trapping and removal of bubbles in microfluidic devices, Lab Chip. 12 (2012) 595– 601. [57] A. Kamitani, S. Morishita, H. Kotaki, S. Arscott, Improved fuel use efficiency in microchannel direct methanol fuel cells using a hydrophilic macroporous layer, J. Power Sources. 187 (2009) 148– 155. [58] X. Zhu, Micro/nanoporous membrane based gas – wat er separation in microchannel, Microsyst. Technol. 15 (2009) 1459– 1465. [59] M.P. David., J. Steinbrenner, J. Miler, K.E. Goodson, Visualization and analysis of venting from a single microchannel twophase copper heat exchanger, ASME. (2009). [60] M.P. Dav id, J.E. Steinbrenner, J. Miler, K.E. Goodson, Adiabatic and diabatic twophase venting flow in a microchannel, Int. J. Multiph. Flow. 37 (2011) 1135– 1146. [61] J. Xu, R. Vaillant, D. Attinger, Use of a porous membrane for gas bubble removal in microfluidi c channels: physical mechanisms and design criteria, Microfluid. Nanofluidics. 21 (2010) 539– 548. [62] R. Nasr Isfahani, A. Fazeli, S. Bigham, S. Moghaddam, Physics of lithium bromide (LiBr) solution dewatering through vapor venting membranes, Int. J. Mult iph. Flow. 58 (2014) 27– 38. [63] R. Nasr Isfahani, S. Moghaddam, Physics of MembraneBased Desorption Process From LiBr Solution Flow in Microchannels, in: ASME 2013 Heat Transf. Summer Conf., Minneapolis, USA, 2013. [64] R. Nasr Isfahani, S. Moghaddam, Physics of MembraneBased Phase Separation in Flow Boiling of a Binary Mixture, in: ASME 2013 11th Int. Conf. Nanochannels, Microchannels, Minichannels, Sapporo, Japan, 2013.

PAGE 137

137 [65] L.E. Ficke, R.R. Novak, J.F. Brennecke, Thermodynamic and Thermophysical Properties of Ionic Liquid + Water Systems, J. Chem. Eng. Data. 55 (2010) 4946– 4950. [66] L.E. Ficke, Thermodynamic properties of imidazolium and phosphonium based ionic liquid mixtures with water or carbon dioxide, 2010. [67] P. Wasserscheid, M. Seiler, Leveraging gigawatt potentials by smart heat pump technologies using ionic liquids., ChemSusChem. 4 (2011) 459– 63. [68] M. Seiler, A. Ku, F. Ziegler, X. Wang, Sustainable Cooling Strategies Using New Chemical System Solutions, Ind. Eng. Chem. Res. 52 (2013) 1651 9 – 16546. [69] K. S. Kim, B. K. Shin, H. Lee, F. Ziegler, Refractive index and heat capacity of 1butyl 3 methylimidazolium bromide and 1butyl 3 methylimidazolium tetrafluoroborate, and vapor pressure of binary systems for 1 butyl 3 methylimidazolium bromi de + trifluoroethanol and 1butyl 3 methylimidazolium te, Fluid Phase Equilib. 218 (2004) 215– 220. [70] Y. Nakata, K. Kohara, K. Matsumoto, R. Hagiwara, Thermal Properties of Ionic Liquid + Water Binary Systems Applied to Heat Pipes, J. Chem. Eng. Data. 56 (2011) 1840– 1846. [71] L.E. Ficke, J.F. Brennecke, Interactions of Ionic Liquids and Water, J. Phys. Chem. B. 114 (2010) 10496– 10501. [72] J. F. Wang, C. X. Li, Z. H. Wang, Z. J. Li, Y. B. Jiang, Vapor pressure measurement for water, methanol, ethanol, and their binary mixtures in the presence of an ionic liquid 1ethyl 3 methylimidazolium dimethylphosphate, Fluid Phase Equilib. 255 (2007) 186– 192. [73] J. Ren, Z. Zhao, X. Zhang, Vapor pressures, excess enthalpies, and specific heat capacities of the binar y working pairs containing the ionic liquid 1ethyl 3 methylimidazolium dimethylphosphate, J. Chem. Thermodyn. 43 (2011) 576– 583. [74] Z. He, Z. Zhao, X. Zhang, H. Feng, Thermodynamic properties of new heat pump working pairs: 1,3 Dimethylimidazolium dimet hylphosphate and water, ethanol and methanol, Fluid Phase Equilib. 298 (2010) 83– 91. [75] X. Zhang, D. Hu, Performance simulation of the absorption chiller using water and ionic liquid 1 ethyl 3 methylimidazolium dimethylphosphate as the working pair, Appl . Therm. Eng. 31 (2011) 3316– 3321. [76] X. Zhang, D. Hu, Performance analysis of the singlestage absorption heat transformer using a new working pair composed of ionic liquid and water, Appl. Therm. Eng. 37 (2012) 129– 135.

PAGE 138

138 [77] M. Shiflett, A. Yokozeki, A bsorption cycle utilizing ionic liquid as working fluid, (2006). [78] M.C. Schneider, R. Schneider, O. Zehnacker, O. Buchin, F. Cudok, A. Khn, et al., Ionic Liquids: New highperformance working fluids for absorption chillers and heat pumps, in: Proc. Int . Sorption Heat Pump Conf., 2011. [79] M. Radpieler, C. Scweigler, Experimental investigation of ionic liquid EMIM EtSO4 as solvent in a single effect cycle with adiabatic absorption and desorption, in: Proc. Int. Sorption Heat Pump Conf., 2011: pp. 6– 8. [ 80] X. Xu, S. Shao, Y. Luo, C. Tian, Feasibility of ionic liquid in solar driven liqid desiccant dehumidification system for air conditioning, in: Int. Conf. Sol. Air Cond., 2011: pp. 12– 14. [81] D. Glebov, F. Setterwall, Experimental study of heat transfer additive influence on the absorption chiller performance, Int. J. Refrig. 25 (2002) 538– 545. [82] A. Kuhn, O. Buchin, M. Seiler, P. Schwab, F. Ziegler, Ionic liquids a promising solution for solar absorption chillers, in: Proc. 3rd Int. Conf. Sol. Air Cond., 2009. [83] G. Zuo, Z. Zhao, S. Yan, X. Zhang, Thermodynamic properties of a new working pair: 1Ethyl3 methylimidazolium ethylsulfate and water, Chem. Eng. J. 156 (2010) 613– 617. [84] ASHRAE Handbook of Fundamentals, Atlanta, 1997. [85] G. a. Floride s, S. a. Kalogirou, S. a. Tassou, L.C. Wrobel, Design and construction of a LiBr – water absorption machine, Energy Convers. Manag. 44 (2003) 2483– 2508. [86] B. Hasse, J. Lehmann, D. Assenbaum, P. Wasserscheid, A. Leipertz, A.P. Frba, Viscosity, Interfacial Tension, Density, and Refractive Index of Ionic Liquids [EMIM][MeSO3], [EMIM][MeOHPO2], [EMIM][OcSO4], and [BBIM][NTf2] in Dependence on Temperature at Atmospheric Pressure, J. Chem. Eng. Data. 54 (2009) 2576– 2583. [87] C.M. Tenney, M. Massel, J.M. Mayes , M. Sen, J.F. Brennecke, E.J. Maginn, A Computational and Experimental Study of the Heat Transfer Properties of Nine Different Ionic Liquids, J. Chem. Eng. Data. 59 (2014) 391– 399. [88] K.J. Kim, N.S. Berman, D.S.C. Chau, B.D. Wood, absorption of water va pour into falling films of aqueous lithium bromide, Int. J. Refrig. 18 (2000) 486– 494.

PAGE 139

139 [89] M.H. Rausch, J. Lehmann, A. Leipertz, A.P. Frba, Mutual diffusion in binary mixtures of ionic liquids and molecular liquids by dynamic light scattering (DLS)., Phy s. Chem. Chem. Phys. 13 (2011) 9525– 33. [90] D. Qin, Y. Xia, G.M. Whitesides, Soft lithography for micro and nanoscale patterning, Nat. Protoc. 5 (2010) 491– 502. [91] J.G. Wijmans, R.W. Baker, The solutiondiffusion model: a review, J. Memb. Sci. 107 (1995) 1 – 21. [92] H.K. Lonsdale, U. Merten, R.L. Riley, Transport properties of cellulose acetate osmotic membranes, J. Appl. Polym. Sci. 9 (1965) 1341– 1362. [93] M. Knudsen, Die gesetze der molekularstromung and der inneren reibungsstromung der gase durch rohr en, Ann. Phys. (1909) 75– 130. [94] E.A. Mason, A.P. Malinauskas, Gas transport in porous media gas model, Elsevier Scientific Pub., New York, 1983. [95] K. Nishikawa, Y. Fujita, H. Ohta, S. Hidaka, Effects of system pressure and surface roughness on nucleate boiling heat transfer, Mem. Fac. Eng. Kyushu Univ. 42 (1982) 95– 111. [96] N. Abuaf, S.H. Black, F.W. Staub, Pool boiling performance of finned surfaces in R 113, Int. J. Heat Fluid Flow. 6 (1985) 23– 30. [97] K.. Rainey, S.. You, S. Lee, Effe ct of pressure, subcooling, and dissolved gas on pool boiling heat transfer from microporous, square pinfinned surfaces in FC 72, Int. J. Heat Mass Transf. 46 (2003) 23– 35. [98] X. Hu, G. Lin, Y. Cai, D. Wen, Experimental study of flow boiling of FC 72 in parallel minichannels under subatmospheric pressure, Appl. Therm. Eng. 31 (2011) 3839– 3853. [99] G. Kocamustafaogullari, M. Ishii, Interfacial area and nucleation site density in boiling systems, Int. J. Heat Mass Transf. 26 (1983) 1377– 1387. [100] T. Hibiki, M. Ishii, Active nucleation site density in boiling systems, Int. J. Heat Mass Transf. 46 (2003) 2587– 2601. [101] S. Jani, M.H. Saidi, a. a. Mozaffari, Second Law Based Optimization of Falling Film Single Tube Absorption Generator, J. Heat Transfer. 126 (2004) 708. [102] S. Hasaba, T. Uemura, H. Narita , Refrigeration. 36 (1961) 622.

PAGE 140

140 BIOGRAPHICAL SKETCH Rasool was born in Esfahan, Iran in 1985. He received his bachelor of Mechanical Engineering from Univer sity of Tehran in September of 2007. During his undergrad study, he was involved in the research concerning the effects of magnetic fields on delaying flow separation in diverging flows of viscoelastic fluids . Rasool pursued his education in University of Tehran and he received his Master degree in Mechanical Engineering in 2010. The main focus of his M.Sc. research was the optimal control of temperature profile within glasswork considering thermal radiation effects. After his M.Sc., Rasool got PhD admission from University of Florida in 2010. Throughout his PhD, Rasool was working on development of a new generation of absorption chillers in Dr. Saeed Moghaddam’s group. Rasool defended his PhD dissertation in June 2014 and obtained his PhD in Mechanical Engineering in August 2014.