ABSTRACT Title of Document: ENERGY SAVINGS AND THERMAL COMFORT OF SEPARATE SENSIBLE AND LATENT COOLING AIR-CONDITIONING SYSTEMS Jiazhen Ling, Doctor of Philosophy, 2011 Directed By: Professor, Reinhard Radermacher, Mechanical Engineering Conventional air conditioning (AC) systems have limited control of sensible cooling and latent cooling capacities; therefore additional energy-consuming devices, i.e. electric heaters, are often used to reheat the conditioned air in order to provide thermal comfort for the building occupants. Separate sensible and latent cooling (SSLC) AC systems are capable of providing better control of cooling at no extra overload in the form of energy input. Moreover, because of a higher coefficient of performance (COP) in the sensible cooling cycle, the SSLC technology reduces total energy input to vapor compression systems (VCS), and makes AC systems more energy efficient. This dissertation explores and compares two main methods for implementing the SSLC concept: cycle options for SSLC systems and methods of indoor heat transfer. One of these options consists of two independent VCS, and the other consists of one VCS removing sensible load only and one solid desiccant wheel (DW) regenerated with the waste heat from the VCS. The objectives of the system option study are to understand the reasons behind energy savings and explore the best possible configurations of SSLC systems in different summer outdoor conditions. The simulation results of the first kind of SSLC system show that the energy savings come from a reduced compressor power input of the sensible cycle. Under wide ranging ambient conditions, the amount of energy savings varies from 22% to 50% over conventional system energy input. However, such a system has limited independence of varying sensible to latent load ratio and the extra cost of an internal heat exchanger. The integration of VCS and DW overcomes these limitations. An experimental setup was constructed in an environmental chamber to test the performance of the second kind of SSLC system using carbon dioxide as refrigerant. The experimental results show only a 7% improvement by using SSLC systems, and two negative factors hindering SSLC systems from achieving more energy savings were later identified. As a result, the application of divided heat exchangers is proposed as a solution to address one of the issues. An optimal SSLC system, which incorporates the application of divided heat exchangers, an enthalpy wheel and other energy-saving methods, was modeled and demonstrated a doubling of the COP as compared to a conventional AC system. The second method crucial to implementing SSLC is a so called ?low ?T indoor heat exchanger? which is being introduced as an improved sensible heat exchanger design for the successful implementation of SSLC system concept. Its capability of providing both radiative heat transfer and convective heat transfer leads to better thermal comfort to occupants. Compared to the baseline fan-coil unit, the low ?T indoor heat exchanger creates better thermal comfort in terms of reducing temperature stratification from head to feet by 0.8 K and providing higher operative temperature at the foot level in winter. Numerical models were developed to simulate the operative temperature field created by the low ?T indoor heat exchanger. The model had only an average deviation of 0.4 K compared to the experimental data. The air temperature simulation in the model was later replaced by the proper orthogonal decomposition (POD) method. The POD method provides simulation results almost identical to CFD simulation (maximum deviation of 0.1 K), and moreover reduces the computation time from 24 hours to only minutes. The major contributions in this dissertation are listed as follow: Exploration of energy saving potential of the SSLC systems: ? Design, fabricated and tested an SSLC air conditioning system and compared its performance to a conventional system ? Compared the performance of SSLC systems using two refrigerants, R-410A and CO2 ? Based on experimental results, established models to simulate SSLC systems o Simulated SSLC system performance under different ambient conditions o Optimized the vapor compression cycle operation under each ambient condition o Explored maximum energy saving options (configurations) of an SSLC system Thermal comfort study of the low ?T heat exchanger: ? Established a low ?T heat exchanger test facility with sensors for operative temperature measurement ? Compared the thermal comfort zone created by the baseline fan-coil unit and low ?T heat exchanger system ? Developed models to simulate the thermal comfort zone in an office setting o Simulate natural convection by a commercial CFD tool and obtain 2D air temperature field in the conditioned space o Simulate radiation cooling (heating) and obtain 3D mean radiation temperature field in the conditioned space ? Developed a reduced-order POD model to replace the CFD simulation of air temperature in the conditioned space and verify the POD model by comparing its results to the original CFD model ENERGY SAVINGS AND THERMAL COMFORT OF SEPARATE SENSIBLE AND LATENT COOLING AIR-CONDITIONING SYSTEMS By Jiazhen Ling Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park, in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2011 Advisory Committee: Professor Reinhard Radermacher, Chair Assistant Professor Amir Riaz Associate Professor Bao Yang Associate Professor Gary Pertmer, Dean?s representative Professor Gregory Jackson Research Associate Professor Yunho Hwang ? Copyright by Jiazhen Ling 2011 ii Dedication To Vivian Ling (???) My beloved daughter who will be born in September 2011 iii Acknowledgement I am forever grateful for my advisor, Dr. Radermacher, for giving me the opportunity to conduct research at Center for Environmental Energy Engineering (CEEE). I came in as a student interested in obtaining a Ph.D. but through his guidance and advice, today I am a researcher interested in finding solutions to all problems. I am also grateful to my dissertation committee members, Dr. Jackson, Dr. Yang, Dr. Riaz and Dr. Pertmer for their time and effort in making this dissertation more valuable to the engineering and scientific community. I would like to offer a special note of thanks for Dr. Hwang who gave me numerous advices and comments on my day-to-day research. He has shown unlimited amount of patience in answering my questions. His professional and meticulous working attitude also sets a great example for me. I would also like to thank Mr. Osamu Kuwabara from Sanyo Electric. He taught me the skills of conducting experiments and made me feel like I am a real engineer. The working experience with him has become a valuable asset in my PhD study. The experiments related to the low ?T heat exchangers would not exist in my dissertation without the help from Mathias Koepke. He set up the experimental facility and conducted the tests with great care. The data from the tests are priceless for my later research. I would like to thank every colleague in CEEE for the help I received in the past four and half years, especially Jan Muehlbauer. His creative ideas on building iv test facility saved me huge amounts of time and provided me with better results than I can ever imagine. Most importantly, I am indebted to my wife Yuan Zhang and my parents. Without their continuous love and sacrifice, the achievement would not be possible. v Table of Contents Table of Contents .......................................................................................................... v List of Tables .............................................................................................................. vii List of Figures ............................................................................................................ viii Nomenclature ............................................................................................................... xi Chapter 1: Introduction and Literature Review ............................................................ 1 1.1 Introduction to the Separate Sensible and Latent Cooling Systems ................... 1 1.2 Thermal Comfort of SSLC System ..................................................................... 5 1.3 Literature Review................................................................................................ 8 1.3.1 Literature Review on SSLC Systems ........................................................... 8 1.3.2 Literature Review on Thermal Comfort Standards .................................... 11 1.3.3 Literature Review on Natural Convection and Its Enhancement ............... 19 1.3.4 Literature Review on View Factor Calculation ......................................... 37 1.3.5 Literature Review of Multi-mode Heat Transfer ....................................... 41 1.4 Summary of Literature Review ......................................................................... 44 1.5 Research Objectives .......................................................................................... 46 Chapter 2: Energy Saving Analysis of SSLC Systems ............................................... 48 2.1 SSLC Systems Using Two Vapor Compression Cycles ................................... 48 2.1.1 System Description .................................................................................... 48 2.1.2 System Modeling Approaches ................................................................... 50 2.1.3 Modeling Results ....................................................................................... 51 2.1.4 Parametric Studies ..................................................................................... 55 2.1.5 Air Distribution Methods ........................................................................... 60 2.2 SSLC Systems Using One Vapor Compression Cycle and One Desiccant Wheel ...................................................................................................................... 64 2.2.1 Experimental Setup .................................................................................... 64 2.2.2 Test Results ................................................................................................ 68 2.2.3 Exploration of Better SSLC Systems ......................................................... 78 2.2.4 Improved System Modeling Approach ...................................................... 84 2.2.5 Modeling Results and Discussion .............................................................. 87 Chapter 3: Experimental Assessment for the Low ?T Heat Exchangers ................... 96 3.1 Chilled Ceiling Panels, Heated Floor Systems and the Low ?T Heat Exchanger ................................................................................................................................. 96 3.2 Sensors for Operative Temperature Measurement ........................................... 99 3.2.1 A Simplified Operative Temperature Calculation ..................................... 99 3.3 Low ?T Heat Exchanger Test Facility ........................................................... 104 3.3.1 The Hot Water Supply Loop .................................................................... 104 3.3.2 The Assembly of Low ?T Heat Exchanger Panels.................................. 105 3.3.3 Room Selection ........................................................................................ 107 3.4 Low ?T Heat Exchanger Experiments Results............................................... 110 3.4.1 Baseline System Experiment Results....................................................... 110 3.4.2 Low ?T Heat Exchanger Experiment Results ......................................... 115 Chapter 4: Modeling the Operative Temperature Field in an Office Setting ........... 121 4.1 The Objectives of Operative Temperature Field Modeling ............................ 121 vi 4.2 The Calculation of Mean Radiation Temperature (MRT) .............................. 123 4.2.1 Model Description ................................................................................... 124 4.3 Calculation of Air Temperature inside an Enclosure ...................................... 129 4.3.1 Model Description ................................................................................... 129 4.3.2 CFD Simulation ....................................................................................... 131 4.3.3 Linear Curve Fit for CFD Results ............................................................ 136 Chapter 5: A Reduced-order Simulation Method for Air Temperature Calculation 147 5.1 Introduction on the POD Method ................................................................... 147 5.2 Application of the POD Method on the Natural Convection in an Enclosure 152 5.3 Introduction on the Galerkin Projection Method ............................................ 163 5.4 POD Simulation Results and Validations ....................................................... 166 Chapter 6: Summary and Conclusions ...................................................................... 172 6.1 SSLC System Using Two Vapor Compression Cycles .................................. 172 6.2 SSLC System Using One Vapor Compression Cycle and Desiccant Wheel .. 173 6.3 The low ?T heat exchanger test ...................................................................... 175 6.4 Modeling the operative temperature field in an office setting ........................ 176 6.5 A Reduced-order Simulation Method for Air Temperature Calculation ........ 178 Chapter 7: List of major contributions and future work ........................................... 179 7.1 List of major contributions.............................................................................. 179 7.2 List of related publications.............................................................................. 181 7.3 Future work ..................................................................................................... 184 References ................................................................................................................. 186 vii List of Tables Table 1: Thermal sensation based on PMV scale ....................................................... 12 Table 2: Thermal manikin: main dimensions and skin areas of the 16 segments....... 17 Table 3: Dimensionless group and description in Eq. (27) ......................................... 42 Table 4: Power savings under different climate conditions ........................................ 60 Table 5: Pressure lift and air flow rate requirements for fans used in the baseline and SSLC systems ............................................................................................................. 62 Table 6: Specifications of experimental components and instruments used in the SSLC test .................................................................................................................... 65 Table 7: Detailed test condition settings in the SSLC test .......................................... 67 Table 8: Operating conditions tested in the DW experiments .................................... 75 Table 9: COP Comparison between Baseline and SSLC Systems ............................. 77 Table 10: Optimization results of SW-DW-assisted SSLC system ............................ 93 Table 11: Optimization results of EW-DW-assisted SSLC system ............................ 94 Table 12: Operative temperature calculation .............................................................. 99 Table 13: Load calculation components ................................................................... 108 Table 14: Fluctuation terms of u-velocity snapshot .................................................. 158 Table 15: Fluctuation terms of v-velocity snapshot .................................................. 158 Table 16: Fluctuation terms of temperature snapshot ............................................... 158 Table 17: Kernel matrix of u-velocity ...................................................................... 159 Table 18: Kernel matrix of v-velocity ...................................................................... 159 Table 19: Kernel matrix of temperature.................................................................... 160 viii List of Figures Figure 1: Psychrometric process of conventional AC operation. ................................. 2 Figure 2: Psychrometric process of one kind of SSLC system (VCS + DW) .............. 2 Figure 3: ASHRAE thermal comfort zones for winter and summer .......................... 13 Figure 4: Block diagram for evaluating the perception of human thermal comfort ... 16 Figure 5: A 2D enclosure with boundary conditions and dimensions for natural convection study ......................................................................................................... 23 Figure 6: Isotherms and streamlines for Ra = 107, stainless steel walls ..................... 27 Figure 7: Comparison of the Nusselt number in the enclosure with (right) and without radiation ...................................................................................................................... 29 Figure 8: Schematic diagram and coordinate system for a square enclosure with inclined fin at the center of the hot wall ..................................................................... 29 Figure 9: Schematic diagram of an enclosure with partition ...................................... 30 Figure 10: Streamlines and isotherms of the partitioned enclosure ........................... 30 Figure 11: Fin with equilateral triangular perforations to enhance natural convection ..................................................................................................................................... 33 Figure 12: Schematic diagram of the physical system................................................ 34 Figure 13: Local cycle-averaged Nusselt number results for the surface of the fin facing the neighboring fin. Curve designation: (A) baseline, no oscillation, (B) 10 Hz, (C) 50 Hz and (D) 100 Hz. .......................................................................................... 35 Figure 14: Experimental apparatus, simple heat transfer promoter and split heat transfer promoters ....................................................................................................... 36 Figure 15: Visualized fluid motions in the downstream region of a split heat transfer promoter (x-y plane, Nz = 5) ....................................................................................... 36 Figure 16: Visualized fluid motions in the downstream region of a split heat transfer (y-z plane, Nz = 4) ....................................................................................................... 37 Figure 17: Schematic diagram of view factor between two infinitesimal surfaces .... 39 Figure 18: (a) Physical model of two-fin enclosure. (b) Different surfaces in radiation enclosure ..................................................................................................................... 43 Figure 19: Schematic diagram of the 2VCC SSLC system ........................................ 49 Figure 20: Comparison of SSLC systems and baseline system in psychrometric chart ..................................................................................................................................... 52 Figure 21: COP and total air flow rate under different air temperature leaving sensible evaporator ................................................................................................................... 56 Figure 22: Power savings and ratio of sensible to latent load under different ambient relative humidities ....................................................................................................... 57 Figure 23: Power savings and ratio of sensible to latent load under different ambient temperatures ................................................................................................................ 58 Figure 24: Six different climate conditions plotted on the psychrometric chart ......... 59 Figure 25: Air distribution method for the SSLC system ........................................... 61 Figure 26: Schematic diagram of the experimental setup for SSLC system test ........ 64 Figure 27: Pressure range of R-410A SSLC test ........................................................ 68 Figure 28: Temperature range of R-410A SSLC test ................................................. 69 Figure 29: Pressure range of CO2 SSLC test .............................................................. 70 Figure 30: Temperature range of CO2 SSLC test ....................................................... 71 ix Figure 31: COP profile of R-410A SSLC tests with error bars .................................. 72 Figure 32: COP profile of CO2 tests with error bars .................................................. 73 Figure 33: The refrigeration cycle of the DW-assisted SSLC system with divided HXs ..................................................................................................................................... 79 Figure 34: New DW-assisted SSLC system option 1: evaporative cooling ............... 80 Figure 35: New DW-assisted SSLC system option 2: evaporative cooling + fresh air ..................................................................................................................................... 81 Figure 36: New DW-assisted SSLC system option 3: EW + evaporative cooling + fresh air ....................................................................................................................... 82 Figure 37: New DW-assisted SSLC system option 4: SW + evaporative cooling + DOS application .......................................................................................................... 82 Figure 38: New DW-assisted SSLC system option 5: EW + evaporative cooling + DOS application .......................................................................................................... 83 Figure 39: Psychrometric process of new SSLC system option 1 .............................. 87 Figure 40: Psychrometric process of new SSLC system option 2 .............................. 88 Figure 41: Psychrometric process of DW-EW-assisted SSLC system (option 3) ...... 88 Figure 42: Psychrometric process of DW-SW-assisted DOS system (option 4) ........ 89 Figure 43: Psychrometric process of DW-EW-assisted DOS system (option 5)........ 89 Figure 44: The system COP comparison of different SSLC options and baseline system ......................................................................................................................... 90 Figure 45: P-h diagram of CO2 DW-assisted SSLC systems ..................................... 91 Figure 46: Picture of RTD sensor ............................................................................. 100 Figure 47: Picture of assembled operative temperature sensor ................................ 101 Figure 48: Positions of four operative temperature sensors ...................................... 102 Figure 49: OT sensors readings in one day ............................................................... 102 Figure 50: Uncertainty analysis of operative temperature sensors ........................... 103 Figure 51: Schematic diagram of low ?T HX test facility ....................................... 105 Figure 52: Low ?T HX?s sheet and tube .................................................................. 106 Figure 53: Insulation for the backside panel of low ?T HX ..................................... 106 Figure 54: Installed panels in the test office ............................................................. 107 Figure 55: Sketch of the office under low ?T study ................................................. 108 Figure 56: Heating load analysis of the test office ................................................... 109 Figure 57: Cooling load analysis of the test office ................................................... 109 Figure 58: Operative temperature measurement in the baseline test ........................ 111 Figure 59: Window surface temperature variation during the baseline test ............. 112 Figure 60: Thermal comfort analysis of the baseline test: operative temperature stratification .............................................................................................................. 113 Figure 61: Thermal comfort analysis of the baseline test: operative temperatures in the comfort zone ....................................................................................................... 114 Figure 62: Water flow rate variation during the low ?T HX.................................... 116 Figure 63: Heater inlet and outlet temperatures? variations...................................... 117 Figure 64: Comparison of measured heating capacities of heater and HX ............... 117 Figure 65: Thermal comfort analysis of the low ?T HX test: Operative temperature stratification .............................................................................................................. 119 Figure 66: Thermal comfort analysis of low ?T HXs: operative temperatures in the comfort zone ............................................................................................................. 119 x Figure 67: Adopted radiation model setup................................................................ 125 Figure 68: Calculation of sun light area .................................................................... 126 Figure 69: Calculation of view factor between two infinitesimal areas .................. 127 Figure 70: View factor from a sphere to a non-intersected rectangular area ............ 128 Figure 71: Boundary conditions and dimensions of the adopted natural convection model......................................................................................................................... 130 Figure 72: 1st generation of mesh generated by Gambit .......................................... 132 Figure 73: Screenshot of viscous model GUI in Fluent ............................................ 132 Figure 74: Isotherms of air in the enclosure (1 m by 1 m) ....................................... 133 Figure 75: Streamlines of air in the enclosure (1 m by 1 m) .................................... 133 Figure 76: 2nd generation of mesh generated by Gambit ......................................... 134 Figure 77: Streamlines of air in the enclosure (3 m by 3 m) .................................... 134 Figure 78: Isotherms of air in the enclosure (3 m by 3 m) ....................................... 135 Figure 79: Comparison of temperature readings from CFD and curve fitting (bulk air region) ....................................................................................................................... 137 Figure 80: Comparison of temperature readings from CFD and curve fitting (close-to- cold wall region) ....................................................................................................... 137 Figure 81: Comparison of temperature readings from CFD and curve fitting (close-to- hot wall region) ......................................................................................................... 138 Figure 82: Simulated operative temperature of air in the bulk flow region.............. 139 Figure 83: Simulated operative temperature of air near the cold wall ...................... 140 Figure 84: Simulated operative temperature of air near the hot wall ....................... 140 Figure 85: Comparison between OT simulation results and experimental data (low ?T HX off) ...................................................................................................................... 142 Figure 86: Comparison between OT simulation results and experimental data (low ?T HX on) ...................................................................................................................... 144 Figure 87: Indication of the search for the optimum basis of u ................................ 149 Figure 88: Isotherms of CFD snapshots (Ra 106) ................................................. 152 Figure 89: Streamlines of CFD snapshots (Ra ~ 106) ............................................... 153 Figure 90: Isotherms of CFD snapshots (Ra ~ 107) .................................................. 153 Figure 91: Streamlines of CFD snapshots (Ra ~ 107) ............................................... 154 Figure 92: Isotherms of CFD snapshots (Ra ~ 108) .................................................. 154 Figure 93: Streamlines of CFD snapshots (Ra ~ 108) ............................................... 155 Figure 94: Isotherms of CFD snapshots (Ra ~ 109) .................................................. 155 Figure 95: Streamlines of CFD snapshots (Ra ~ 109) ............................................... 156 Figure 96: Eigenvalue spectrum of temperature snapshots ...................................... 161 Figure 97: Eigenvalue spectrum of u-velocity snapshots ......................................... 161 Figure 98: Eigenvalue spectrum of v-velocity snapshots ......................................... 162 Figure 99: The matrix of average temperature ......................................................... 167 Figure 100: Temperature POD mode #1 ................................................................... 167 Figure 101: Temperature POD mode #2 ................................................................... 168 Figure 102: Temperature POD mode #3 ................................................................... 168 Figure 103: Comparison of POD calculation and CFD simulation (Ra = 106) ........ 169 Figure 104: Comparison of POD calculation and CFD simulation (Ra = 109) ........ 170 Figure 105: Comparison of POD simulated OT and experimental results ............... 171 xi Nomenclature a Coefficient of the POD modes AR Aspect ratio ARI Air-Conditioning and Refrigeration Institute C Kernel matrix CHP Cooling, heating and power COP Coefficient of performance cp Specific heat (kJmol -1k-1) D Diameter (m) e Internal energy (kJkg-1) EES Engineering equation solver F View factor g Gravitational acceleration (ms-2) Gr Grashof number HX Heat exchanger ? Mass flow rate (kgs-1) N Rotation speed (rpm) Nu Nusselt number P Pressure lift (Pa) Pr Prandtl number PR Pressure ratio ? Source term (per volume) Q Air flow rate (m3s-1) Ra Rayleigh number xii RH Relative humidity SHF Sensible heat Factor SSLC Separate sensible and latent cooling T Temperature (K) UA Overall heat transfer conductance of heat exchanger (kWK-1) V, v Velocity (ms-1) VFR Volume flow rate (m3s-1) Greek Letters Thermal diffusivity (m2s-1) ? Thermal expansion coefficient (K-1) ? Turbulent dissipation Compressor efficiency ? Turbulent kinetic energy ? Dynamic viscosity (Pa s) ? Kinematic viscosity (m2s-1) Density (kgm-3) ? Time term ? POD mode ? Dissipation term ?P Pressure drop ?T Degree of subcooling/superheat Subscripts xiii 1 SSLC system 2 Baseline system iso Isentropic ref Refrigerant vol Volumetric r Radiation c Convection s Surface ? Quiescent fluid 1 Chapter 1: Introduction and Literature Review 1.1 Introduction to the Separate Sensible and Latent Cooling Systems During operation of a conventional air-conditioning (AC) system, two kinds of cooling, i.e., sensible cooling and latent cooling are provided to a conditioned space. The sensible cooling is provided by supplying cold air to reduce the temperature of the conditioned space. The source creating the cold air is the evaporator which is filled with two-phase refrigerant to absorb heat. When the refrigerant temperature in the evaporator is below the dew point of the room air, it causes water vapor in moist air to condense on the evaporator and therefore reduces the humidity ratio of air. The drier air removes the latent load in the space. Figure 1 shows a psychrometric process where point B refers to the dew point of room air and point D refers to supply air. 2 Figure 1: Psychrometric process of conventional AC operation. Figure 2: Psychrometric process of one kind of SSLC system (VCS + DW) There are two limitations related to the operation of conventional AC systems: ? Reheat process reduces system COP 3 Theoretically, the process of supply air flowing through the evaporator follows the path that is composed of a horizontal sensible load removal part (point A to point B) and a latent load removal part along the100% relative humidity (RH) line from B to C. Usually, the temperature of point C is too low for thermal comfort, therefore a reheat process, typically adopted in commercial buildings, is added to increase the temperature of point C to the temperature of point D. The reheat process, usually carried out by electric heaters, requires extra energy input and increases the total net energy input. Hence, the reheat process reduces the COP. COP = useful refrigerating effect net energy input (McQuinston, 2005) ? Conventional AC systems lack the independent control of sensible and latent cooling The reheat process in conventional systems is in fact caused by the lack of independent control of sensible and latent cooling. The path from point B to point C along the 100% RH line reveals that the amount of latent cooling and the amount of sensible cooling are co-dependent to each other and relevant to the slope of the 100% RH line. That is to say, removing a certain amount of water vapor requires an accompanying ratio of temperature reduction. Therefore, the more the latent cooling, the more likely leads to sensible over-cooling. Such a dependent relationship not only costs a reheat stage but also causes a control issue in a conventional AC systems? operation. For example, when more people enter the room, extra latent cooling (the vertical blue arrow pointing downward) is required. The supply air point moves downwards to point C?. Meanwhile, an unnecessary amount of sensible cooling (the 4 horizontal blue arrow pointing leftward) has to be added to the room as well. This requires more reheat power input to increase temperature for thermal comfort and further reduces the COP. In order to overcome the two limitations of conventional systems, separate sensible and latent cooling (SSLC) systems are hereby proposed as a solution. Figure 2 plots the psychrometric process of one kind of SSLC systems, which consists of one VCS and one solid desiccant wheel (DW). The VCS provides only sensible cooling (point A to point B) required by the conditioned space at both elevated air temperature leaving the evaporator (point B in Figure 2 vs. point C in Figure 1) and a higher air mass flow rate (MFR). The reason for a higher air MFR requirement is to compensate for the reduced enthalpy difference of air across the evaporator, and to maintain the capacity of sensible cooling. Since the VCS operates above the dew point temperature of air and cannot provide latent cooling, the DW is used to reduce the water vapor content in the part of the air leaving from the sensible evaporator. The part of the dry air from the DW mixes with the rest of the air from the evaporator and is delivered to the conditioned space (point D). DWs absorb water vapor in air and provide latent cooling; but they generate heat of adsorption during the process and increase the dry air temperature. Theoretically, the amount of latent cooling is equal to the amount of sensible heat generation. Hence the operation of DWs follows an isenthalpic line, which is from point B to point C in Figure 2. SSLC systems have two features during their operations: 5 ? No reheat process is needed Since the VCS used in an SSLC system operates above the dew point temperature, the supply air temperature (see point D in Figure 2) is thermally comfortable enough to be sent to the conditioned room directly. No reheat is necessary in SSLC systems. ? Independent control on sensible and latent cooling An SSLC system uses a VCS to provide sensible cooling. In consequence, any fluctuations of sensible cooling demand can be simply met by changing the capacity of the VCS. The control method to deal with the fluctuations of latent cooling demand is a little bit more complicated. Of course, the rotation speed of DW can be adjusted to meet the fluctuations of latent cooling demand within a certain range. Any latent cooling demand change beyond the reach of rotation speed adjustment can be met by either increasing the regeneration temperature or increasing the air MFR through a DW. It should be noted that although a DW is a stand-alone device providing latent cooling, any amount of the latent capacity change would theoretically lead to the same amount of change in sensible heat generation. Therefore, the VCS must increase the cooling capacity to cover the extra heat. However, such an increase will not lead to over-cooling because the VCS still operates above the dew point. 1.2 Thermal Comfort of SSLC System To evaluate the performance of an air-conditioning system, thermal comfort is another important factor need to be considered besides energy consumptions. As an 6 example, in summer, people can reduce the energy consumption by raising the thermostat reading. However, there is an upper limit for most of people who refuse to raise the thermostat anymore. According to the ASHRAE standard 55, thermal comfort is defined as the condition of mind which expresses satisfaction with the thermal environment and is assessed by subjective evaluation. Although the evaluation is subjective, there are several parameters that are considered to have significant impacts on the occupants. They are air temperature, mean radiation temperature, humidity ratio of air, air velocity, occupant?s metabolic rate and clothing insulation. The first four parameters are directly controlled by the AC unit in a conditioned space. To be more specific, it is the indoor unit of the AC unit that controls the air conditions. Therefore, in order to study the thermal comfort of the SSLC system, the research should be focused on the design of indoor unit, i.e., the sensible cycle evaporator. There are several issues regarding to the sensible evaporator design that need to be addressed in the thesis. How to solve the problem of large air side pressure drop? What is the thermal comfort condition of using sensible evaporators? The first question comes from the requirement of larger amount of air MFR through the sensible heat exchanger in order to compensate the smaller air enthalpy difference than that of a conventional system. To reduce the air side pressure drop, the frontal area of the sensible heat exchanger has to be larger than conventional heat exchanger so that the air velocity can be reduced. Some current products, such as chilled ceiling panels and heated floor systems, utilize large frontal area to provide sensible cooling and heating. The common characteristic of the two systems is that it has a low 7 temperature difference between the working fluid (refrigerant) and the indoor air. For example the chilled ceiling panels use cold water temperature typically of 16 ? 18?C to keep the room air temperature at around 25?C. However, conventional evaporators use refrigerant temperature around 7 ? 10?C to keep the same indoor condition. This dissertation introduces a new term called ?low ?T heat exchanger? to describe the improved design of sensible heat exchanger such as chilled ceiling panels. Another benefit of using the low ?T heat exchanger comes from its capability of providing radiative heat transfer to the occupants. This unique capability helps low ?T heat exchanger control the mean radiation temperature (MRT). MRT, as mentioned above, is one of the factors affecting occupant?s thermal comfort, but it cannot be effectively controlled by convectional systems. 8 1.3 Literature Review 1.3.1 Literature Review on SSLC Systems There exist different methods to achieve the separation of the sensible cooling and latent cooling. Ling et al. (2009) proposed the most straightforward method. Ling?s idea was to separate the two forms of cooling by two vapor compression systems. The first system removes sensible load only, while the second system removes both latent load and a small amount of sensible load. Under the standard ambient conditions (35?C, 44% relative humidity (RH)), the energy consumption of such an SSLC system was reduced by 30% compared with that of a conventional system, and the savings was reported to be up to 50% under the hot and dry condition (37?C, 15% RH). Although Ling?s separation method is straightforward, there are two problems need to be addressed. First the sensible cycle cannot remove the entire sensible load in the system. There is always a small amount of sensible load associated with the process of latent load removal in the latent cycle. Because the energy savings of the SSLC system comes from the high-COP sensible cycle, an incomplete separation means such configuration is not the best option (Ling et al. 2009). Second, an internal heat exchanger is required in the SSLC configuration to recover the cooling from the latent cycle, but the extra cost of the internal heat exchanger was not considered in the paper. More studies were focused on the application of using a vapor compression cycle for sensible load removal and solid/liquid desiccant equipment for latent load removal. Yadav (1995) investigated a hybrid system consisting of a liquid desiccant and a vapor compression system. The objective of the study was to find the best operating condition of such system, and the 9 conclusion was that either having a low sensible heat factor (SHF) condition or when the ambient humidity ratio was high. The SHF is defined as the ratio of sensible heat over the total heat load. Dai et al. (2001) studied the application of integrating a liquid desiccant device and a vapor compression cycle. The test was conducted under the AHRI standard 210/240 conditions (35?C, 44% RH, AHRI, 2008) and the cooling capacity was 5 kW. The coefficient of performance (COP) of the vapor compression cycle improved from 2.2 to 3.39 because of the assistance from the liquid desiccant. Ma et al. (2006) utilized a similar configuration to a larger scale application. A green building demonstration project in Shanghai required a total 60 kW cooling capacity, and the latent cooling was provided by a liquid desiccant unit powered by waste heat from a heat pump. The sensible heat was removed by two 10 kW adsorption chillers powered by a 150 m2 solar collector and the heat pump powered by electricity. The performance of this complicated system was evaluated at two different SHFs, 0.7 and 0.58, and the COP?s were 44.5% and 73.8%, respectively, higher than a conventional VCS. Other than the energy savings results, no economic analysis was conducted. Similar study was also conducted by Katejanekarn et al. (2008) in Thailand. Dhar and Singh (2001) simulated a hybrid system of a solid desiccant wheel (DW) and a vapor compression cycle. They demonstrated that the hybrid system had maximum energy savings under hot and dry weather. In hot and humid region, energy savings was still possible but the space latent load should be high. Depending on different desiccant materials, the temperatures of regeneration can vary between 50?C and above 100?C, therefore different heat sources are reported to drive desiccant devices. Jia et al. (2006) studied the performance of a solid DW using lithium chloride as the adsorbent. 10 The temperature required to regenerate the wheel was set to be 100?C, and one regeneration heater was used as a heat source. Ghali (2008) numerically simulated a hybrid system in the ambient conditions of Beirut. The main feature of this hybrid system was that the regenerative heat needed by the desiccant wheel was partly supplied by the condenser dissipated heat while the rest was supplied by an auxiliary gas heater. The hybrid air conditioning system was compared with a 23 kW vapor compression unit for a typical office in Beirut characterized by a high latent load. The size of the vapor compression subsystem was reduced to 15 kW at the peak load when the regeneration temperature was fixed at 75 ?C. Also the sensible heat ratio of the combined hybrid system increased from 0.47 to 0.73. The paper also conducted a preliminary economic analysis. The annual running costs savings for the hybrid system was 418.39 USD for a gas cost price of 0.141 USD/kg. The payback period of the hybrid system was less than five years when the initial cost of the hybrid air conditioning system priced an additional 1712.00 USD. Hence, for a 20-year life cycle, the life cycle savings of the hybrid air conditioning system were 4,295.19 USD. Casas and Schmitz (2004) studied the integration of a DW and a cooling, heating, and power (CHP) unit. In their study, the waste heat from the CHP unit could be utilized for lithium chloride regeneration. However, the regeneration temperature was only in the range between 50?C and 60?C. The difference in regeneration temperatures in these works may be caused by different dehumidification requirements. Besides lithium chloride, silica gel is another widely accepted candidate for desiccant material, and its regeneration temperature is usually higher than 70?C (Neti, 2000). Different energy efficiency evaluation method was 11 also reported in literature. Exergy analysis of a solar driven hybrid system was investigated by Ahmed et al. (1998). They compared the performance of the hybrid system operated at different ambient conditions and different mass flow rates through the desiccant wheel. The conclusion was that the maximum irreversibility was generated at an ambient vapor pressure of 3.33 kPa and a desiccant mass flow rate of 5 kg hr-1m-2. 1.3.2 Literature Review on Thermal Comfort Standards In 2004, ASHRAE released the standard 55-2004, Thermal Environmental Conditions for Human Occupancy, to specify the combinations of indoor thermal environmental factors and personal factors that will produce thermal environmental conditions acceptable to a majority of the occupants within the space. A PMV index, predicted mean vote, is used to measure a large group of persons? thermal sensation on a seven-level scale which uses +3 to be hot and -3 to be cold. Table 1 lists the detail definition of the PMV. The seven-level scale is usually enough to describe the thermal sensation in building applications, however higher numbers than 3 and lower numbers than -3 are also found in some literature to describe thermal sensation in extreme conditions, such as the cabin of a car parking in direct sunlight in summer for a long time. 12 Table 1: Thermal sensation based on PMV scale PMV scale Thermal sensation +3 hot +2 warm +1 slightly warm 0 neutral -1 slightly cool -2 cool -3 cold Six primary factors, which are metabolic rate, clothing insulation, air temperature, radiant temperature, air speed and humidity, are included in the standard when defining conditions for thermal comfort. Besides the primary factors, there are a number of other secondary factors affecting comfort in some circumstances. Figure 3 plots two thermal comfort zones on a psychrometric chart. Compared with other conventional psychrometric chart, the x-axis in the figure represents operative temperatures instead of dry-bulb temperatures. It is because the operative temperature takes both the dry-bulb temperature and the radiant temperature into consideration. The operative temperature can be calculated from Eq. (1). Two different thermal comfort zones are plotted for summer and winter because clothing insulation is 13 different. The boundaries of these two zones are determined by using Eq. (2) and Eq. (3), which calculate the maximum and minimum operative temperature ranges of the comfort zones. There is no unanimous agreement on what should be the low side comfort range of humidity, but ASHRAE suggests it should be higher than 2 gkg-1 dry air. Air speeds greater than 0.2 ms-1 (40 ftmin-1) may be used to increase the upper operative temperature limit for the comfort zone in certain circumstances to= hrtr?+hcta hr+hc (1) Figure 3: ASHRAE thermal comfort zones for winter and summer Tmin,Icl= (Icl-0.5 clo)Tmin,1.0clo+(1.0clo-IclTmin,0.5clo /0.5 clo (2) 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 0.018 0.020 10 15 20 25 30 35 H um id ity ra tio (k g/ kg -D ry A ir) Temperature (degC) Summer Winter 14 Tmax,Icl= (Icl-0.5 clo)Tmax,1.0clo+(1.0clo-Icl)Tmin,0.5clo /0.5 clo (3) where Tmax,Icl is the upper operative temperature limit for clothing insulation Icl Tmin,Icl is the lower operative temperature limit for clothing insulation Icl Icl is the thermal insulation of the clothing in question (clo) Besides ASHRAE Standard 55-2004, International Standards Organization (ISO) has also established a set of standards to address the thermal comfort issue. ISO standard 7730: 2005, Ergonomics of the thermal environment?Analytical determination and interpretation of thermal comfort using calculation of the PMV and PPD indices and local thermal comfort criteria, describes the PMV (Predicted Mean Vote) and PPD (Predicted Percentage Dissatisfied) indices and specifies acceptable conditions for thermal comfort. ISO 8996:2004, Ergonomics of the thermal environment?Determination of metabolic rate, describes six methods for estimating metabolic heat production, which are divided into three levels according to accuracy. ISO 9920:2007, Ergonomics of the thermal environment?Estimation of thermal insulation and water vapor resistance of a clothing ensemble, provides an extensive database of the thermal properties of clothing and garments. The properties are based upon measurements on heated manikins where basic (or intrinsic) thermal insulation is measured as well as vapor permeation properties of garments and ensembles. Since most vehicles have a HVAC system to control the thermal environment of the cabin, thermal comfort is also extensively studied by automotive thermal engineers. Compared to the building thermal load, 50 per cent of the automotive 15 thermal load is due to the solar heat gain (Shimizu et al. 1983). Radiative heat exchange can account for up to 70 per cent of the global sensible heat exchange between human and environment. However, the most significant difference between the field of building thermal comfort study and automotive thermal comfort is how to model the human body. For building thermal comfort modeling, Fanger (1967) proposed a single comfort equation to express thermal comfort of occupants exposed to constant conditions at constant metabolic rate for a long/sufficient period of time and such effort is later included into ASHRAE and ISO standards. However, considering the fact that solar radiation always varies during driving, automotive thermal comfort should be categorized to a dynamic boundary condition problem. The other difference is the volume of the cabin is much smaller to a space in the building, which makes the air flow field in the cabin extremely non-uniform. For example, air velocity can be largest at the outlet of air vents but almost be zero at passengers? backs due to the direct contact with the seat. Because of the non- uniformity and transient nature of automotive thermal comfort problems, automotive engineers have been working on developing detailed human body models to address the transient and non-uniform nature of automotive thermal comfort. The operative temperature is usually replaced by equivalent temperature or other more advanced physiology model to better capture thermal sensation of passengers in the cabin. Figure 4 describes the flow chart of automotive thermal comfort evaluation. 16 Figure 4: Block diagram for evaluating the perception of human thermal comfort (Source: Walgama et al. 2006) Olesen et al. (1988) was among the first to develop a detail human body model for better physiology (skin temperature) measurement. Table 2 shows the 16- segment manikin model as well as the surface areas results. The paper also studied five different clothing ensembles with the same total thermal insulation, but very different distributions of the insulation on the body in experiments with 16 sedentary subjects. The asymmetry was ranging from unclothed upper part to unclothed lower part of the body. Their experimental study provides a method for quantifying the non- uniformity of a clothing ensemble and examines how it influences local thermal discomfort. 17 Table 2: Thermal manikin: main dimensions and skin areas of the 16 segments (Source: Olesen et al. (1988)) Tanabe et al. (1994) later added sensible and latent heat exchange between each part of the body and the environment. The 16-segment body used in the paper was a female manikin and the total surface area of the body was therefore smaller than Olesen?s model. In order to better address the thermal comfort of a person in a non-uniform condition such as car cabin, the paper used an equivalent temperature (teq), which was defined as the temperature of a uniform enclosure in which a thermal manikin with realistic skin surface temperatures would lose heat at the same rate as it would in the actual environment, to calculate PMV. The idea of equivalent temperature has been widely accepted in automotive thermal comfort calculation. Han et al. (2001) also used a 16-segment male manikin model and improved the model by dividing each segment into four body layers (core, muscle, fat and skin tissues) and a clothing layer. The improved model had the ability to predict local thermal comfort level of an occupant in a highly non-uniform thermal environment, 18 and formulated the thermal comfort as a function of air velocity, humidity, direct solar flux, as well as the level of activity and clothing type of each individual. The author also emphasized the use of equivalent homogeneous temperature (EHT), which is similar to teq in Tanabe?s work, to quantify thermal comfort in the non- homogeneous (non-uniform) area. Kaynakli et al (2005) used Olesen?s manikin model but was focused on thermal comfort of passengers during vehicle?s initial warm-up and cooling period. In early minutes of warm-up, heat loss from body to environment is very high due to low inside air and surface temperatures. The average skin temperature of the body and contact temperature of hand with steering wheel were reported to fall to 32?C and 17.5?C, respectively. The thermal sensation (TS) was recorded as low as - 4.5 which was in the range between very cold and painfully cold and became neutral in 12 minutes. Opposite to the warm-up period, the TS starts from a very high value of 8 due to large sensible heat exchange between body and high temperature cabin. It takes almost 30 minutes to reach TS neutral. Walgama et al. (2006) presented a comprehensive survey of research studies regarding automotive thermal comfort. The work was classified according to whether it is concerned with the passenger compartment environment or the condition of the passengers and their interaction with the compartment. The review included factors associated with passenger compartment conditions, such as flow field and temperature field, which affect the thermal comfort of the occupants. The evolution of thermal comfort models was reviewed. Also included were various computational and empirical models for predicting physiological response and the sensation of thermal comfort in the non-uniform transient environment of a vehicle. 19 1.3.3 Literature Review on Natural Convection and Its Enhancement Natural convection refers to the motion of a flow is driven simply by the interaction of a difference in density in a gravitational field. The driving force of such density difference is a temperature difference, as in atmospheric and oceanic circulations, or in the air current arising from a cooling object. It may also be due to variation in composition or phase of a fluid, such as in most air rising. Natural convection is quite a different transport process compared with forced convection. The flow and temperature fields are invariably completely coupled and must be considered together. The flows are relatively weak, because the velocities are always relatively small and the inertial and viscous effects of momentum transport are usually of the same order. The basic equations of natural convection are continuity, momentum and energy equations, which are listed as Eq. (4) through Eq. (6) (Gebhart, 1971). The ? in the energy equation represents the source term over volume, and the ? represents the dissipation term. ? + ( V?? )=0 (4) DV?? D? = [ V?? ? +(V?? )V?? ]= g? - p+ 2V?? + 3 ( V?? ) (5) De D? = [ e ? +(V?? )e]= k t+q -p V?? + ? (6) The complexity and coupling inherent in natural convection processes are apparent in this set of equations. Motion results because is subject to change 20 according to temperature. The density term in equations can only be solved by considering the ?temperature? equation. The energy equation, in turn, inevitably involves velocity. Thus, the distributions of ? and e in space (x, y, z), and perhaps also in time , must be found simultaneously from these governing equations. In spite of the aforementioned complexity, governing equations are still useful to provide a great amount of information on natural convection problems. Most of the information comes from the simpler forms of the governing equations which are applicable in most physical circumstances. The most widely applied assumption is Boussinesq approximation. It assumes a linear density variation corresponding to temperature in the ? term (see Eq. (7)) and neglects the density variation in other places. - = ?t= (t-t ) (7) where is the density ? is the thermal expansion coefficient t is the temperature Natural convection problems can be further classified as either external one (free convection) or internal one (natural convection). Ostrach (1964), Ede (1967) and Gebhart (1979) conducted comprehensive reviews on the first problem. The second problem is considerably more complex than external one. This is because at large Rayleigh numbers (see Eq. (8) for definition), classical boundary-layer theory yields the same simplifications for external problems that are so helpful in other fluid-flow 21 problems, i.e., the region exterior to the boundary layer is unaffected by the boundary layer. Ra= g (Ts T )x 3 (8) where Ra is the Rayleigh number g is the gravitational acceleration Ts is the surface temperature T? is the quiescent temperature ? is the kinematic viscosity ? is the thermal diffusivity ? is the thermal expansion coefficient x is the characteristic length For confined natural convection, on the other hand, boundary layers form near the walls but the region exterior to them is enclosed by the boundary layers and forms a core region. Because the core is partially or fully encircled by the boundary layers, the core flow is not readily determined from the boundary conditions but depends on the boundary layer, which, in turn, is influenced by the core. The interactions between the boundary layer and core constitute a central problem that the flow pattern cannot be predicted a priori from the given boundary conditions and geometry. In fact the situation is even more intricate because it often appears that more than one global core flow is possible and flow sub-regions, such as cells and layers, may be imbedded in the core (Ostrach 1972, Ostrach 1982, Ostrach and Hantman 1981). Figure 5 shows one kind of widely studied enclosures. It is a square 22 enclosure with the side length of L. The upper and lower walls (ceiling and floor) are assumed to be well thermally insulated, and the left and right walls are at constant temperatures of Tc and Th, respectively. The stability of the natural convection problem is another widely discussed topic in literature. Taking the problem of a horizontal layer of fluid heated from below as an example, Schluter et al. (1965) pointed out that when a horizontal layer of fluid is heated from below, thermal expansion causes a density gradient opposite to the direction of gravity. In cases where the temperature gradient exceeds a certain critical value the static state of the fluid becomes unstable because the buoyancy force is sufficient to overcome the dissipative effects. The resulting cellular convective flow cannot be uniquely determined by the momentum equation and boundary conditions (Schluter et al. (1965)). Malkus and Veronis (1958) showed for special solutions that the degeneracy persists for finite amplitude solutions. They demonstrated that flows with rectangular or hexagonal cell pattern are finite amplitude solutions and that their number is infinite because the ratio of side lengths of a rectangular is a free parameter. Schluter et al. (1965) concluded that the instability of the hexagonal cell pattern was in a range between the critical Rayleigh number and a certain supercritical value. Beyond that, the rolls are stable. 23 Figure 5: A 2D enclosure with boundary conditions and dimensions for natural convection study Henkes and Hoogendoorn (1995) presented the governing equations, Eq. (9) through Eq. (14), for turbulent natural convection in an enclosure with simplifications. 2D steady state continuity equation for incompressible flow is: u x + v y =0 (9) Momentum equations with Boussinesq assumption are: u u x +v u y =- 1 p x + x ( + t) (2 u x )+ y ( + t) ( u y + v x ) (10) u v x +v v y =- 1 p y +g (T- Th+Tc 2 )+ x ( + t) ( v x + u y )+ y ( + t) (2 v y ) (11) 24 Energy equation is: u T x +v T y = x ( Pr + t T ) T x + y ( Pr + t k ) ( T y ) (12) Turbulent kinetic energy equation (k-? model) is: u k x +v k y = x ( + t k ) k x + y ( + t k )( k y )+Pk+Gk (13) Turbulent dissipation equation (k-? model) is: x +v y = x ( + t ) x + y ( + t ) ( y )+,C 1(Pk+c 3Gk) c 2 - . k / (14) with Pk= t *2 ( u x ) 2 +2 ( v y ) 2 + ( u y + v x ) 2 + Gk=- t T g T y t=c k 2 de Vahl Davis (1968) was probably among the first who studied the natural convection problem in an enclosure, although he only considered the problem in two conditions and both of them were limited in the laminar region. One condition was natural convection in a square enclosure with the Rayleigh number of 2?105 and the other was natural convection in a rectangular enclosure (aspect ratio of 5) with the Rayleigh number of 1.25?106. 25 Cormack et al. (1974) extended the geometry of the enclosure to any small aspect ratio which was defined as the width of the enclosure should be at least 12 times larger than the height. The walls of the enclosure were heated differentially. One analytic solution in terms of an asymptotic expression for the Nusselt number was presented in the study. The streamline profile was also plotted and it demonstrated that there were two regimes: a parallel flow in the core region and a second, non-parallel flow near the ends of the enclosure. Ostrach. (1988) provided a wide review on previous natural convection problem. It covered different geometries of enclosures including cylinders, rectangular enclosures of large aspect ratio and rectangular enclosures of small aspect ratio. As being pointed out in the conclusion, earlier research focused on searching analytical or experimental solutions. However, there was an ever-increasing proliferation of numerical solutions to such problems with more complications. Henkes and Hoogendoorn (1995) reported the outcomes of a workshop on turbulent natural convection in enclosures. The problem was defined as: a 2D square enclosure whose side length is unit with hot left and cold right vertical walls and adiabatic horizontal walls. Air with Pr = 0.71, assuming a Boussinesq fluid, at Ra = 5?1010. All the participants solved the problem by different CFD software packages. Their results were compared and numerical accuracies were reported against each other. Aounallah et al. (2005) used a commercial CFD software package, Fluent, to solve the similar problem as Henkes and Hoogendoorn but at a smaller Rayleigh number of 1.58?109. Three different mesh grids, 100?100, 120?120 and 200?200, 26 were studied but no significant improvement of accuracy were reported by refining the grid. K-? SST model was adopted to solve the turbulent equations. Both temperature and stream function field were very similar to the results of Henkes and Hoogendoorn?s work. One of the paper?s contributions was development of a reduced order correlation of the heat flux at the hot wall in terms of the normalized Nusselt number. Besides 2D convection problems, Sigey et al. (2004) studied a three dimensional enclosure in the form of a rectangular enclosure containing a conventional heater built into one wall and having a window in the same wall. In this problem, the Rayleigh number was varied from 5?1010 to 5?1011. Temperature stratification was reported in the paper: The room was stratified into three regions, a cold upper region, a hot region in the confluence of the hot and cold streams and a warm lower region. The results also showed that the location of the heater, as well as the size of the window, has an important influence on the overall heat transfer through the room. Ogut. (2009) used water-based nanofuids as the working fluid in an inclined square enclosure. The enclosure was heated at a constant heat flux on the left wall and cooled on the right. The floor and ceiling were kept adiabatic. Five types of particles were taken into consideration: Cu, Ag, CuO, Al2O3, and TiO2. Polynomial differential quadrature (PDQ) method was applied to solve the governing Equations. A parametric study was performed for inclination angles from 0? to 90?. It was found that nano-particles increased the average heat transfer rate, and the largest 27 improvement (63.9%) came from using Ag at a concentration of 20% volume fraction. Instead of applying an adiabatic-wall assumption, a further effort was made to make the wall assumption more realistic. Liaqat and Baytas (2001) studied a square enclosure with constant wall temperature and finite conductive properties. The space was filled with a Bousinessq fluid with a Prandtl number of 7.0 containing a uniform volumetric heat source. Control volume method was applied and the SIMPLER algorithm (Patankar, 1980) was utilized to handle the pressure and velocity coupling of governing equations. Isotherms and streamlines (in normalized form) obtained from the study are plotted in Figure 6. Figure 6: Isotherms and streamlines for Ra = 107, stainless steel walls (Source: Liaqat et.al (2001)) Sharma et al. (2007) added radiation into the natural convection problem. The conjugate turbulent natural convection problem was also defined in a rectangular 28 enclosure filled with air, and the walls were of finite thickness tw and finite conductivity kw. The floor of the enclosure was maintained at a constant temperature th, and its external surface was exposed to ambient temperature at tc. K-? model was applied to solve the turbulent problem. A correlation, Eq. (16), was developed to predict the overall heat transfer rate in terms of the Nusselt number (see Eq. (15) for definition): Nu= hL kf (15) where L is the characteristic length kf is the thermal conductivity of the fluid h is the convective heat transfer coefficient Nuc=0.152AR 0.267Ra0.34 (16) where AR is the aspect ratio varied from 0.5 to 2.0, and the Rayleigh number is from 108 to 1012. Bouali et al. (2005) also considered radiation and applied his model into an inclined rectangular enclosure. From the comparison shown in Figure 7, it was found that radiation improved the heat transfer in the enclosure. 29 Figure 7: Comparison of the Nusselt number in the enclosure with (right) and without radiation (Source: Bouali et al. (2005)) Ben-Nakhi and Chamkha (2007) added an inclined thin heated fin into the square enclosure (Figure 8). Surprisingly, rather than improving the heat transfer, the existence of the fin reduces the average Nusselt number. Two reasons were given: restraining natural convection and increasing heating surface. Figure 8: Schematic diagram and coordinate system for a square enclosure with inclined fin at the center of the hot wall (Source: Ben-Nakhi et al. (2007)) Cuckovic-Dzodzo (1999) complicated the enclosure problem by partitions which separated the enclosure into three parts. (Figure 9) 30 Figure 9: Schematic diagram of an enclosure with partition (Source: Cuckovic- Dzodzo et al. (1999)) The isotherms and streamlines of the problem with the Rayleigh number of 2.6?105 were plotted as shown in Figure 10. Figure 10: Streamlines and isotherms of the partitioned enclosure (Left two are streamlines and the right one is isotherms) (Source: Cuckovic-Dzodzo et al. (1999)) Because natural convection has a lower heat exchange rate compared with forced convection, many researchers have been working on the exploration of natural convection enhancement methods. The addition of different configurations of fin and active oscillators are two most widely reported practices. Both practices aim to disturb the growth of thermal boundary layer and promote effective heat transfer. 31 Frederick (2006) numerically studied the enhancement of natural convection by placing a thick vertical fin in the middle of a differentially heated cubical enclosure. He focused on investigating the variation of the Nusselt number with different Rayleigh numbers and thermal conductivity ratios. A main circulation and a flow restriction were observed but they were only significant at low Rayleigh numbers. A secondary circulation cell was reported at high thermal conductivity ratios. Long fins were found to be more effective in promoting heat transfer. Fujii (2007) studied the effect of the inclination angle of a finned surface to enhance natural convection. The fins were constructed on a 250 mm ? 240 mm ? 3 mm aluminum base plate. The angle of air flow through these fins changed from 30? to 90?. It was concluded that the enhancement was most significant at 60?. At this angle, the convective heat transfer rate was 19% higher than that of a vertical finned surface. A correlation of the Nusselt number and inclination angle was also obtained. Instead of vertically placed fins, Dialameh (2008) investigated the fluid flow and heat transfer through horizontal rectangular thick fin arrays. The lengths of these fins are under 50mm. Results showed that there were two types of flow depending on the ratio of height to length. If the ratio is smaller than 0.24, the air can only enter into the channel from fin end regions. However, the air flow can also enter into the middle parts of the fins if the ratio is greater than 0.24. With regards to the enhancement of natural convection, the natural convection heat transfer coefficient (HTC) increases with the fin spacing and the temperature difference, but decreases with the fin length. It was reported that the fin thickness and fin height did not affect the HTC considerably. Two correlations (Eq. (17) and Eq. (18)) were proposed to predict the 32 average Nusselt number of an array of fins based on the Rayleigh number, the ratio of fin height and fin length, the ratio of fin spacing and fin length and the ratio of fin height and fin thickness. Nus=0.625(Ra) 0.2382 . H L / 0.3674 . S H / 0.3303 . H t / -0.0504 for Ra < 1500 (17) Nus=0.5007(Ra) 0.2828 . H L / 0.4468 . S H / 0.3901 . H t / -0.083 for Ra > 1500 (18) Heindel et al. (1995) investigated the single phase natural convection enhancement by a discrete heat source with parallel plate fin arrays. Compared to unfinned conditions, the parallel plate fin arrays provided 24 and 15 times more heat flux for vertical and horizontal cavity orientation, respectively. Horizontal orientation was more favored because it generated nearly uniform heat transfer from the source. A porous medium model was also developed to simulate fluid flow and heat transfer from a dense array of parallel plate fins mounted to on wall of a vertical cavity. Fluid penetration and heat transfer was found to increase within the porous regions as the applied power (modified Rayleigh number) increased. Numerical predictions were in reasonable agreement with experimental results for the vertical orientation, with the Brinkman-Forchheimer-extended Darcy model following the data more closely than the Brinkman extended Darcy model. AlEssa et al. (2008) numerically examined the heat transfer enhancement from a horizontal rectangular fin embedded with triangular perforations under natural convection (see Figure 11). The fin?s heat dissipation rate was compared to that of an equivalent solid one. Several conclusions were drawn: i) The temperature drop along the perforated fin length is consistently larger than that on an equivalent non-perforated fin. 33 ii) For certain values of triangular dimensions, the perforated fin can enhance heat transfer. The magnitude of enhancement is proportional to the fin thickness and its thermal conductivity. iii) The extent of the heat dissipation rate enhancement for perforated fins is a complicated function of the fin dimensions, the perforation geometry and the fin thermophysical properties. iv) The gain in the heat dissipation rate for the perforated fin is a strong function of both the perforation diameter and lateral spacing. This function attains a maximum value at a given perforation diameter and spacing, which are called the optimum perforation dimension bo, and the optimum spacing Syo, respectively. v) The perforation of fins enhances the heat dissipation rates and at the same time decreases the expenditure for fin materials. Figure 11: Fin with equilateral triangular perforations to enhance natural convection (Source: AlEssa et al. (2008)) 34 Van Lear and Sparrow (2010) developed a numerical simulation code to study an active enhancement device for natural convection in the interfin spaces of a fin array (see Figure 12). A baseline solution for the non-enhanced situation revealed that the confinement created by the walls of adjacent fins and the base surface contributed to a drastic reduction of the HTC values compared with those for the standard vertical plate. Enhancement was achieved by alternately introducing and extracting air into and from the space. The frequency of introduction/extraction cycle was varied over values of 0, 10, 50 and 100 Hz. Even at a low oscillation frequency of 10 Hz, the interfin HTCs were significantly enhanced but not sufficiently to overcome the confinement effect. At 100 Hz, the enhancement gave rise to coefficient values that about 64 times greater than the unenhanced values. Figure 12: Schematic diagram of the physical system (Source: Van Lear and Sparrow (2010)) 35 Figure 13: Local cycle-averaged Nusselt number results for the surface of the fin facing the neighboring fin. Curve designation: (A) baseline, no oscillation, (B) 10 Hz, (C) 50 Hz and (D) 100 Hz. (Source: Van Lear and Sparrow (2010)) Tsuji et.al (2007) conducted an experimental study (Figure 14) on heat transfer enhancement for a turbulent natural convection boundary layer in air along a vertical flat plate by inserting a long flat plate in the span-wise direction (simple heat transfer promoter) and a short flat plate aligned in the span-wise direction (split heat transfer promoter) with clearance into the near-wall region of the boundary layer. For the simple heat promoter, the HTC increases by a peak value of approximately 37% in the downstream region of the promoter compared with those in the usual turbulent natural convection boundary layer. For the split heat transfer promoter, the enhancement was approximately 60% in the downstream region of the promoter. 36 Figure 14: Experimental apparatus, simple heat transfer promoter and split heat transfer promoters (Source: Tsuji et al. (2007)) Figure 15: Visualized fluid motions in the downstream region of a split heat transfer promoter (x-y plane, Nz = 5) (Source: Tsuji et al. (2007)) 37 Figure 16: Visualized fluid motions in the downstream region of a split heat transfer (y-z plane, Nz = 4) (Source: Tsuji et al. (2007)) 1.3.4 Literature Review on View Factor Calculation As described in the previous chapter, thermal comfort can be valued in terms of operative temperature. Eq. (1) points out that the mean radiant temperature (MRT) should be evaluated as part of the effort to obtain operative temperature as well as air temperature. The MRT is a concept arising from the fact that the net exchange of radiant energy between two objects is approximately proportional to their temperature difference multiplied by their ability to emit and absorb heat (emissivity). MRT is simply the area weighted mean temperature of all the objects surrounding the body. This is valid as long as the absolute temperatures of objects in question are large compared to the temperature differences, allowing linearization of the Stefan- 38 Boltzmann Law in the relevant temperature range. Technically, MRT is defined as the uniform temperature of a surrounding surface giving off blackbody radiation (emissivity e = 1) which results in the same radiation energy gain on a human body as the prevailing radiation fluxes which are usually very varied under open space conditions. Eq. (19) formulates the expression for the MRT, which is the summation of surface temperature multiplying by a view factor from a person to the surface. tr=?Fp-itsi (19) where, is the MRT, is the view factor (angle factor, configuration factors, form factors or shape factors) from a person to surface i, and is the surface temperature. The view factor in the Eq. (19) refers to the proportion of all that radiation which leaves person and strikes surface p. Figure 17 and Eq. (20) explain the calculation of the view factor between two infinitesimal surfaces, where dA1 and dA2 are two infinitesimal surfaces, and s is the distance between the two surfaces. The angle between line s and the surfaces dA1 and dA2 are and , respectively. 39 Figure 17: Schematic diagram of view factor between two infinitesimal surfaces FdA1 dA2= ? cos 1cos 2 s2 A1 A2 dA1dA2 (20) Because of the double surface integral in the Eq.(20), analytical solutions of view factor between two finite surfaces can be difficult, or sometimes impossible to obtain, if the geometries of the surfaces are complicated such as spheres, cylinders. Dunkle (1963) calculated the view factors from the inner-wall of a room to an occupant who either stands or sits inside the room. In order to simplify the complicated surface integral of a human body, the body was treated as a sphere. The equivalent sphere radii of the standing occupant and the seated occupant are expressed by: R2=0.65+cos (0.715+0.52|cos |) (21) R2=1.365+(0.2+0.673sin )cos cos (22) where ? is the vertical angle from horizontal to point on surface, ? is the azimuth angle between direction faced and point on surface. dA1 dA2 s 40 Eq. (21) refers to the radius of a standing person and Eq. (22) refers to the radius of a seated person. The two view factors which from a point on a wall to the two spheres are provided as below. Eq. (23) refers to the view factor of a standing person, and Eq. (24) refers to the one of a seated person. f= R2D (x2+y2+z2) 3 2 (23) f= R2Dh? 1 3 30.8(x2+y2+z2) 3 2 (24) Feingold and Gupta (1970) found the analytical solution to evaluate the view factors from a sphere to a coaxial disk; from a sphere to an infinitesimal area lying in a plane which does not intersect the sphere; from a sphere to a segment of a coaxial disk, from a sphere to a coaxial rectangle; from a sphere to a coaxial right circular cylinder; from a sphere to a Polygon; from a sphere to a noncoaxial disk. Sabet and Chung (1987) proposed a general Equation to calculate the view factor from a sphere to any nonintersecting planar surfaces: F2-1= d 2 ? f(x) (x2+d2?x2+,f(x)-2+d2) dx x2 x1 (25) where: 2 is the sphere surface; 1 is any nonintersecting planar surface; d is the distance of the center of the sphere to the plane; f(x) is a characteristic function depending on the shape of the planar surface. 41 1.3.5 Literature Review of Multi-mode Heat Transfer In the field of air-conditioning and refrigeration, the combination of radiative and natural convective heat transfer was first studied to investigate the performance of wire-and tube condensers of refrigerators. Tagliafico and Tanda (1997) investigated the air-side heat transfer from wire- and-tube heat exchangers in the application of refrigeration. The radiation heat transfer between refrigerant and ambient air was modeled using a diffuse, gray-body network method. The natural convection part was modeled by a semi-empirical correlation. It was reported that the fractional contribution of convection to the combined-mode heat transfer was between 40% and 70% depending on the temperature difference and heat exchanger configuration (pitch-to-diameter ratio of the wires, pitch-to-diameter ratio of the tubes and normalized heat exchanger height). Nu=0.66 ( RaH dt ) 0.25 *1- *1-0.45 ( dt H ) 0.25 + exp (-sw/ ) (26) where = . C1 H / 0.4 sw 0.9st -1.0 + . C1 H / 0.8 0 C2 Tt-T 1 0.5 sw -1.5 st -0.5 s1= p t dt dt ;sw= p w dw dw ;C1=28.2 m;C2=264 K Melo and Hermes (2009) proposed a more complicated correlation by considering more parameters, Eq. (27), and Table 3 describes the dimensionless used in the equation. 42 0=5.68 1 0.6 2 0 28 3 0.49 4 0.08 (27) Table 3: Dimensionless group and description in Eq. (27) (Source: Melo and Hermes (2009)) Dimensionless group Description 0=(hc+hr)/hr Combined heat transfer coefficient 1=Aw/(At+Aw) Heat transfer surface 2=(pt dt)/dt Tube spacing 3=(pw dw)/dt Wire spacing 4=(tavg tair)/tfilm buoyancy Bansal et al (2003) developed a model using FORTRAN 90 code and simulated the wire-and-tube condenser under different ambient conditions. The modeling results show that the dominant heat transfer mode for wire-and-tube condenser is by convection, which contributes up to 65% of the total heat transfer. Gupta et al (2008) improved Bansal?s model by considering the effects of aluminum tape to a hot-wall condenser. Rao et al. (2006) focused on a more fundamental study whose subject were only two fins in an enclosure (Figure 18). Alternating direction implicit (ADI) method was used to solve the governing equations. Isotherms and stream lines were obtained in the paper as well as the average Nusselt number and a fin effectiveness correlation. 43 Figure 18: (a) Physical model of two-fin enclosure. (b) Different surfaces in radiation enclosure (Source: Rao et al. (2006)) Kuznestov and Sheremet (2008) numerically studied the convective-radiative heat transfer in an enclosure having finite thickness heat-conducting walls and a heater at the bottom. Air (Pr = 0.7) was the fluid inside the enclosure, and the Grashof number in the problem was varied from 105 to 107. Isotherms and streamlines were obtained and the influence of some parameters, such as the Grashof number, the transient factor, the optical thickness and the heat conductivity ratio, on formation of thermo-hydrodynamic modes was analyzed. It was determined, that taking into account of the radiative heat transfer leads to the temperature increase in the gas cavity on the average of 11% at 0 < <200. Talukdar (2004) studied the multi-mode heat transfer in a porous channel bounded by isothermal parallel plates. Chiu (2007) investigated the problem in rectangular ducts rotating about a parallel axis. Both Chiu et al. (2007) and Premachandran (2006) studied the problem in a horizontal channel. 44 Krishnan et al. (2004) conducted an experimental and semi-experimental investigation on steady-state natural convection and surface radiation between three parallel vertical plates with a hot plate in the middle and the other two unheated ones each side. The radiative heat transfer (in terms of Nusselt number) calculation was conducted by the radiosity-irradiation method, and the convective heat transfer (in terms of the Nusselt number) was obtained from an experiment featuring six plate spacing ranging from 12.66 to 52.2 mm and for an order of magnitude range of wall- to ambient temperature difference. It was concluded that even at low temperature, 310K, the significance of radiation heat transfer rate cannot be ignored. A correlation for the average convective Nusselt number was also developed at the range of 2,370 2,500 kJkg-1, Gao et al. 2005), which varies according to desiccant material, but is usually higher than the heat of evaporation of water vapor (~ 2,500 kJkg-1). The difference of these two forms of heat added an additional heat load to the vapor compression cycle, which reduced the effective cooling from the vapor compression cycles. Test data indicated that the extra heat load was around 300 W, with the resulting effective cooling at 3.2 kW and the COP of SSLC systems reduced by 8%. Second, in order to provide the regeneration temperature of 50?C, the airflow rate through the condenser (or gas cooler) needs to be reduced from 0.37 m3s-1 (baseline system) to 0.25 m3s-1. Such a reduction raised the condensing pressure (high side pressure) of both refrigerants. As described in Table 4, the high side pressure of CO2 increases from 10.37 MPa to 10.91 MPa, and the condensing pressure of R-410A increases from 3.16 MPa to 3.42 MPa. Excessive high side pressure increases the compressor input and in turn decreases the COP. The first negative factor is difficult to eliminate because it results from the inherent characteristic of desiccant material. However, the impact of the second factor could be minimized. We hereby propose the application of divided condensers (or gas coolers). Instead of heating the entire amount of air through a heat exchanger, (referred to in this dissertation as a condenser or gas cooler) to a required regeneration 79 temperature, divided heat exchangers use only one section (the first section) to provide hot air for DW regeneration, while the other sections (second section or third section) are used for heat rejection only, without meeting the temperature requirement for regeneration. Therefore, the refrigerant high side pressure is restrained while the system is still able to effectively regenerate the DW. Furthermore, as a common practice to improve the performance of the CO2 cycle, the addition of a suction line heat exchanger helps reduce the refrigerant temperature at the gas cooler outlet. Therefore, its integration with the SSLC system was also investigated. Figure 33: The refrigeration cycle of the DW-assisted SSLC system with divided HXs Figure 33 describes the refrigeration cycle of the new SSLC system with divided heat exchangers (HXs). The condenser (gas cooler) in the vapor compression cycle was divided into two or three sections. The refrigerant that discharged from the compressor entered the first part of the condenser (or gas cooler), and then entered the second and third parts in sequence. The ambient air flowing to the first part of the heat exchanger was sent directly to the DW for regeneration. The ambient air or Evaporator Compressor 1st. Condenser (1st. Gas cooler) 2nd. Condenser (2nd. Gas cooler) 3rd. Condenser (3rd. Gas cooler) Expansion valve 80 exhaust air from the space served as a heat sink for the refrigerant in the second and third parts of the heat exchanger, after passing through the evaporative cooling process. The refrigerant leaving from the third part was sent to the expansion device. Some of the options listed below do not have the third part heat exchanger, which is shown in dashed rectangular box, so the refrigerant leaving the second part heat exchanger was sent to the expansion valve directly. Figure 34: New DW-assisted SSLC system option 1: evaporative cooling Return air from the space Sensible evaporator Supply air to the space DW Condensers (Gas coolers) Ambient air Exhaust air Ambient air through evaporative cooling 5 6 7 1 23 4 81 Figure 35: New DW-assisted SSLC system option 2: evaporative cooling + fresh air Both Figure 34 and Figure 35 describe the DW-assisted SSLC systems. Option 1 is a zero-ventilation system, and the condensers (or gas coolers) was divided into two parts. Option 2 has the required amount of fresh air for the capacity, and the condensers (or gas coolers) were divided into 3 parts. Each part faced different air conditions, which varied from the ambient condition (35?C, 44% RH), ambient air condition after the evaporative cooling process (24.8?C, 100% RH) to the exhaust air from the space after the evaporative cooling process (19.5?C, 100% RH). Return air from the space + fresh air Sensible evaporator Supply air to the space DW Condensers (Gas coolers) Ambient air Exhaust air Ambient air through evaporative cooling Exhaust air from the space through evaporative cooling 1 4 23 5 6 7 8 9 82 Figure 36: New DW-assisted SSLC system option 3: EW + evaporative cooling + fresh air Figure 36 shows the option with an added enthalpy wheel (EW) into the system. The heat and mass transfer between hot-and-humid ambient air and cool-and-dry indoor exhaust air helped recover both the sensible and latent cooling from the space. Figure 37: New DW-assisted SSLC system option 4: SW + evaporative cooling + DOS application Sensible evaporator Supply air to the space DW Condensers (Gas coolers) Ambient air Exhaust air Ambient air through evaporative cooling EW Fresh air Pre-conditioned fresh air Return air from the space 1 23 4 1 5 6 7 8 Sensible evaporator Supply air to the space DW Condensers (Gas coolers) Ambient air Exhaust air Ambient air through evaporative cooling Fresh air Return air through evaporative process SW 1 1 23 4 5 6 7 9 83 Figure 38: New DW-assisted SSLC system option 5: EW + evaporative cooling + DOS application Figure 37 and Figure 38 describe two new applications for such a configuration on a dedicated outdoor system (DOS). In the DOS, only fresh air is pre-conditioned to such a level that the temperature and humidity ratio are the same as those of the indoor air. Moreover, the application of the enthalpy wheel (EW) and the sensible wheel (SW) were compared to each other in order to find out which component is more suitable for the DOS. In the configuration of SW, one of the air streams was outside fresh air and the other was return air after the evaporative cooling process. Since there was only sensible heat transfer through the wheel, the evaporative cooling provided a large heat recovery potential without adding any water vapor to the fresh air stream. On the other hand, in the configuration of EW, no evaporative cooling process was added. Sensible evaporator Supply air to the space DW Condensers (Gas coolers) Ambient air Exhaust air Ambient air through evaporative cooling Fresh air Return air from the space EW 1 1 23 4 5 6 7 84 2.2.4 Improved System Modeling Approach EES was again used to model the vapor compression cycle and the DW. CoilDesigner (Jiang, 2006), an in-house heat exchanger simulation software package, replaced the previous multi-segment UA-LMTD method to model all the heat exchangers in the vapor compression cycle. The built-in optimization tool in EES was utilized to optimize the system COP. The detailed assumptions adopted for calculating the vapor compression cycle are listed as follows: The integration between EES and CoilDesigner: For each heat exchanger calculation, a database was created by a multiple-variable parametric study in the CoilDesigner. Specifically, the database of evaporator (evaporating pressure, inlet quality and mass flow rate) were selected as variables, and for the database of condenser (gas cooler) (condensing pressure (gas cooling pressure), inlet temperature, and mass flow rate) were selected as variables. Each database had 1,000 records. EES imported all the records and saved them in 3-dimension arrays. Linear interpolation method was applied to calculate the results from the database records. Optimization approach: The EES built-in optimization tool was used to maximize the system COP. Because of the nature of the system model, which was non-linear and the possible existence of multiple local maximums, the genetic method was applied for the optimization. The optimization function was defined as, Min f=-(COP of the system) (34) s.t. : system capacity 3,800 W (SSLC system) or 1,820 W (DOS system) Air discharge temperature off the 1st condenser (GC) 50?C 85 The normalized objective function with penalty factor was demonstrated below, f=- COP baseline COP +rp*(max . -Qeva+3800(1820) 3800(1820) 0/+max . -tairoff+50 50 0/ ) (35) where: Baseline COP = 3, rp(penalty factor) = 1,000 The following assumptions were made for modeling the vapor compression cycle: ? System capacity: 3.5 kW (SSLC), 1.82 kW (DOS) (smaller capacity because it only conditions ventilation air flow) ? Sensible heat factor: 0.7 ? Latent capacity: 1.0 kW ? Outdoor/indoor air conditions: 35?C, 44% RH / 27?C, 50% RH ? Refrigerant: R-410A and CO2 ? Regeneration temperature: 50?C ? Regeneration air flow rate: 0.15 m3/s ? Indoor air flow rate: 0.41 m3/s ? Ventilation air flow rate or air flow rate of DOS system: 0.082 m3/s ? Total air flow rate of condensers (gas coolers): 0.42 kg/s (volume flow rate varied in the condition with or without evaporative cooling) Compressor modeling: The compressor?s discharge temperature and power input were calculated from isentropic efficiency, compressor efficiency and volumetric efficiency. The three efficiencies were functions of pressure ratio, degree of superheating, and compressor frequency. Theoretical work input and the functions were obtained by regression analysis on experimental data. 86 ise =f1(PR, Tsup)= hdis,ise hsuc hdis hsuc (36) vol =f2(PR,Tsup,compfreq)= m? RPM*disp* ref (37) comp =f3(theoretic work input)= theoretic work real work (38) HX modeling: all the heat exchanger calculations were conducted in CoilDesigner. Expansion device modeling: The expansion process in the vapor compression cycle was treated as isenthalpic. hin=hout (39) DW modeling: The DW was also modeled in the EES. The dehumidification performance was assumed to be a function of DW rotation speed, regeneration temperature, air velocity through the wheel and inlet humidity ratio in the regeneration side. The enthalpy of air in the process side off the DW was calculated as a function of DW rotation speed, regeneration temperature and air velocity. All the input data described above were from experimental data. WVR=f4(Treg,RPH,Vair ?regin) (40) hgain=f5(Treg,RPH, Vair) (41) SW and EW modeling: both of them were treated as counter-flow heat exchangers in modeling. A free heat exchanger software package (Heatex Select, 2009) was used to calculate the efficiencies of both wheels. For the SW, since the sensible heat was 87 transferred between two air streams, the sensible heat transfer efficiency was defined to be 0.85. For the EW, both sensible and latent heat were transferred between two air streams, and the efficiencies of temperature (sensible heat) transfer and humidity (water vapor) transfer were defined to be 0.82 and 0.65, respectively. 2.2.5 Modeling Results and Discussion Figure 39: Psychrometric process of new SSLC system option 1 88 Figure 40: Psychrometric process of new SSLC system option 2 Figure 41: Psychrometric process of DW-EW-assisted SSLC system (option 3) 89 Figure 42: Psychrometric process of DW-SW-assisted DOS system (option 4) Figure 43: Psychrometric process of DW-EW-assisted DOS system (option 5) 90 Figure 39 through Figure 43 demonstrate the psychrometric processes of the aforementioned options. Different line styles represent different air processes. In those figures, the bold numbers represent each air state point corresponding to ones in their respective schematic drawings. Figure 44: The system COP comparison of different SSLC options and baseline system In Figure 44, the system COPs of different options were compared with the convectional systems and DW-assisted SSLC system with single condenser (gas cooler). The system COP was defined as the ratio of the space cooling capacity (3.5 kW) to the compressor input. For conventional systems, the system COP could be considered same as the vapor compression cycle COP when neglecting heat load of fans, which is defined as the ratio of the evaporator air-side capacity to the compressor input. However, for the DW-assisted SSLC system, the vapor compression cycle provided an extra 300 W cooling to compensate for the difference 2 3 4 5 6 7 8 9 10 Baseline SSLC baseline Option 1 Option 2 EW assisted SSLC Sy st em C O P R410A systems CO2 systems 91 of heat of adsorption and heat of evaporation. After conducting the optimization, both the previous conventional systems and DW-assisted SSLC systems had slight improvements. In the conventional systems, the system COP of R-410A system improved from 3.6 to 3.7 and the system COP of CO2 system improved from 2.6 to 2.7. For the SSLC systems with single condenser (gas cooler), R-410A system improved from 3.9 to 4.1 and CO2 system improved from 2.8 to 2.9. All these system COP results were plotted in Figure 44. In order to demonstrate the effect of the application of divided heat exchangers, the CO2 vapor compression cycles were plotted in the P-h diagram as shown in Figure 45. Figure 45: P-h diagram of CO2 DW-assisted SSLC systems (1?: compressor outlet, 2?: gas cooler outlet, 3?: expansion device outlet, 4?: evaporator outlet 1: compressor outlet, 2: 1st gas cooler outlet, 3: 2nd gas cooler outlet, 4: expansion device outlet, 5: evaporator outlet) -300 -250 -200 -150 -100 -50 4x10 3 10 4 2x10 4 Enthalpy (kJ/kg) P re ss ur e (k P a) 42?C 37?C 34?C 0.2 0.4 0.6 0.8 CarbonDioxide DW-assisted SSLC without divided gas coolers DW-assisted SSLC with divided gas coolers 1'2' 3' 4' 1 23 4 5 92 The dotted cycle represents the DW-assisted SSLC system without gas cooler being divided, and the solid cycle represents the one with divided gas coolers. It clearly shows that the application of divided heat exchangers reduce the high side pressure and also reduce the approach temperature. In more detail, the divided gas coolers reduce the high side pressure from 10.4 MPa to 9.7 MPa, and the refrigerant temperature off the gas cooler is reduced from 42?C to 38?C. For the R-410A system, the pressure reduction is insignificant (3.46 MPa to 3.45 MPa); however, the condenser outlet temperature reduction is significant (from 40?C to 36 ?C). The hollow legends in Figure 44 represent the case of option 1 without evaporative cooling in the second HX. It was found that the COP improved by 20% for the R- 410A system and 44% for the CO2 system. It is also interesting to notice that when the evaporative cooling is applied to the second gas cooler (solid legends); the CO2 system had a better COP than the R-410A system. The COP of the CO2 systems tends to decrease quickly under high ambient temperature conditions; however, the application of the evaporative cooling lowered the actual operating conditions of the system. Therefore, the COP improvements of the CO2 systems were larger than those of the R-410A systems. Option 2 divides the condensers (gas coolers) into three parts, and the evaporative cooling of return air provides the lowest possible temperature to further cool down the refrigerants. The COP had an extra 4.8% and 16.8% improvement for the R-410A and CO2 systems, respectively. Again, the CO2 system outperformed the R-410A system because the same reason explained above. The EW- assisted SSLC system had the highest COP as shown in Figure 44. Application of the EW provided ?free? sensible and latent cooling, and reduced the sensible cooling 93 requirement of the vapor compression cycle from 3.8 kW to 2.7 kW. The heat exchangers were modeled in reduced sizes in proportion to the reduced capacity. The compressor power input was reduced to a minimum with this option. The system COP improvements were 68% and 73% for the CO2 and R-410A systems, respectively. Table 10: Optimization results of SW-DW-assisted SSLC system Refrigerant R-410A CO2 Vapor compression cycle COP 5.09 3.64 Cooling capacity of VCC (kW) 1.25 1.25 Power input (kW) 0.24 0.34 Regeneration temperature (?C) 48.21 48.19 The ratio of heating to cooling 1.12 1.12 Sensible wheel capacity (kW) 1.27 1.27 System cooling capacity (kW) 1.82 1.82 Table 10 lists the optimization results of the SW-DW-assisted DOS system. The vapor compression cycle COP is defined as the ratio of cooling provided by the VCC over the VCC power input. Unlike aforementioned system COP which adds the cooling effect from cooling recovery device, like EW or SW, the VCC COP only considers the cooling from VCC system. For both refrigerant systems, the most important information was that they could not provide enough heat (to reach 50?C) to regenerate the DW. The table shows the maximum regeneration temperatures are only around 48?C for the two systems. Since the SW did not provide any latent cooling, the DW had to deal with the entire latent load (1 kW). However, the vapor compression cycle had a reduced cooling capacity (from 1.8 kW to 1.25 kW), enhanced by the fact that the SW provided a sensible cooling. Compared with the required amount of heat to regenerate the DW (1.4 kW), the reduced cooling capacity 94 output from the vapor compression cycle was too small to maintain the COP at a high level while still rejecting enough heat to the ambient air. If the system was forced to deliver the required amount of heat to the DW, the system COP would be lower than the conventional system, rendering ineffective the point of using the separate sensible and latent cooling technology. The EW-DW-assisted system, on the other hand, worked well. Table 11: Optimization results of EW-DW-assisted SSLC system Refrigerant R-410A CO2 Vapor compression cycle COP 5.80 4.54 Cooling capacity of VCC (kW) 0.80 0.81 Power input (kWe) 0.14 0.18 Regeneration temperature (?C) 50.06 50.01 The ratio of heating to cooling 0.95 0.95 Enthalpy wheel capacity (kW) 1.32 1.32 Latent capacity of EW (kW) 0.65 0.65 System cooling capacity (kW) 1.82 1.82 Table 11 lists the optimization results of such a system. The EW provided both the sensible and latent cooling to the system, helping to reduce both the cooling load of the vapor compression cycle and the amount of heat required to regenerate the DW. When comparing the cases of using an EW against using a SW, it was found that the ratio of heating to cooling could serve as a measure to determine whether or not the configuration was suitable for the DW-assisted SSLC system. If the ratio was greater than 1, meaning the heat requirement of the regeneration is greater than the 95 cooling output of the vapor compression cycle; such a configuration would not be suitable because the regeneration heat from the cooling system is not large enough. 96 Chapter 3: Experimental Assessment for the Low ?T Heat Exchangers 3.1 Chilled Ceilin Panels Heated Floor Systems and the Low ?T Heat Exchan er In order to assess the thermal comfort provided by the SSLC system, it is necessary to focus the study onto its indoor heat exchanger. The previous study shows that one of the major design challenges for the indoor heat exchangers is larger air- side pressure drop than that of conventional ones. It is because the air flow rate through the sensible heat exchangers has to be typically 3 to 4 times higher than that that through conventional systems in order to compensate the reduced air enthalpy difference across the heat exchanger. As a solution to this challenge, the indoor heat exchangers are designed to have a larger frontal area; so that the air velocity can be reduced. The larger frontal area has another advantage. It makes the radiative heat transfer between the heat exchanger and occupants more significant than that of conventional systems. Such heat transfer helps to adjust the mean radiant temperature (MRT), and hence provides better thermal comfort. In some European and Asian countries, the idea has been applied to products like chilled ceiling panels (cooling) and heated floor systems (heating). Chilled ceiling panels are typically suspended on the ceiling or plenum which refers to a drop ceiling. The panels consist of one or multiple serpentine-shaped tubes fixed on metal sheets. The tubes can be made of copper because of a better heat conductance so that the heat from the tube can be better transferred to the sheet. However they are more often made of PEX, cross-linked polyethylene, for the easy 97 handling and lower cost. The metal sheet is typically made of aluminum. Paint may be applied on the surface for a better emissivity. The working fluid flowing inside the tubes is water which is in the temperature range from 14?C to 18?C. Since condensation should be avoided, the fluid temperature has to be 1 or 2 K higher than the dew point. Therefore, for the entire air-conditioning application, certain latent load removal system has to be installed such as desiccant wheels. Riffat et. al. (2004) provides a detailed review on the origination, development and current state of the chilled ceiling panels. The heated floor system instead provides hot water inside the tubes which are embedded under floors. The two papers from Bean et al. (2010a, 2010b) offer a review of the development of heated floor systems in both Asia and Europe. It is widely accepted that both the chilled ceiling panels and heated floor systems have the benefits of low energy consumption and better thermal comfort. However, there are drawbacks to these two systems. First of all, as restricted by the motion of buoyancy airflow, chilled ceiling panel cannot effectively provide heating while heated floor system cannot effectively provide cooling. Therefore, each system cannot replace the other with the same function. Second, especially for the heated floor systems, the installation requires an entire overhaul of the floor in order to embed the system. Such overhaul work is almost impossible without affecting the residents. Therefore, the systems are generally limited to be installed for new houses. However, considering the fact that in the US, the number of retrofitting old houses is 98 at least three times more than that of building new homes, it is obvious that the market for these two systems are low. The idea of low ?T heat exchanger is proposed hereby in order to address the above two drawbacks. The low ?T refers to the small temperature difference between indoor air and refrigerant, i.e., in winter the low ?T heat exchanger uses hot fluid of temperature just several degrees higher than room air temperature to provide heating; and in summer it uses cold fluid of temperature just several degrees lower than room air temperature to provide cooling. The structure of the heat exchanger is similar to chilled ceiling panels. It consists of serpentine tubes and metal sheets. However, unlike the installation of a chilled ceiling panel which has to be suspended on the ceiling, the low ?T heat exchanger can be hung on the wall. It also has a large frontal area that can cover as much as the entire wall or even multiple walls so it can effectively provide radiative heat transfer as well as convective heat transfer. Since the heat exchanger can be installed on the wall, the temperature difference between the heat exchanger wall and the opposite wall will drive a ring-shape air motion due to natural convection. Hence, there is no restriction for it to provide both heating and cooling. Moreover, the installation of heat exchanger against the wall is much more convenient. It does not require any overhaul to the existing structure of the house. The convenience makes it possible to apply the low ?T heat exchangers and low temperature lift heat pump systems onto retrofit market. In order to test the performance of the low ?T heat exchangers and to prove the potential for better thermal comfort, a series of experiments have been conducted. 99 The experiments started with the fabrication of operative temperature sensor, followed by a baseline system test, and finally the test of low ?T heat exchangers. 3.2 Sensors for Operative Temperature Measurement 3.2.1 A Simplified Operative Temperature Calculation According to Eq. (1), the operative temperature is defined as the heat transfer coefficient - weighted arithmetic mean between air temperature and (MRT). However, the direct measurement of either radiative heat transfer coefficient or convective heat transfer coefficient is cumbersome; hence proper simplifications to the equation have to be made. Table 12 lists a simplified calculation method by ASHRAE standards 55 (ASHRAE 2004). It is assumed that (1) convective heat transfer coefficient depends on air velocity; (2) when the velocity is small, the convective heat transfer coefficient is comparable to the radiative heat transfer coefficient. Table 12: Operative temperature calculation (based on ASHRAE 2004) Air speed v < 0.2 m/s 0.2 v 0.6 m/s v>0.6 m/s Correlation top=0.5Tair+0.5MRT top=0.6Tair+0.4MRT top=0.7Tair+0.3MRT 100 3.2.2 Operative Temperature Sensors The operative temperature (OT) is measured with self-built sensors. The sensors are designed according to the ISO standard 7726 (ISO 7726:1998), which provides a guideline for OT sensor?s design. Each one consists of a resistive temperature sensor (RTD), which has an uncertainty of 0.15 K, inserted into the center of a hollow aluminum sphere. The sphere of 0.152 m in diameter was welded from two halves and polished on the outer surface. Multiple layers of black paint having an emissivity of 0.95 were then uniformly applied on the polished surface. The emissivity of 0.95 is selected because it is close to the emissivity of a human body. Figure 46 and Figure 47 are the pictures of the RTD sensor and the assembled operative temperature sensor, respectively. Figure 46: Picture of RTD sensor 101 Figure 47: Picture of assembled operative temperature sensor The four assembled operative temperature sensors were later attached to a vertically-placed pole at different heights. Figure 48 provides detailed positions of the four sensors. The selection of the different positions is based on the height of an ordinary person so that different sensors can record the operative temperature from the feet to the head. The RTD sensors are calibrated at both 0?C and 100?C. Figure 49 plots the operative temperature readings of four sensors in one day. OT 1 is located at 1.7 m, OT 2 is located at 1.1 m, and OT 3 and OT 4 are located at 0.6 m and 0.1 m, respectively. Because the average temperatures of exterior wall (21.9?C) and floor (23.6?C) is much colder than air temperature in the room (25.1?C), it resulted in lower MRT. Therefore, the OT sensor readings are lower than air temperature. Since OT 1 has the largest view factor to the exterior wall, followed by OT 2, consequently, the OT sensors readings are low. OT 3 has the least view factors to the exterior wall and floor, therefore the reading is highest. In Figure 50, the uncertainties of the sensors (error bars) are applied to the average values of data in Figure 49. Both 102 systematic error and random error were considered in the uncertainty analysis. The highest uncertainty, 0.36 K, came from OT 1, and the lowest uncertainty, 0.30 K, came from OT 3. Figure 48: Positions of four operative temperature sensors Figure 49: OT sensors readings in one day 103 Figure 50: Uncertainty analysis of operative temperature sensors 104 3.3 Low ?T Heat Exchan er Test Facility 3.3.1 The Hot Water Supply Loop Figure 51 describes the hot water supply loop for the low ?T HX test facility. A tank made of polypropylene panels is used as a reservoir for hot water. The water inside the loop is circulated by a Grundfos? circulator pump. The pump, having a 205 W power input, is capable of delivering 1.77 l s-1 water flow rate at zero pressure lift or 90 kPa pressure lift at zero flow rate. A 1,500 W electric heater was installed in the loop to reheat the water. The heater came with an analog manual temperature control dial but was later replaced with a solid state relay to achieve a better on and off control. The relay itself is controlled by the PID module of the data acquisition system. The components were connected together by PEX tubes. Three T-type thermocouples (systematic error of 0.5 K) were installed at different locations to obtain an energy balance between the heater output and the heating capacity of the low ?T heat exchangers. The detailed positions of these thermocouples are marked in the Figure 51. The water mass flow rate in the loop was measured by a MicroMotion? Coriolis mass flow meter. The mass flow meter was re- calibrated to a range of 25 to 100 g/s for a better accuracy of 2.5% reading. Finally, the water flow rate was adjusted through a needle valve to achieve the capacity control of the heat exchangers. 105 Figure 51: Schematic diagram of low ?T HX test facility 3.3.2 The Assembly of Low ?T Heat Exchanger Panels Two low ?T heat exchanger panels (which are called comfort panels in the catalog) were purchased from Uponor?. The heat exchangers consist of 60 aluminum sheets and PEX tubes of a total length of 75 m. Each sheet has a smooth back-side, but its front-side is divided into small channels so that the PEX tube can be clipped into channels in a serpentine shape. The dimension of each sheet is 1.22 m by 0.089 m. 34 sheets were combined together to make one heat exchanger, while the rest 26 sheets were combined together to form another one. Insulation foams were taped to fill the gaps between channels in order to prevent conduction loss (see Figure 52). The total frontal area of the two heat exchangers is 6.52 m2. According to the Uponor? catalog (Uponor, 2010), it should be able to provide up to 658 W cooling/heating. The capacity was determined by a maximum capacity output of 101 106 W/m2 at 11 K temperature difference. The amount of cooling/ heating is sufficient to cover the heating/cooling load in the test office (Koepke 2011). Figure 52: Low ?T HX?s sheet and tube To minimize the heat loss from the back, insulation foams of fiberglass were applied to the backside of the sheet. The applied foam has a thickness of 3.8 cm and a thermal conductivity of 0.042 W(m K)-1 (see Figure 53). Figure 53: Insulation for the backside panel of low ?T HX 107 Due to the restriction of office layout, the heat exchanger can only be installed on the north wall opposite to the window. The heat exchangers were assembled together by aluminum frames. The frame was made of 80/20? aluminum profiles with a side length of the square cross section of 0.0381 m. The total frame size per panel is 1.6 m 2.44 m. They were connected together by aluminum angles. Figure 54 is the picture of the final installation of low ?T heat exchangers inside the test office. Figure 54: Installed panels in the test office 3.3.3 Room Selection Among various criteria, the most important one for room selection is that its dimension should be as close as possible to that of the CFD model (3 m by 3 m). The CFD modeling will be discussed in detail in the following chapter. Based on this criterion, an office on the first floor of Potomac Building at College Park, MD campus was selected. The office is 2.95 m from north to south, 4.25 m from west to 108 east and 3.2 m in height. It has one frequent occupant, a desk, a chair, two cabinets, one PC, two monitors and two fluorescent ceiling lights. Table 13 shows the breakdown of components included in the load calculation. Figure 56 and Figure 57 demonstrate the results of load calculation for winter and summer, respectively. Figure 55: Sketch of the office under low ?T study Table 13: Load calculation components Load Components Summer Winter Exterior walls Heat gain Heat loss Exterior windows Heat gain Heat loss Interior walls Heat gain Heat loss Lighting Heat gain Heat gain Electronic appliances Heat gain Heat gain Occupants Heat gain Heat gain Ventilation Heat gain Heat loss 109 Figure 56: Heating load analysis of the test office Figure 57: Cooling load analysis of the test office 110 3.4 Low ?T Heat Exchan er Experiments Results 3.4.1 Baseline System Experiment Results The objective of conducting a baseline system test is to evaluate the thermal comfort (in terms of operative temperature) of the space provided by convectional fan-coil unit in winter or window-type AC unit in summer. It is also the purpose of the baseline system to be served as a comparison to the later low ?T heat exchanger system test. Due to the season restriction, only heating test has been conducted. One weekend was randomly picked from January and February of 2011 to perform the baseline test. The test lasted for 60 hours. The four OT sensors were placed in the center of the office. A relative humidity sensor was also attached to the pole which holds OT sensors, and it was placed at the height of 0.6 m. Four aluminum tape-shielded thermocouples (accuracy of 0.5 K) were located at the center of the heater, air outlet of the heater, window and outside of the office. The purpose of shielding the thermocouples is to minimize the impact from direct solar radiation. Figure 58 demonstrates the operative temperatures recorded by the four OT sensors. It is clear to find that the operative temperature readings are stratified. This is due to the air temperature stratification caused by the buoyancy force. The fluctuation of the readings mainly comes from the change of exterior wall temperature. This statement can be supported by the fact that the trend of fluctuation of each OT sensor 111 is almost identical to each other. Hence, the fluctuation must have come from the same source. There are two temperature peaks in Figure 58 (in red circles) worthy of a further investigation. It was found that the two peaks happened approximately at the same time in two consecutive days. The peaks are probably because during one time in a day, the direct solar radiation reaches one or all spheres (depending on the cloudiness) through the window. This assumption was further supported by Figure 59. It is clear that the two peaks of window surface temperature are coincident in time with the peaks in Figure 58. Figure 58: Operative temperature measurement in the baseline test 112 Figure 59: Window surface temperature variation during the baseline test The hot water to the convectional fan-coil unit comes from the campus central heating system. Its temperature is around 80?C. The heater is located 0.4 m away from the floor. A strong stratification of air temperature was anticipated due to the buoyance force. The strong stratification should also be reflected on the operative temperature measurement. Figure 60 plots the average operative temperature readings of each OT sensor during the test day. The average operative temperatures of OT sensors are 23.0?C, 22.6?C, 21.5?C and 19.9?C, respectively. The maximum temperature difference of 3.24 K is between OT 1 (1.7 m high) and OT 4 (0.1 m). It represents the fact that when using the conventional fan-coil unit, occupants are subject to cold feet and warm head with the temperature difference over 3 K. According to the ASHRAE standard 55, this should be considered as uncomfortable. Generally speaking, people feel thermally comfortable when the feet are warmer than 113 the head. Moreover, the temperature stratification between the head and the feet should not exceed 3 K. Figure 60: Thermal comfort analysis of the baseline test: operative temperature stratification The other problem from the use of conventional fan-coil units in winter is the temperature at lower body, especially the feet, can be too low. Figure 61 overlays the 60-hour operative temperature data onto a psychrometric chart. The blue quadrilateral in the psychrometric chart is the simplified ASHRAE thermal comfort zone. Most of the OT 4 data falls out of the comfort zone to its left side. It means that occupants may suffer too low temperature around their feet area, although the other parts of body are subject to comfortable temperatures. 114 Figure 61: Thermal comfort analysis of the baseline test: operative temperatures in the comfort zone In summary, the conventional fan-coil unit with hot water temperature of 80?C is able to provide enough heating capacity to cover the office load. However, in terms of thermal comfort, occupants can still suffer from too large temperature difference between the feet and the head, and the feet area temperature is lower than required by ASHRAE comfort zone. 115 3.4.2 Low ?T Heat Exchanger Experiment Results The experiment of low ?T heat exchanger was conducted after the baseline test. In order to eliminate the effect from the fan-coil unit, both the supply water valve and fan were shut off during the test. The objectives of the experiment are to evaluate the general performance of the low ?T heat exchangers and its hot water supply loop, to study the thermal comfort created by the low ?T heat exchanger and to provide experimental data for the future validation of simulation codes. The experiment started with filling distilled water into the water tank. The usage of distilled water is to reduce the possibility of fouling inside the tubes and HXs. The water pump was then turned on to fill water into the water loop system and the heat exchangers. Since the manufacturer?s catalog (Uponor, 2010) suggests a minimum water flow rate to be 62 gs-1 (1 GPM), the pump was adjusted for a water flow rate of 78.5 gs-1 (1.25 GPM). The water flow rates during the entire test were plotted in Figure 62. The water temperature was set to be 35 ?C, which is approximately 10 K above the space air temperature. Compared to the baseline system having a hot water temperature of 80?C, the ?T between hot fluid and air temperature was reduced by around 45 K. The thermocouples located at the heater inlet and outlet recorded the temperature readings which are plotted in Figure 63. The heater capacity can then be calculated based on the water flow rate and the temperature difference across the heater (see Eq. (42)). Without any heat loss, the heater capacity should be the same as the low ?T heat exchangers (see Eq. (43) for HX capacity calculation). The two capacities were plotted against each other in 116 Figure 64. The average capacities of the heater and the HXs are 527 W and 546 W, respectively. It results in an average deviation of only 3.7%. The small deviation reflects good measurement accuracy. Qheater=m? cp (theater out theater inlet) (42) Qheater=m? cp (tHX inlet tHX outlet) (43) Figure 62: Water flow rate variation during the low ?T HX 117 Figure 63: Heater inlet and outlet temperatures? variations Figure 64: Comparison of measured heating capacities of heater and HX 118 It is then important to study the thermal comfort created by the low ?T heat exchangers. In Figure 65, the average operative temperatures of each sensor during the 60-hour test period were plotted against the sensors? heights. The highest sensor (OT 1) has an average temperature of 24.3?C, while the lowest sensor (OT 4) has an average temperature of 22.3?C. Compared with the results of the baseline test, the average operative temperature readings from OT 1 and OT 4 was found to increase by 2.4 K and 1.3 K, respectively. The reason behind the increase is the elevated surface temperature. The north wall of the office had an average surface temperature of 23.2?C during the baseline test, while it increased to 35?C during the low ?T HX test. The increased surface temperature increased the MRT of all OT sensors; therefore the operative temperature readings were increased. Since OT 4 sensor had the lowest air temperature reading due to the buoyancy force, the increased MRT had the largest effect on its increase of operative temperature. Consequently, the temperature difference between occupants? head (OT 1) and feet (OT 4) was reduced to 2.0 K. This demonstrates a better thermal comfort in terms of a reduced head-to-feet temperature stratification. 119 Figure 65: Thermal comfort analysis of the low ?T HX test: Operative temperature stratification Figure 66: Thermal comfort analysis of low ?T HXs: operative temperatures in the comfort zone 120 Figure 66 overlays the entire OT sensors readings on the psychrometric chart. The blue quadrilateral represents a simplified thermal comfort zone according to the ASHRAE standard 55. Almost all measured points fall into the thermal comfort zone. The higher surface temperature of north wall increased the operative temperature reading of the OT 4 from the baseline test. Hence, the second issue of the baseline test, too low temperature of OT4, has been addressed. 121 Chapter 4: Modeling the Operative Temperature Field in an Office Setting 4.1 The Objectives of Operative Temperature Field Modeling The experimental test demonstrates encouraging results when using the low ?T heat exchanger in terms of obtaining better thermal comfort than the baseline fan- coil unit. However, there are several questions still unanswered. First of all, both systems show the temperature stratification with respect to the height. However, the test data provides only several data point. How to get a complete picture of temperature stratification elsewhere inside the room? Second, the operative temperature sensor measures only operative temperature, but how MRT affects the operative temperature in detail? Finally, how to extrapolate the results to other conditions? In order to answer all the above questions and make further investigation on the low ?T heat exchangers, the simulation of operative temperature field created by low ?T heat exchangers must be conducted. It should focus on the simulation of operative temperature and velocity profiles created by the low ?T heat exchangers in the air-conditioned space. The objectives of the modeling are to achieve a complete understanding of the thermal comfort zone created by low ?T heat exchangers from solving the physical governing equations; to expand the database of low ?T heat exchanger?s operation from experimental test. 122 The OT modeling started from the MRT calculation, followed by the modeling of air temperature and velocity inside a square enclosure and finally the combination of MRT and air temperature to obtain the operative temperature. 123 4.2 The Calculation of Mean Radiation Temperature (MRT) To calculate the operative temperature, the MRT has to be obtained besides the air temperature. The MRT can be calculated as: tr=?Fp-itsi (44) Where: Fp-i is the angle factor (view factor) from a person to surface i; is the surface temperature of surface i. From the above equation, it is clear that in order to get the MRT, we have to calculate the view factor from all solid surfaces, i.e., walls, floor, ceiling, to the person (occupant). Moreover, from Eq. (44), one can get a better interpretation of MRT as a view factor- weighted average surface temperature. The view factor is mainly affected by the distance between the interested point and the surface, and the size of the surface. The following discussion within the section focuses on the model of the MRT temperature calculation, the view factor calculation and finally the MRT calculation results 124 4.2.1 Model Description Figure 67 describes the radiation model adopted in the study. The room is assumed to be 3 m by 3 m by 3 m with four walls (front wall is not shown), a ceiling and a floor. The left wall has an indoor heat exchanger, and it is assumed that the heat exchanger make the entire wall be at a constant temperature. The dimension of the MRT model matches the model of the CFD simulation of air temperature, which will be discussed later in the chapter. The right wall has a window facing south whose area is smaller than the wall. The window has a higher temperature than the rest of the wall because of the solar radiation. The four walls, i.e., the frontal wall, the back wall, the left wall and the right wall, are assumed to have uniform temperatures. It was through the later literature review (Rohan et al., 2010) that an additional surface has to be added to the model. There is a heated-up area on the floor receiving the solar radiation through the window. The temperature of the area can be higher than the rest of the floor, and the size and temperature of the area vary depending on the solar altitude angle and azimuth angle. Figure 68 describes the detail of the sun light area calculation. For simplification, the person in the room is simplified to be a sphere. Dunkle (1963) defined the equivalent sphere radius of both a standing person and a sitting person. However, since the sphere?s radius is infinitesimal compared with the room dimension, the radius is neglected in the model. The sphere is free to be moved anywhere inside the room because there are actually no restrictions for occupant?s activity inside the room. The most important function of the model is to output the MRT of the sphere inside the room. 125 Figure 67: Adopted radiation model setup 126 Figure 68: Calculation of sun light area (where and are the solar altitude angle and azimuth angle, respectively) (Source: Rohan et al., 2010) 127 In order to obtain the mean radiation temperature anywhere in the room, we have to first obtain the view factor between the sphere (representing occupant?s activity area) inside the room and all the eight surfaces. The calculation of view factors started with the one between two infinitesimal areas, and then integrated the result to obtain the one between a sphere and a wall surface. The detailed steps are as follows: first, use Eq. (45) to calculate the view angle between two infinitesimal areas of the sphere and wall (see Figure 69). Then, integrate the two infinitesimal areas to the entire sphere and a quarter of the wall as in Eq. (46) (see Figure 70). Figure 69: Calculation of view factor between two infinitesimal areas F1-2= ? cos 1cos 2 s2 dA1dA2 A1A2 (45) Where 1 are the angle of line s to surface dA1 and dA2, respectively. d A1 d A2 s 128 Figure 70: View factor from a sphere to a non-intersected rectangular area ? ( )? (46) where.B1= b d B2= q d , = c d (Sabet and Chung, 1987) From Eq. (46), the view factor between a sphere and the entire wall can be calculated by summing up four view factors which are from the sphere to each quarter-wall. With different d and rectangular dimensions, one can calculate view factors from anywhere inside the room to all the walls. 129 4.3 Calculation of Air Temperature inside an Enclosure 4.3.1 Model Description Based on the literature review and previous radiation model, the enclosure model for natural convection calculation is defined as a square space with the left wall having a uniform temperature of Tc and a right wall having uniform temperature of Th. Both the floor and ceiling are assumed to be adiabatic. The governing equations, which are Eq. (47) to Eq. (49), and boundary conditions: Continuity equation: u? =0 (47) Momentum equation: u? t + u? u? =- P+ gj (T-Tc)+ 2u? (48) Energy equation: cp T t + cpu? T=? 2T (49) Boundary conditions: At the top and bottom walls of the cavity u=v=0 T n? =0 At the left wall u=v=0 T=TC At the right wall 130 u=v=0 T=TH Figure 71: Boundary conditions and dimensions of the adopted natural convection model In Figure 71, x and y are used to represent the horizontal and vertical direction, respectively. The side length of the square is L. The bottom left corner has the coordinate of (- L 2 - L 2 ), while the top right corner has the coordinate of ( L 2 L 2 ). 131 4.3.2 CFD Simulation For the airflow inside the square enclosure, its Rayleigh number reaches when the dimension of the square is over 1 m by 1 m. Therefore, the flow should be considered as turbulent flow. The previous literature review proves difficult to analytically solve such problem. Hence, a commercial CFD software package, Fluent (ANSYS, Inc., 2006), was chosen to model the natural convection inside square enclosure. Figure 72 shows the mesh generated by Gambit (ANSYS, Inc., 2011). The grid has 150 by 150 quad meshes with enhanced mesh density in the boundary layer to capture the complicated flow characteristics. To be specific, the boundary layer has the first row of 1 mm and the growth of 1.15, i.e., the entire depth of the boundary layer is 20 mm. The mesh is proved to be sufficient to solve the square of 1 m by 1 m case. The turbulence model used in the model is k-? SST model. Figure 73 is the screenshot of the viscous model GUI in Fluent. The isotherms and streamlines are plotted in Figure 74 and Figure 75. The model was later expanded to be a 3 m by 3 m square space. It is because the new dimension is much closer to the actual size of a room. The original mesh structure did not lead to a convergent solution. The reason is that an increased space dimension increases the size of a cell as well; therefore, the cell cannot capture the characteristics of the boundary layer. As a solution, the mesh was refined (see Figure 76). The new mesh provides a converged solution as plotted in Figure 77 and Figure 78. 132 Figure 72: 1st generation of mesh generated by Gambit? Figure 73: Screenshot of viscous model GUI in Fluent? 133 Figure 74: Isotherms of air in the enclosure (1 m by 1 m) Figure 75: Streamlines of air in the enclosure (1 m by 1 m) y x y x 134 Figure 76: 2nd generation of mesh generated by Gambit? Figure 77: Streamlines of air in the enclosure (3 m by 3 m) y x 135 Figure 78: Isotherms of air in the enclosure (3 m by 3 m) y x 136 4.3.3 Linear Curve Fit for CFD Results It can be found from Figure 78 that the temperature profile of bulk air in the enclosure can be simplified as a function of y axis (the opposite of g? direction) only. The temperature of air close to the side walls has a rapid temperature change, so different correlations have to be used to predict the air temperature near the cold wall and hot wall. Eqs. (50) through (52) are the correlations used to calculate the air temperatures inside the room. More specifically, Eq. (50) is used to calculate the bulk air temperature in the room (-1.48