Development and Control of a condensation system using Peltier Cells

Pages 100
Views 16
of 100
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Description
Development and Control of a condensation system using Peltier Cells António Maciel Thesis to obtain the Master of Science Degree in Electrical and Computer Engineering Examination Committee Chairperson:
Transcript
Development and Control of a condensation system using Peltier Cells António Maciel Thesis to obtain the Master of Science Degree in Electrical and Computer Engineering Examination Committee Chairperson: Prof. Name Supervisor: Professor Leonel Augusto Pires Seabra de Sousa Co-Supervisor: Ph.D. Ana Sofia Oliveira Henriques Moita External Co-Supervisor: Engenheiro Emanuele Teodori Member of Committee: Prof. Name Member of Committee: Prof. Name April 2014 ii iii To... iv Acknowledgements Acknowledgements I would like to thank v vi Abstract Abstract This work proposes the development and control of a condensation system using Peltier cells. The condensation system is used to study the feasibility of direct immersion cooling of a CPU. The design of an efficient controller requires a good understanding of the working principles of the Peltier cells. So, a secondary objective is to characterize the Peltier cells under real working conditions, namely their internal properties and the dynamic response. This is expected to be useful to provide alternative methods and guiding values to characterize the internal properties of the cells in works focused on the application and not on the fundamental research, as these methodologies are still sparsely reported in the literature. Such characterization of the system is important for a closed-loop control and is vital to an open-loop control, so it is an important task to accomplish. A PID controller was designed to control the amount of cooling the Peltier cells provide. From the experimental procedures followed, results show that the internal properties of Peltier cells are within the range described in the literature and follow the same behaviour. Its dynamic response allowed to gain a sensibility of the limiting values and how the system responded. The PID controller showed very good results that were within the requirements previously proposed. A final test of 72 hours under 100% of CPU Load was made to study the feasibility of direct immersion cooling and the condensation system. The results obtained show that the combination of direct immersion cooling with the condensation system developed is indeed feasible. Keywords Peltier cell, PID Controller, Direct Immersion Cooling, CPU, Arduino, Condensation vii Resumo Resumo Este trabalho propõe o desenvolvimento e controlo de um Sistema de condensação usando células de Peltier. O sistema de condensação é usado para estudar a viabilidade do arrefecimento de um processador por imersão directa. Para desenvolver um controlador eficiente, é preciso ter um bom conhecimento dos princípios de funcionamento das células de Peltier. Portanto, um objectivo secundário é a caracterização das células de Peltier em condições reais, nomeadamente as suas parâmetros internos e resposta dinâmica. Com isto é esperado obter-se algum conhecimento e valores de referência que permitam caracterizar os parâmetros internos da célula numa aplicação real e não apenas através de teoria. Esta caracterização é relevante para desenhar um controlador em malha-fechado mas é principalmente importante para desenhar um controlador em malha-aberta, portanto é uma caracterização relevante que deve ser feita. Um controlador PID foi desenvolvido para controlar o arrefecimento as células de Peltier produziam. Através da realização dos procedimentos experimentais, foram obtidos resultados que mostram que os parâmetros internos estão dentro da gama descrita na literatura e que têm as mesmas dependências em relação a certas variáveis. Além disso a resposta dinâmica permitiu ganhar uma maior sensibilidade de alguns valores limites e a resposta que sistema dava. Os resultados do teste do controlador PID demonstram que efectivamente o controlador consegue seguir muito fielmente a referência imposta tendo em conta os requisitos previamente definidos. Um teste final de 72 horas com a carga máxima do processador foi feito para demonstra a viabilidade do arrefecimento por imersão directa e sistema de condensação. Os resultados mostram que de facto que a combinação do sistema de arrefecimento por imersão directa com o sistema de condensação é de facto uma solução viável. Palavras-chave Células Peltier, Controlador PID, Arrefecimento Imersão directa, Processador, Arduino, Condensação viii Table of Contents Table of Contents Acknowledgements... v Abstract... vii Resumo... viii Table of Contents... ix List of Figures... xi List of Tables... xiii List of Acronyms... xiv List of Symbols... xv List of Software... xvi 1 Introduction Overview and Motivation Objectives Contents State of the Art Electronic Cooling Techniques Peltier Cells State of the Art Pool Boiling as a cooling technique Indirect and Direct Contact cooling Boiling Curve and regimes and Condensation Condensation Cooling configurations using phase change Peltier Cells Introduction Peltier cells in various applications: advantages and drawbacks ix 4.3 Thermoelectric Effect Thermoelectric Cooling Selection of the Peltier cells: Relevant characteristics and procedures Experimental Apparatus, Procedures and PID Controller Conceptual Design Central Processing Unit (CPU) Pool Peltier Cell Zalman CNPS7500-AlCu LED Fan Final Setup Final Pool Working fluid - HFE PID Controller Experimental Procedures Characterization of Peltier Cells Proportional Integral Derivative (PID) Controller Final Setup Results Characterization of the Peltier cells Dynamic Response Internal Resistance Seebeck Coefficient Thermal Conductance PID Controller Kp Ki Kd Final Setup Conclusions References Bibliografia x List of Figures List of Figures Figure 1.1 Conceptual Design... 4 Figure 3.1 Indirect Contact...15 Figure 3.2 Direct Contact...15 Figure Drew and Muller Experiment (Black) and Nukiyama Curve (Red arrows)...18 Figure 3.4 Nomenclature for filmwise condensation on a vertical plane surface [44]...21 Figure 4.1 Single N-type pellet...28 Figure 4.2 Single P-type pellet...28 Figure 4.3 Multiple P-type pellets connected in parallel...29 Figure 4.4 Single N-P-type pellet...30 Figure 4.5 Multiple N-P-type pellets...30 Figure 4.6 Typical Configuration of a Peltier device [56]...30 Figure 4.7 Current Flowing through N-P junction...31 Figure 4.8 Heat flow through a Peltier cell...32 Figure 5.1 Conceptual design. The numbers identify each of the aforementioned essential components of the system Figure 5.2 Multicomp MCPF E...38 Figure 5.3 Zalman CNPS7500-AlCu LED...40 Figure 5.4 Setup developed to characterize the Peltier cells...41 Figure 5.5 Potentiometer and Thermocouples placement in the resistance...41 Figure 5.6 Hewlett Packard 6274B DC Power Supply and Rear-Panel connections...43 Figure 5.7 Arduino Due...44 Figure 5.8 MAX31855 Pin Configuration...46 Figure 5.9 Final Setup...47 Figure 5.10 Dimensions of the final pool...48 Figure 5.11 Dimensions of the final pool...48 Figure Detail of the Coupling Fan/Peltier/Insulation/Aluminum wall...49 Figure Absolute temperature of the cold surface for 1A of current...59 Figure 6.2 Dynamic response for 1A of current...60 Figure 6.3 Temperature difference between both surfaces of the Pelier cell for 1A of current...61 Figure 6.4 Absolute temperature of the hot surface for 1A of current...61 Figure Dynamic response for 2A of current...62 Figure Temperature difference between both surfaces of the Pelier cell for 3A of current...62 Figure Dynamic response for 3A of current...62 Figure Dynamic response for 4A of current...63 Figure Dynamic response for 5A of current...63 Figure Dynamic response for 6A of current...64 Figure Dynamic response for 7A of current...64 Figure Dynamic response for 8A of current...64 Figure Dynamic response for 8.5A of current...64 Figure Absolute temperature of the hot surface for 8.5A of current...64 Figure for, for each of the currents tested...64 xi Figure Figure 6.17 Relation between Seebeck Coefficient and between both surfaces of the Peltier cell...66 Figure Relation between Thermal Conductance and between both surfaces of the Peltier cell...67 Figure 6.19 Kp=5.0; Ki=0.01; Kd=1.0 for...68 Figure Kp=0.2; Ki=0.01; Kd=1.0 for...68 Figure Kp=1.0; Ki=0.01; Kd=1.0 for...69 Figure 6.22 Comparison of the proportional terms...69 Figure Ki=0.05; Kp=1.0; Kd=1.0 for...70 Figure Ki=0.002; Kp=1.0; Kd=1.0 for...70 Figure Ki=0.01; Kp=1.0; Kd=1.0 for...70 Figure Comparison of the integrative terms...70 Figure Kd=5.0; Kp=1.0; Ki=0.01 for...71 Figure Kd=0.2; Kp=1.0; Ki=0.01 for...71 Figure Kd=1.0; Kp=1.0; Ki=0.01 for...71 Figure Comparison of the derivative terms...71 Figure 6.31 PID Controller test with...72 Figure 6.32 Final test of 72 hours under CPU full load...73 Figure CPU Temperature...73 Figure A.1. Comparison between the Gamma PDF and the PDF equivalent modified histogram obtained from the Gamma RNG ( =9, =100).... Erro! Marcador não definido. Figure B.1. BR versus User Scenario (G).... Erro! Marcador não definido. xii List of Tables List of Tables Table Convective Heat Transfer Coefficient for several processes...16 Table Multicomp MCPF E Characteristics...39 Table Remote programming coefficients...43 Table MAX31855 Specifications...45 Table MAX31855 Pin Description...46 Table Thermophysical properties of the liquids used in the present study, taken at saturation, at 1.013x10 5 Pa...50 Table A.1. Observation parameters and MSE of each RNG.... Erro! Marcador não definido. xiii List of Acronyms List of Acronyms CPU IC TEC emf MTBF C.O.P. DC AC PWM PID Central Processing Unit Integrated Circuit Thermoelectric Cooler Electromotive force Mean time between failures Coefficient of Performance Direct Current Alternate Current Pulse-Width Modulation Proportional Integrative Derivative xiv List of Symbols List of Symbols h V Convective Heat Transfer Coefficient Temperature difference Current xv List of List of Software Arduino IDE TEC Calculator Everest OriginPro 8.0 xvi Chapter 1 Introduction 1 Introduction This chapter gives a brief overview of the work, as well as motivation and objectives. 1 1.1 Overview and Motivation Since the birth of electronic technology, the heat flux produced by electronic devices has increased and this trend is expected to continue. In 1965, co-founder of Intel Gordon E. Moore stated that the number of components in an integrated circuit (IC) would approximately double every two years [1]. This trend, which was later known as Moore s law has been correct by half a century and is expected to continue until 2020 according to Kanellos [2]. Moore also posed the question Will it be possible to remove the heat generated by tens of thousands of components in a single silicon chip?. Since the end of the seventies, processors have improved their performance by 35% each year [3], while the size of integrated circuits has decreased [4]. However, this led to an increase in the power dissipated by electronic devices. This approach of reducing the size and increasing the frequency of operation, thus increasing the power dissipated has been possible until From this moment, the existing architectures were unable to dissipate to heat generated, so that new architectures have been adopted to continuously increase the performance while maintain the power dissipated. These architectures mainly consisted in distributing the workload across multiple less powerful processors. Computer architectures nowadays continue to find new ways to increase performance by distributing the workload and thus distributing the heat dissipation. However, the increase of heat generated by the processors is inevitable. This trend led Hennessy and Patterson [3] to state that thermal management of electronic devices would be, in the near future, the biggest limitation in the development of new processors. Additionally, a constant temperature below the junction temperature is required, which is approximately 85ºC for reliable operation and increased electronic device lifespan. Bar-Cohen et al. [5] alert for the performance degradation and reliability that this increase in heat dissipation can cause. Moreover, Tonapsi et al. [6] state that an increase by 10ºC in the temperature is enough to reduce the device life by half, while Paik et al. [7] state that an increase by 15ºC can increase time response (interconnect delay) from 10% to 15%. In line with this, the development of new techniques to cool down electronics becomes crucial. Cader et al. [8], state that the conventional air cooled systems have trouble accompanying the heat flux generated by constant increase of clock frequency and number of transistors. In recent years, various promising techniques for electronic cooling have been studied. In this work the feasibility of Direct Immersion Cooling in a dielectric liquid is tested, which addressed the liquid phase change. Two-phase cooling techniques, in which there is a phase change of the liquid, have much potential because they have a higher convective heat transfer coefficient than conventional computer cooling techniques (eg. air cooling and water cooling). As the name implies, the heated surface which in this case is the Central Processing Unit (CPU) is immersed in a fluid. For this to be possible the fluid must have dielectric properties. To take full advantage of this technique, the fluid used must have a lower boiling point than the heated surface temperature. At a given temperature difference between the boiling point of the fluid and the temperature of the surface, the liquid starts to boil. As will be explained later, when boiling occurs there is a phase change in which the convective heat transfer coefficient rises exponentially allowing higher heat fluxes to be removed. 2 However, in these techniques the boiling liquid vaporizes very fast and will be lost if it is not properly condensed. Given the particular properties that these liquids must have, the feasibility of such cooling systems strongly depends on an efficient condensation system. This way, losses of the fluid are avoided and the system can continuously work for long periods of time. The vapour will only condense in a stable condition for a well-defined temperature difference between the boiling point of the fluid and the temperature of the surface responsible for the condensation of the fluid. However, the CPU works in transient conditions for which the heat flux dissipated may change significantly along time. It is possible that for some periods of time the CPU may work under full load and generate its Thermal Design Power (TDP). This is usually given by the manufacturer and corresponds to the maximum heat power released by the CPU. When the heat flux generated by the CPU changes, i.e. the pace at which the fluid evaporates, the temperature of the surface responsible for condensation must also be adjusted, to assure that the temperature difference is high enough to promote the vapour condensation. Hence, the good performance of the condensation system depends on how well it can adapt the temperature of the surfaces responsible for condensation to the different heat fluxes created by the CPU during its working period. So, the key for a good performance of the condensation system lays in the accurate control of that temperature. In this context the main purpose of this work is to develop a controlled condensation system using Peltier cells, which can fix the temperature of the surface responsible by the condensation of the fluid, accordingly to the power supplied to them. This adjustment is made taking into account the variations in the heat dissipation associated to different CPU loads, which occur during its functioning. Despite their relatively low efficiency, Peltier cells were chosen to include the condensation system as they allow precise temperature control, independently of the ambient temperature which is, as aforementioned, one of the biggest limitations of conventional cooling techniques. The condensation system is integrated in a pool boiling cooling strategy. The conceptual design for the system is represented in Figure 1.1. The CPU, identified by number 1, is immersed in a pool, identified by number 2, filled with a dielectric fluid (the fluid used here is HFE-7000, as further explained). As the temperature difference between the CPU and the boiling point of HFE-7000 (34ºC) increases, boiling starts to occur and HFE-700 evaporates. The vapour is transported due to a gradient density and it comes in contact with the ceiling of the pool, which must be at a lower temperature than the HFE-7000 saturation temperature, so it can condensate. The temperature of the ceiling of the pool is fixed by the Peltier cell, identified by number 3. The Peltier cell allows to precisely fix a temperature accordingly to the current supplied to it. On top of the system is the heatsink, identified by number 4, and the fan, identified by number 5, which are responsible for dissipating the heat flux created by the CPU and the Joule effect generated by the Peltier cell. The Peltier cells may seem useless, given that at this stage of the work they could not be completely removed from the cooling components and are actually being used to cool the cells. As it will be explained later, in Chapters 3 and 5, the heatsink cools the CPU by conduction and the fan cools the heatsink by forced convection of the air. So, in this context we have still the physical 3 limitation imposed by the temperature of the ambient air. However, when this temperature is above the saturation point of the dielectric fluid, condensation will not occur. If the pool is not sealed, the fluid is lost. If the pool is sealed, the fluid is not condensed, so the pressure inside the pool may rise up to dangerous values. In addition, the boiling point of most liquids, including HFE7000 increases with pressure, so the temperature of the liquid will increase until values which do not assure a safe temperature for the CPU to work. In this context, the Peltier cells solve both these issues, while allowing a precise control of the temperature of the condensation surface, almost independently from the ambient temperature. Furthermore, if the control of the Peltier cells is robust enough, it may allow the system developed here to operate in any other applications requiring cooling and a precise control of the temperatures (which does not necessarily need to include condensation). Figure 1.1 Conceptual Design 4 1.2 Objectives Following the context presented above, the objective of this work is to design a controlled condensation system using Peltier cells. This work is integrated in a project whose objective is to study the feasibility of Direct Immersion Cooling and particularly pool boiling as a commercial cooling system. As aforementioned the system to be developed must fix the temperature of the surface responsible by the condensation of the fluid, accordingly to the power supplied to the Peltier cells. This adjustment is made taking into account the variations in the heat dissipation associated to different CPU loads. The design of an efficient controller requires a good understanding of the working principles of the Peltier cells. So, a secondary objective is to characterize the Peltier cells under real working conditions, namely their internal properties and the dynamic response. This is expected to be useful to provide alternative methods and guiding values to characterize the internal properties of the cells in works focused on the application and not on the fundamental research, as these methodologies are still sparsely reported in the literature. Such characterization of the system is important for a closedloop control and is vital to an open-loop control, so it is an important task to accomplish. Although one cannot aim at the development of a fully operational cooling system, as this is a complex task, which is out of the scope of a single Master Thesis, a preliminary configuration is proposed and tested in a real application tes
Advertisements
Related Documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x