The numerical simulation and experimental study on rack cooling effect in date center with UFAD system

Transcription

1 International Journal of Low-Carbon Technologies Advance Access published December 28, 2013 The numerical simulation and experimental study on rack cooling effect in date center with UFAD system... Yunzhi Ling 1 *, Xiaosong Zhang 1, Shuhong Li 1, Guannan Han 1 and Xiaogeng Sun 2 1 School of Energy and Environment, Southeast University, Nanjing, China; 2 Wujin Zhongtian Computer-room Equipment Co., LT, Changzhou, China... Abstract In order to analyze the way of improving the cooling effect of the data center with the UFAD system, the HVAC simulation software Airpak was applied. Numerical simulations for the interior temperature of the rack at different porosity ratios were carried out, and the simulated results indicate that the way of partly enhancing the porosity ratio of the front door and back door can improve the cooling effect. To solve the issue of the thermal current recycle in the exhaust outlet and airflow distribution disorder, the measure of the cold aisle regional closure was taken, and the comparison of data before and after the closure of the cold aisle proved its positive influence on the cooling effect: to lowering the inlet air temperature of racks, to improve airflow distribution inside and outside of the rack and to solve the thermal current recycle in the exhaust outlet. *Corresponding author: 5123bnmbnm@163.com Keywords: date center; UFAD; cold aisle; Airpak Received 26 November 2012; revised 6 January 2013; accepted 17 February INTRODUCTION With the rapid development of information industry, data equipment covers an increasing proportion in the data center. The majority of data equipments, which covers a small area and has a high level of integration, adopt alternating current supply. The average power of existing data processing equipment in each rack is 3 kw, and with data processing equipment being integrated in a narrow and limited space, the intensity of the energy consumption load and the density of the heating load inside the communication room increase sharply. There are higher and higher demands for air-conditioning refrigeration in the communication room in order to guarantee the normal operation of data processing equipment. Concurrently, the consumption of the airconditioning system in the communication room is a major component of energy consumption of communication constructions [1, 2]. Therefore, realizing its energy-conserving operation has active and pro-found significance to the continuous, rapid and sound development of information industry [3]. The target of date center s design and services is mainly heating devices, and then working staffs. Owing to various factors such as the distribution of heating equipment inside the room, overheating question will easily occur in operation. At present, the operating temperature of some communication rooms has been approaching the regulated upper limit. Excessively high operating temperature will directly influence the stable operation of data processing equipment, and at the same time, this potential question will also have adverse influence on the commencement of many projects as scheduled, and lead to certain economic loss. As a whole, questions of date center can be divided into several aspects as follows: first, regional air-supply temperature cannot satisfy the requirement of data processing equipment (inside the rack) regarding its operating environment; cold supplying airflow can only cool down a part of electronic devices, and other electronic devices on the rack absorb surrounding hot airflow which has relatively higher temperature; secondly, insufficient refrigerating output inside the date center, and data processing equipment which mainly adopts alternating current supply generates higher heat in the process of integration and operation: heat productivity of the equipment based on alternating current ¼ heat productivity of input devices ¼ equipment power 4 0.7; thirdly, short-circuiting often happens to air-conditioning airflow in the communication room, which leads to uneven distribution of refrigerating output in each region; fourthly, the efficiency of heat International Journal of Low-Carbon Technologies 2013, 00, 1 6 # The Author Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( by-nc/3.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com doi: /ijlct/ctt011 1 of 6

2 Y. Ling et al. Table 1. Design parameter of the experimental date center. Category The height of the date center The height of the static pressure plenum The height of the rack The supply air temperature Value 3.45 m 0.45 m 2.2 m 178C Category The power of each rack The power of each heater The air volume of the fan The size of the vent Value 3 kw 0.75 kw 800 m 3 /h m Category The amount of the rack The amount of the heater in each rack The amount of the vent The number of the layer in the rack Value dissipation inside and outside the rack is not high, which is also known as airflow distribution question; fifthly, recycle of hot airflow on the air exhaust sides of the rack: hot airflow combines and cycles with cold air surrounding the rack and enters the heating equipment, thus causing excessive temperature at the entrance of the rack, and severely influencing the operation stability of data processing equipment. In accordance with the above questions which often happen in the data room, the author of this dissertation adopts several measures when experimenting the data room design as follows: (1) disperse load, adopt multiple racks to bear heat source, classify the heat source to layers according to certain rules and monitor the heat source in each layer for reference; (2) larger air flow should be chosen in practical project; (3) install ventilators to offer certain auxiliary heat dissipation effect; (4) install baffles inside each rack; (5) make the front door of the rack (air inlet) as the cold aisle through the concept of the cold and hot aisle to form a region which brings in cold airflow; set the direction of the back door of each rack as the hot aisle to form a region which discharges hot airflow. At the same time, the removable insulation plate could be set so as to satisfy the requirement of regional closing the cold aisle. Internationally, the simulation and experimental research of thermal environment, airflow distribution and energy consumption in a room are much mature [4 6]. As to the date center, heat density, energy efficiency as well as the inlet temperature of data processing equipment have been considered by several scholars [7 9]; however, compared with research on an office environment, there are still many deficiencies; therefore, the research is well worth considered [10 13]. Therefore, in order to solve the issue of the hot air flow recycle on the air exhaust sides of the rack, this article will change the porosity ratio of the front and back door of the rack and compare the temperature and flow field inside each rack before and after regional closing the cold aisle, to conduct an analysis on its influence on the heat dissipation effect of the equipment in the rack. Figure 1. Layout of the experimental date center. Figure 2. Photo of the experimental date center. 2 EXPERIMENTAL EQUIPMENT AND METHOD 2.1 Experimental date center interior design parameter For the data processing equipment, the requirements of the inlet temperature of the low side and outlet temperature of the high side should be 158C+ and 358C+, respectively. The Figure 3. Layout of the experimental date center. average power of the single rack, the existing data processing equipment, is 3 kw [11]. The design parameter of the experimental equipment room should comply with data in Table 1. 2 of 6 International Journal of Low-Carbon Technologies 2013, 00, 1 6

3 Study on rack cooling effect in date center 2.2 Layout of experimental date center The layout of the plenum and room is shown in Figures 1 3. Each rack is divided into six layers. In addition to the set of four experimental heat sources, an alternative heat source was set. The thermocouple, used for recording experiment data via data collecting instrument, was set in an appropriate place in each layer and the front door as well as back door of each rack. 2.3 Experimental method The research combines experimental methods and simulation methods; the former method is to record temperature data at each point via data collecting instrument, including front and back door temperature, temperature at each point inside the rack. The porosity ratio of the front and back door should be 25% in the experiment. To compare experimental data under such conditions with simulation data to verify whether the simulation method was correct, and then to predict the influence on the cooling effect from the porosity ratio of the front and back door through the simulation with different porosity ratios (respectively, 25, 32, 40, 45, 50%). Finally, to compare experimental results of before and after the regional closure of the cold aisle with simulation results, to analyze the influence on the temperature field, cooling effect and air distribution after the setting of the cold aisle. CFD simulations were performed in Airpak version 3.0. Fluent Airpak is a kind of professional systematic analysis software of artificial environment, aiming at engineers, architects, interior designers and other engineers in specialized fields, especially in the field of HVAC. Airpak is popular business-oriented CFD software in the world, and certain results have been achieved by many scholars with it [12 15]. 3 EXPERIMENTAL DATE CENTER NUMERICAL SIMULATION AND ANALYSIS 3.1 Mathematical model Airpak provides the interior zero-equation model, RNG k 1 turbulence model, k 1 two-equation model and other turbulence models [16, 17]. Table 2. Meshing control. Coarse Normal Num elements Num nodes Num elements Num nodes In 1925, Prandtl proposed the mixing length theory, and based on which, Chen proposed a simple algebraic equation [18] to represent turbulent viscosity, m t ¼ rnl, n stands for the local time-averaged velocity. Turbulent viscosity was taken as a function of the local time-averaged velocity and length scale and length scale defined as the distance with the nearest wall. The interior zero-equation model was developed to simulate interior air flow, to provide a simple and stable turbulence model for the study on HVAC airflow distribution. It calculates with fast and stable convergence, being applicable to predict interior airflow distribution; ideal effect could be reached under mixed convection conditions. It can also be applied under natural convection, forced convection and displacement ventilation, which can help spare computational resources [14]. For the objects of study in this article, the interior zero-equation is adopted [19]. The meshing control is shown in Table 2, and relevant parameters of the discretization solution are shown in Table Physical model As is shown in Figure 4, the experiment room model is equipped with 10 racks, a rectification metallic mesh, 10 square outlets with a porosity ratio of 40%, two return air inlets and a data collecting instrument. The porosity ratio of the front and back door of the rack is 25%. 3.3 Analysis of simulated results It can be seen from Figure 5, which is the comparison of the experimental and simulated result of the interior temperature under the condition that the porosity ratio of both the front and back doors of the rack being 25%, the overall tendency of temperature is the same, and the deviation belongs to the acceptable scope, meaning that the model and solution has a high goodness of fit with practical situation, and can reflect the latter. As Figure 6 shows, by using the above model and changing the porosity ratio of the front and back door of racks to explore its influence on the cooling effect inside the rack, we can find out that partly increasing the porosity ratio of the front and back door of the rack can improve the cooling effect inside the rack, because the porosity ratio of the front and back door of the rack can improve airflow distribution inside the rack, and with the porosity ratio gets higher, heat generated by the equipment inside the rack can be discharged more effectively, which improves the heat dissipation efficiency of the rack. However, when the porosity ratio reach to a certain degree, as the velocity vector diagram in Figure 7a and c Table 3. Relevant parameters of the discretization solution. Turbulence model Convergence criteria Variables Pressure Temperature Momentum Interior zero equation Flow: Discretization scheme Second Second Second Under-relaxation Power: Solving format AMG AMG AMG Type of the linear solver V Flex Flex International Journal of Low-Carbon Technologies 2013, 00, of 6

4 Y. Ling et al. shows, exhaust air volume from the back door of the rack is markedly increased, and the cold airflow pass the U-shape region inside the rack from up to down are markedly Figure 4. The model of the test date center. decreased. In addition, the inlet temperature is increased as a result of the hot air recirculation, and the upper temperature of the rack is increased apparently too. The simulated result indicates that the advisable porosity ratio is 40%+, which is shown in Figure 8. It can be seen from Figure 7b that under the condition of non-sealed region in the cold aisle, due to reasons as low air output (speed), only a part of cold airflow enters from air supply outlet scours the lower region of the rack. Therefore, we can see from Figure 6 that the cooling effect in the lower region of the rack is far better than the upper region. A part of cold airflow is discharged from the back door of the rack after chilling the hot source, and another part continues to scour the hot source inside the U-shape region of the rack from up to down, hot airflow generated at the top after chilling all the hot source is discharged out of the rack under the function of ventilator and hot airflow discharged from the back door of the rack flows upwards and enters the upper mixing region. A part of hot airflow is discharged by the return air inlet, and another part of hot airflow joins lower cold air and enters the rack from the front door, thus leading to the hot airflow recycle and temperature rise of the inlet of the rack, which Figure 5. Experimental and simulated date of the interior temperature of rack 1. Figure 6. At a porosity ratio of 25%, the simulated result before and after the closure of the cold aisle. The temperature field (x ¼ 5 m) (a) before and (b) after the closure of the cold aisle. 4 of 6 International Journal of Low-Carbon Technologies 2013, 00, 1 6

5 Study on rack cooling effect in date center Figure 8. At different porosity ratios, the simulated result of the interior temperature of rack 1. It can be seen from Figure 7b that after the closure of the cold aisle, cold airflow enters from the air supply outlet is filled in the cold aisle, which scours the front door of the rack quite roundly. The hot airflow is isolated outside the aisle and discharged by the air return outlet, thus basically solving the hot airflow recycle on the air exhaust sides of the rack. Then it can be seen from Figure 9, the air inlet temperature is relatively lower, and the air outlet temperature is slightly higher than that before regional sealed. The temperature difference indicates that the cooling effect of the rack has been substantially improved. 4 CONCLUSIONS Figure 7. The velocity vector (x ¼ 5 m) before and after the closure of the cold aisle. At a porosity ratio of 25%, the velocity vector (x ¼ 5 m) (a) before and (b) after the closure of the cold aisle. (c) At a porosity ratio of 50%, the velocity vector (x ¼ 5 m) before the closure of the cold aisle. even exceeds the temperature limit and leads to the equipment s failure in operating stably, and the hot spaces are mainly centralized on the upper rack. Therefore, we can see from Figure 6 that the temperature of the upper rack is higher. For the indoor thermal environment, the lower part of the room mainly presents the single-direction flow, which has obvious thermal stratification, the upper part is mixing region, where temperature field is relatively even and the average temperature is close to the temperature of air return. Through the analysis of the above numerical simulations and experimental results, the following conclusions were reached: (1) partly enhancing the porosity ratio of the front and back door could improve interior and external airflow distribution, raise cooling efficiency and improve the cooling effect of the rack. However, it will carry out adverse effect when the porosity ratio increases to certain degree. This article recommends that the advisable porosity ratio should be 40%+. (2) Hot air recirculation problems of the exhaust side are common in the equipment room with the underfloor air-conditioning system. Under this circumstance, regional closure of the cold aisle is applicable, the adoption of which has a positive effect on reducing inlet air temperature, ameliorating air flow distribution, ensuring stable operation of devices and improving the cooling effect of the equipment in the rack. (3) Since temperature inside the rack is not well distributed, the cooling effect of lower part is better than that of the upper part, we propose that heating devices should be appropriately arranged in practice, devices with higher requirements of cold should be put in International Journal of Low-Carbon Technologies 2013, 00, of 6

6 Y. Ling et al. Figure 9. The experimental result of the rack temperature before and after the closure of the cold aisle. Inlet and outlet temperature of the rack (a) before and (b) after the closure of the cold aisle. the under layer, while devise with lower requirements of cold should be put in the upper layer. REFERENCES [1] Alajmi A, El-Amer W. Saving energy by using underfloor-airdistribution(ufad)system in commercial buildings. Energy Convers Manage 2010;51: [2] Fong KF, Hanby VI, Chow TT. HVAC system optimization for energy for energy management by evolutionary programming. Energy Build 2006;38: [3] Yao J. Effect of variable ventilation modes on interior thermal comfort and building energy consumption. Int J Low-Carbon Tech 2012; 7: [4] Lee KH, Schiavon S, Bauman F, et al. Thermal decay in underfloor air distribution (UFAD) systems: fundamentals and influence on system performance. Appl Energy 2012;91: [5] Xu H, Niu J. Numerical procedure for predicting annual energy consumption of the under-floor air distribution system. Energy Build 2006;38: [6] Chung JD, Hong H, Yoo H. Analysis on the impact of mean radiant temperature for the thermal comfort of underfloor air distribution systems. Energy Build 2010;42: [7] Patterson MK. The effect of data center temperature on energy efficiency, thermal and thermomechanical phenomena in electronic systems. In: Thermal and Thermomechanical Phenomena in Electronic Systems, ITHERM th Intersociety Conference, 2008, [8] Herrlin MK, Belady C. Gravity-assisted air mixing in data centers and how it affects the rack cooling effectiveness. In: Thermal and Thermomechanical Phenomena in Electronics Systems, ITHERM 06. The Tenth Intersociety Conference, 2006, 438. [9] Schmidt R, Cruz E. Raised floor computer data center: effect on rack inlet temperatures of chilled air exiting both the hot and cold aisles. In: 2002 Inter Society Conference on Thermal Phenomena. [10] Bauman FS, Webster T. Outlook for underfloor air distribution. ASHRAE J 2001;43: [11] Webster T, Bauman FS. Design guidelines for stratification in UFAD systems. HPAC Eng 2006;78:6 16. [12] ASHRAE Handbook-Fundamentals. America Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc [13] Chen X, Zhu Y, Wang Y. Airflow simulation in air-conditioned and ventilated rooms with zero-equation model. HVAC 2006;36: [14] Gao CF, Lee WL. Optimized design of floor-based air-conditioners for residential use. Build Environ 2009;44: [15] Vera S, Fazio P, Rao J. Interzonal air and moisture transport through large horizontal openings in a full-scale two-story test-hut: Part 2-CFD study. Build Environ 2010;45: [16] Airpak 3.0 user s guide. [17] Kong Q, Yu B. Numerical study on temperature stratification in a room with underfloor air distribution system. Energy Build 2008;40: [18] Wan M, Chao C. Numerical and experimental study of velocity and temperature characteristics in a ventilated enclosure with underfloor ventilation systems. Interior Air 2008;15: [19] Chen Q, Xu WA. Zero-equation turbulence model for interior airflow simulation. Energy Build 1998;28: of 6 International Journal of Low-Carbon Technologies 2013, 00, 1 6