Modeling Rack and Server Heat Capacity in a Physics Based Dynamic CFD Model of Data Centers Sami Alkharabsheh, Bahgat Sammakia 1/28/213
2 ES2 Vision To create electronic systems that are self sensing and regulating, and are optimized for energy efficiency at any desired performance level Project Vision Toward a full physics-based experimentally verified 3D computational fluid dynamics model for data centers
3 Outline Introduction Physics Based Steady State Baseline Model CRAC model Server model Tile model Dynamic Model- Server Heat Capacity Effect Server level model Room level model Case studies Conclusions and Future Work
2.5<= 2.4-2.49 2.3-2.39 2.2-2.29 2.1-2.19 2.-2.9 1.9-1.99 1.8-1.89 1.7-1.79 1.6-1.69 1.5-1.59 1.4-1.49 1.3-1.39 1.2-1.29 1.1-1.19 1.9>= Response percentage 4 Introduction EPA (27):1.5 % of total U.S. electricity consumption in 26. (Total cost of $4.5 billion) 14% 12% 1% 8% 6% 4% 2% % Datacenter Dynamics (212) Global Census : power requirements grew by 63% globally to 38 GW from 24 GW in 211. PUE M. Stansberry and J. Kudritzki, Uptime Institute 212 Data Center Industry Survey, Uptime Institute, 213. Uptime Institute 212 Data Center Industry Survey: PUE>1.8 for more than 55% of data centers HVAC Cooling Others IT M. Iyengar and R. Schmidt, Energy Consumption of Information Technology Data Centers, 21.
5 Nature of Problem Cooling Power treehugger.com In real time, cooling is difficult to control due to long lag times Complexity of transport in data centers Overprovisioning is commonly used for safe operation Solutions for improving the energy efficiency in data centers have been isolated System-level and holistic solutions are a MUST Fromtimes.com Performance is not proportional to power Server overprovisioning is a common practice
6 Bench Mark Numerical Model CRAC Raised Floor Rack Perforated tile Parameter Value Room size 6.5 m x 13.42 m x 3.65 m Plenum depth.6 m Tile perforation ratio 5% Perforated tiles area.61 m x.61 m CRAC fan speed 1%
7 CRAC Model Based on manufacturer data Liebert 114D CW Operating fan curve is obtained from the manufacturer, Liebert Consulting The CRAC model is calibrated such that the flow rate can be predicted accurately at different operating pressures Static pressure (in. H2O) * Alkharabsheh et al. Utilizing Practical Fan Curves in CFD Modeling of Data Centers, SEMITHERM213. 3.5 3 2.5 2 1.5 1.5 Calibrated operating point Uncalibrated operating point Emersonnetworkpower.com.5 1 1.5 2 2.5 3 3.5 4 Flow rate (CFM) CRAC internal resistance x 1 4
8 Server Model A standard testing procedure following the AMCA 21-99 guidelines are used to measure the pressure fan curves 2U server for testing Flow bench apparatus 9 RU server simulators (load banks) and a 2 RU commercial server are tested 25 2 2 RU server 9 RU load bank The measured fan curves include the internal resistance of the server The measured fan curve can be imbedded directly into the CFD Static pressure (Pa) 15 1 5-5.5.1.15.2.25.3.35.4.45 Flow rate (m 3 /s)
9 Tile Model The CFD tile model is validated using experimental data in Schmidt et al.*.1 Computerfloorpros.com The CFD tile model is modified to compensate for the momentum loss in the CFD flow resistance model The CFD tile model is able to capture the tile flow distribution and can be used in room level simulations Airflow rate (m 3 /s) -.1 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15.1 -.1.1 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 Row A Row B Row C -.1.1 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 Row D -.1 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 Tile Solid line: experimental data, Dashed line: CFD results *Experimental data: Schmidt et al, Measurements and Predictions of The Flow Distribution Through Perforated Tiles in Raised-Floor Data Center, InterPACK21
1 Steady State Room Level Simulations In addition to affecting the power dissipation, the servers power scenario also the airflow pattern by operating 15 kw/ rack The room can be underprovisioned/ overprovisioned based on the servers power level Several parametric studies can be conducted using this model * Alkharabsheh et al. Numerical Steady State and Dynamic Study in a Data Center Using Calibrated Fan Curves for CRACs and Servers, InterPACK213 2 kw/ rack 32 kw/ rack 15 17.5 2.5 24 28.1 32.9 38.5 45 Temperature (C)
11 Simple Dynamic Model The thermal capacity of the equipment is not taken into account Complete CRAC failure simulated at 2 seconds Inlet temperature (degc) 11 1 9 8 7 6 5 4 No backup power Blower backup power Critical temperature Failure 3 Supporting the CRAC blower with backup power provides the room with extra cooling and time that can be utilized in increasing the reliability of operation 2 1 2 3 4 5 6 7 8 9 Time (s) Unused plenum cold air * Alkharabsheh et al. Numerical Steady State and Dynamic Study in a Data Center Using Calibrated Fan Curves for CRACs and Servers, InterPACK213
12 Server Heat Capacity The server level CFD model is developed based on the lumped mass approximation T server 6 5.5 5 4.5 4 3.5 Exp. data [*] CFD model Experimental data is used to calibrate and validate the server level CFD model An increase in the rate of change in temperature is observed at low values of heat capacity until instantaneous change in temperature is noticed when server heat capacity is completely neglected *Ibrahim et al., "Thermal Mass Characterization of a Server at Different Fan Speeds," ITHERM212. T server T server 3 1 2 3 4 5 6 7 8 9 5 4.75 4.5 4.25 4 3.75 3.5 3.25 Time (s) 3 1 2 3 4 5 6 7 8 5 4.5 4 3.5 Time (s) Exp. data [*] CFD model 3 1 2 3 4 Time (s) No HC 1% Cap. 1% Cap. 5% Cap. 1% Cap. 12% Cap. 15% Cap.
13 Room Level Model The detailed rack model is capable of hosting the server model, blanking panels, leakage through the mounting rails, and internal supports n=2 n=2 n=1 Blanking panel Mounting rails Server Each server consists of an experimentally characterized fan curve and thermally calibrated heated mass Each rack is populated with twenty of the 2 RU servers Raised Floor Detailed rack model Rack Perforated tile CRAC
14 Case I: Servers Shutdown It is assumed that all the servers inside the modular data center are shutdown at time 2 seconds Three different room level models are compared in this transient analysis Including the servers heat capacity is crucial in dynamic modeling. However, the heat capacity of the rack chassis can be neglected without affecting the accuracy of the results and reducing the computational time Power (kw/rack) Rack inlet tempeature 2 2 1.8.6.4.2 Time (s) No HC Servers HC Only Servers & Racks HC 5 1 15 2 25 Time (s) Where: T Tss Tˆ T T o ss
15 Case II: Server Power Short Pulses Fluctuations in the dissipated power is simulated in the form of 3 second pulses Power (kw) 15 The temperature increases immediately in the model if we ignore the heat capacity The heat capacity damps down the effect of short duration power fluctuations on the inlet temperatures Rack A1 inlet temperature 1 3 1218 1 Time (s) 1.8.6.4.2 Temperature without HC Temperature with HC 2 4 6 8 1 Time (s)
16 Conclusions and Future Work Experimentally validated models of different data center components are developed A steady state and dynamic, physics based, room level CFD model for a bench mark data center is developed It is found that the heat capacity of the servers affects the rate of change in temperature significantly The effect of rack frames heat capacity is found to be small and can be neglected in room level simulations Future work will include adding cooling unit heat capacity
17 Acknowledgement This material is based upon work supported by the National Science Foundation under Grants No.1134867 and CNS-14666