Data Center Journal

Volume 28 | August 2013

Issue link: https://cp.revolio.com/i/148754

Contents of this Issue

Navigation

Page 28 of 36

Table 2 - Data Center Airflow Metrics Comparisons Table 3 - Basic Cost Analysis of Data Center Scenarios Scenario RCIHI RCILOW RTI 55 °F SA – no containment 100% < 0% 98.8% 55 °F SA – containment 100% < 0% 98.8% New UPS Installation No Yes 65 °F SA – no containment 99.7% 100% 98.8% 65 °F SA – containment 100% 100% 98.8% Capital Cost $888,460 $2,070,316 $916,585 $2,098,441 75.04% 100% 98.8% Annual Savings $154,919 $267,900 $167,687 $280,108 100% 100% 98.8% Simple Payback 5.7 yrs. 7.7 yrs. 5.5 yrs. 7.5 yrs. 75 °F SA – no containment 75 °F SA – containment COST ANALYSIS A detailed cost analysis is an important part of the decision making process. When choosing between a small number of final designs a life cycle cost analysis is appropriate, whereas a simple cost analysis is adequate in the early phases. The main difference between the two is the inclusion of interest rates, operational costs and equipment depreciation. For this case a simple cost analysis was done in order to make high level comparisons. It is also important to note that IT is typically 40-50% on day 1, although it is assumed 100% and uniform in this example. Item 65 °F Supply Air 75 °F Supply Air No Yes analysts may calibrate models by reviewing UPS load information and extrapolating it to server load variations / diversity. Others may measure airflows from cooling units and airflow inlets and adjust boundary conditions within the model as needed. Thermal imaging is sometimes utilized as well for model calibration and can provide further insight into the data center dynamics (see Figures 13 and 14). Figure 13 - CFD Model Calibration, Simulated Cold Aisle Several assumptions were made in order to carry out the basic cost analysis; 1) a 45°F chilled water supply temperature (55 °F SA) at standard conditions, 2) hot-aisle containment is required at 65°F CHWS (see Table 2), 3) total cost for heat exchanger and pump is $72,500 (provided by design engineer, 4) total cost for containment is $28,125 (based on $25/SF installed cost of curtains and $10/lb. of ductwork), 5) Total cost for two new chillers is $480,000, 6) Total cost for nine (9) new CRAC units and associated installation is $362,585, 7) Total cost for new UPS is $1,181,856, and 8) $.102/kWh cost of electricity provided by utility company. Remember that we are assuming a 65 °F supply air temperature corresponds to a 55 °F chilled water supply temperature. The baseline here is considered to be 55 °F supply air and 45 °F chilled water supply. Notice that the simple payback is slightly less when the supply air temperature is raised to 75 °F (65 °F CHWS), despite the increased cost for containment infrastructure. It should also be noted that the total project payback can be reduced significantly through addition of utility incentives. Many utility companies will provide incentive money on projects with payback similar to those shown in Table 3. The analysis outlined within this article is sufficient documentation when applying for these incentives. 13F D Figure 14 - CFD Model Calibration, Actual Cold Aisle Thermal Image ADDITIONAL CONSIDERATIONS Model calibration plays a crucial role in obtaining accurate results, regardless of the software and type of modeling being performed. In the energy modeling world, models of existing buildings are typically "calibrated" to actual metered electric and steam / natural gas / fuel oil data. A minimum of one year's worth of metered data is desired for improved accuracy. Energy models of new buildings are not typically calibrated since there is no historical usage data available to calibrate against. Model calibration in the CFD world can be similar to methods utilized with energy modeling, but additional tools are available. Data center 26 | THE DATA CENTER JOURNAL When calibrating CFD models it is very important to understand what exactly the model is being calibrated to. Some data center loads can vary significantly from month to month, week to week, or even day to day. If the load is known to vary, then multiple measurements should be taken in order to fully understand the variations. Remember, CFD www.datacenterjournal.com

Articles in this issue

Links on this page

Archives of this issue

view archives of Data Center Journal - Volume 28 | August 2013