Issue link: https://cp.revolio.com/i/148754
The next step is to gather all the information necessary to build the actual model of the data center. This information must be provided to the modeler by the data center manager or design engineer. Site surveys will be required to obtain any data that is not readily available. This includes, but it not limited to; • CAD plan or written description of space • Room dimensions, spatial details and plenum heights (in lieu of survey) • Airflow obstructions within space (survey / photos) • Equipment cut-sheets or operating data / performance curves (verify through survey) • Sequence of operations for white space and central plants • IT equipment inventory (verify through survey) • IT load trending or average kW/rack (in lieu of physical measurements) Depending on which package is selected. This data will be inputted to the model in different ways. More robust software packages may allow integration with 3D CAD tools where the geometry is launched directly from CAD. Other programs enable the user to quickly and easily construct the geometry directly within the CFD interface. After all geometry has been built and boundary conditions set, the simulation can be run and results reviewed. It is important to keep in mind that the primary focus of the simulation output is rack inlet temperatures. These temperatures must be maintained within a certain range to ensure safe operation of the computing equipment. The industry standard guideline for these conditions is provided by the ASHRAE TC 9.9 in Table 1 below: Table 1 - ASHRAE TC 9.9 - Rack Inlet Temperatures Category Minimum T (F) Maximum T (F) Minimum RH Maximum RH Recommended 64 80.6 40% 60% A1 Allowable 59 89.6 20% 80% A2 Allowable 50 95 20% 80% The recommended guidelines also include a 41.9 °F – 59 °F dewpoint range and an allowable limit of 80.6 °F maximum. Dewpoint limits should be included in the energy model calculations when an air-side economizer is also being considered. CASE STUDY – FREE COOLING BASELINE ENERGY MODEL curve should be based on more than four points, but only four points were available here. A curve was fit to the data in order to generate a formula for kW/ton variation with load. Figure 2 below illustrates the curve fit and corresponding equation. Figure 2 – New Chiller Efficiency Curve Here "y" represents the kW/ton efficiency of the chiller and "x" represents the total load on the chiller for the hour. This equation is then entered directly into the energy model as shown in Figure 1, column Q. The energy reduction related to the water-side economizer is included in the energy model as an additional section, as shown in Figure 3 below. The economizer is assumed to be at 100% capacity when the outdoor air wet bulb temperature is a minimum of 4 °F below the chilled water supply temperature, based on the engineered control strategy. The new heat exchanger pump speed will modulate to maintain primary chilled water supply temperature setpoint. A bypass valve will be utilized to ensure the chiller inlet water temperature remains above the manufacturer supplied minimum. The chiller energy consumption is modeled in Column AA of Figure 3. The chiller energy consumption is modeled at 0 kW during 100% economizer mode and varied linearly when the outdoor air wetbulb temperature is between the chilled water supply temperature and the supply air temperature to the data center. It is assumed that the pumps / fans will remain ON in all modes and an additional pump is required during 'free-cooling" and "pre-cooling" modes. Figure 3 - "Snapshot" of the Energy Model for "Free Cooling" Energy Consumption Reductions With this Case Study the cooling demands are met through chilled water produced at a central chilled water plant. The plant consists of; one (1) 400-ton chiller, four (4) 25 HP cooling tower fans, one (1) 50 HP condenser water pump, one (1) 20 HP chilled water pump and one (1) new 400-ton water-side economizer. All pumps and fans related to the central plant are constant speed and are modeled as such. Pump and fan power should be varied within the energy model based on fan and pumps laws if they are variable speed controlled. Part load performance data was obtained from the manufacturer at 25%, 50%, 75% and 100% chiller loading. Ideally a performance www.datacenterjournal.com THE DATA CENTER JOURNAL | 21