Data center equipment was modeled as a continuous 24/7 load of 65 W per person (this does not include energy consumption by cooling equipment), based on a survey of typical data center configurations. The owner selected temperature and humidity criteria consistent with the 2008 ASHRAE Environmental Guidelines for Datacom Equipment (ASHRAE, 2008). Complete hot and cold aisle separation was assumed. The cooling system was simulated with and without airside economizers, with appropriate adjustments made to the fan system pressure drops. The TMY3 climate data were analyzed to understand the psychrometric processes that could be used to achieve the required supply air temperatures. We found that an economizer and direct evaporative cooling could be used to cool the data center for 99.5% of hours with no mechanical cooling, partly because of its assumed broad acceptable supply dry bulb temperature range (66.4-77ºF), as recommended by the ASHRAE Datacom guidelines. At this stage, the economizer and direct evaporative cooling strategies were predicted to save substantial energy, but were not included in the actual design because the incremental capital cost was estimated to be too high. Eventually, these strategies were included in the design, with the ability to cool mechanically for the limited number of hours when analysis of TMY2 data suggests dehumidification will be required. Energy savings by the servers were not included; this is a conservative assumption, as the NREL Information Technologies department will probably use equipment that is more efficient than the industry standard.

The heat from the data center available for recovery was modeled for several scenarios. First, an "economizer" case was modeled assuming direct discharge of the hot aisle air into the crawlspace, where it would warm the mass during non-occupied hours and be used during occupied hours to preheat ventilation air. A "no-economizer" case was also modeled employing a runaround loop to transfer heat from the data center hot aisle to preheat domestic hot water and ventilation air. More efficient data center equipment would reduce the amount of heat available for ventilation preheating; however, future efficiency gains were assumed to be matched by increased data center use so the overall load would remain the same.