Steady-state and transient comparison of cold and hot aisle containment and chimney

Abstract

Aisle containment is increasingly being used in data center cooling solutions because of the benefits of segregating the cold and hot air streams which reduces mixing and enhances energy savings. Containment can be integrated into legacy raised floor or non-raised floor local cooling systems such as in-row coolers. However, containment is not always entirely effective due to leakage. Although containment systems providing better sealing are becoming available, significant leakage exists in currently operational sites. The impact of leakage can be significant leading to backflow, the Venturi effect or entrainment. Comparisons between Cold Aisle Containment (CAC) and Hot Aisle Containment (HAC) indicate that there is a slight advantage of the latter in terms of ambient room temperature, failure scenarios and heat exchanger efficiency, although CAC systems can be easier to install. Detailed comparisons between both containment solutions are not widely available. Each containment approach has different behavior characteristics and air flow paths. For instance, leakage can introduce different flow and temperature conditions at server inlets depending on whether CAC or HAC is being used. An experimental characterization of containment was carried out at the ES2 research data center lab at Binghamton University [1] along with detailed numerical modeling. The numerical model was validated based on the experimentally measured flow rate of the perforated tiles in the cold aisles. Also, for the different types of IT equipment modeled in the simulations, experimentally measured flow impedance and fan curves were utilized. The details of the containment including leakage at the bottom and sides of the racks, the mounting rails, blanking panels and door seams were characterized experimentally and then incorporated in the computational model. Fan curves for the three CRAH blowers were used to accurately represent the targeted flow rates. This study focuses on three containment configurations: cold aisle containment and hot aisle containment and chimney (with a drop ceiling). In first two cases, the CRAH flow rate is controlled based on the pressure difference that was measured by three sensors inside and outside of the containment and for the third case, the CRAH flow rate was controlled based on IT required air flow. The water flow rate of the CRAH unit is also controlled based on the air return temperature and total heat load in the room so as to maintain the supply air temperature set point. The temperature and pressure fields in the aisles and at the server inlets and outlets are monitored for all the cases with leakage.