The choice of cooling architecture,
including hot and cold air containment, is of paramount importance for
minimising the OPEX of a data centre. However, an effective control system is
also essential when hoping to achieve the maximum energy efficiency and PUE.
White Paper 225 details the different levels of cooling control as a basis for
achieving the optimal balance of performance and efficiency.
Achieving efficient use of electrical power is a major concern for data centre operators for both cost and environmental reasons. Next to the IT itself, the largest consumer of power in a typical data centre is the cooling system. Optimising its efficiency, therefore, is greatly to be desired.
Three high-level steps to achieving this goal are: selecting an appropriate cooling architecture; adopting an effective cooling-control system and managing airflow in the IT space. The first of these entails selection of heat rejection method, economiser mode and indoor air-distribution method; the third concerns the separation of air streams into hot and cold airflows using aisle and rack containment. The final step is the continuous monitoring and control of all cooling equipment deployed inside the data centre.
A new White Paper from Schneider Electric, the global specialist in energy management and automation, #225, “Optimize Data Center Cooling with Effective Control Systems” discusses the variables that influence cooling performance and describes the complexity of cooling-system dynamics that frequently increase the need for counter-intuitive measures required to achieve the trade off between cooling performance and efficiency.
Issues affecting the cooling process include the fact that cooling systems capacity is always oversized due to availability requirements and data centres operating under less than total IT capacity; the IT load in terms of equipment population and layout frequently changes over time; cooling system efficiency varies with factors other than IT load such as outdoor air temperature, cooling settings and control approaches; compatibility issues arise due to the installation of cooling equipment from different vendors.
Traditional control approaches involving manual adjustments to individual pieces of equipment such as chillers and air conditioners lead to uneven cooling performance as the effect of adjustments made to one unit can lead to hot spots elsewhere in the data centre. Frequently, too, there is no visibility to the performance of the entire cooling system, a flaw that is often compounded by the presence of poor-quality or badly calibrated sensors and meters.
The paper also recommends approaches for effective control systems entailing the use of automatic controls for shifting between different operation mode such as mechanical mode, partial economiser mode and full economiser mode. Indoor cooling devices should be co-ordinated to work together under a centralised control system with the flexibility to change certain settings based on immediate requirements.
The white paper proposes that control systems be categorised into a hierarchy of four levels namely device-level control, group-level control, system-level control and facility-level control for maximum efficiency.
Device-level control involves the control of individual units such as chillers. Group level control refers to the co-ordination of several units of the same type of device, typically from the same equipment vendor and controlled by the same control algorithm. System-level control co-ordinates the operation of different cooling subsystems within a data centre, for example a pump and a CRAH (computer room air handler). Finally, facility-level control integrates all functions of a building into a common network that controls everything in the building from heating, ventilation, air conditioners and lighting systems to the security, emergency power and fire-protection systems.
The paper demonstrates that effective control of cooling systems can maximise cooling capacity, simplify management, eliminate hot spots, increase performance, reduce costs and enhance data-centre availability.