Sebastian Beyer, test centre manager at Stulz, explains why knowing the actual capabilities of equipment under temperature and air humidity conditions is vital, and how individual performance tests can help to achieve an energy efficient configuration for data centres
With utility bills rising, growing pressure to reduce carbon emissions and an increased demand on power networks, data centre owners and operators are faced with a major challenge when it comes to energy consumption.
Although organisations such as Google and Microsoft are leading the way in measuring and improving energy use –mainly because they consume such vast amounts and it makes economic sense – the fact is that all enterprises need to do the same, at least until such a time as we develop 100% renewable energy sources.
It is widely accepted that data centres consume almost as much energy for non-computing resources – such as cooling and power conversion – as they do actually powering their servers. Cooling and airflow management is a continually evolving science due to the amount of equipment variations possible and the number of options in terms of data centre design and operation.
Air conditioning solutions must be planned and implemented meticulously, and for large-scale projects, in particular, it is not just the investment cost of cooling solutions that have to be considered but the operational expenditure associated with them. In addition, planners and operators face the question of how to achieve energy savings by dimensioning their systems appropriately.
Legislation is increasingly putting pressure on data centre owners and operators and this will only increase. Germany, in particular, is leading the way in this area and its Energy Saving Ordinance (EnEV) legally obliges operators of air conditioning solutions to subject all systems over 12kW to an energy inspection on initial installation, on the replacement of important components, or every 10 years.
This type of legislation is expected to become more widespread across the globe in years to come, as countries try to reduce energy usage and lower CO2 emissions. Therefore, the challenge is to implement high performance, efficient and future-proof dimensioning for these systems in data centres.
Although manufacturers of air conditioning units ascertain the technical specifications of their equipment in accordance with DIN EN 14511 by testing to determine total cooling output and energy efficiency, in practice there can be considerable differences in cooling capacity. This is due, for the most part, to different environmental influences, which cannot be taken into consideration in the standard performance test.
Data centres around the world have varying requirements for their air conditioning units. Therefore, a system’s performance is determined not just by the quality of the individual components but also by its location. Ambient conditions, such as the temperature and humidity of the return air at the unit intake, or of the supply air at the unit outlet, have a significant influence on the performance of the system as a whole. In practice, changes to the equipment’s temperature and air humidity parameters may have a negative influence on actual cooling capacity and efficiency. This affects the operating points of vital components such as pumps, fans and compressors.
If specialist planners and operators rely on the theoretical data provided by manufacturers, they run an increased risk of reduced capacity for cooling their data centres during future operation. The result can be incalculable additional expense, as electricity costs spiral out of control and upgrades or conversions become necessary. As well as jeopardising cost efficiency, badly planned precision air conditioning can also be detrimental to the ability of a data centre to operate efficiently in the future and meet its Power Usage Effectiveness (PUE) targets.
Knowledge is power
There has been a recognised problem, whereby standardised air conditioning systems have not performed to the levels expected once in-situ. To address this, some manufacturers now offer customers simulations and performance tests under realistic operating conditions.
This allows specialist planners and operators to gain essential data on actual performance and efficiency ratings during the planning stages of large air conditioning systems. In the UK – due to strict rules governing the accuracy of the stated performance data – tests of this kind are now standard procedure among manufacturers, with end users increasingly demanding field based information on cooling capacity and energy efficiency, so that they can remain economically competitive. For this reason, manufacturers now also offer individualised customer tests.
Stulz built a test centre for the internal testing of prototypes in the design and development phase as well for customers, data centre planners and operators to test their facilities.
With an area of more than 700m², it features four conditioning systems, where airflow rates from 500m³/h to 55,000m³/h can be achieved. It has two separate climatic chambers, in which air conditioning systems can be put through technical tests either individually or connected via both chambers, as shown in figure 1. Here, different operating parameters, such as environmental influences (-20°C to +50°C) and return air conditions, can be set precisely to match a customer’s requirements, thereby simulating realistic operating conditions. While this is taking place, engineers in the control room record and document test data in real time, and analyse it if necessary.
Specialist planners and operators can use the information gathered from these tests as verification of cooling capacity, efficiency and power, creating an important aid to decision making during the specification of precision air conditioning solutions.
Technical tests of air conditioning units in accordance with DIN EN 14511 (performance of air conditioners, liquid chilling packages and heat pumps), EN 1216 (heat exchangers and forced circulation air cooling and air heating coils) and ISO 9614 (sound power levels), can be completed.
Using the air enthalpy method, system performance is ascertained by measuring the airflow rate and the associated intake and outlet conditions of the air. The calorimetric method, on the other hand, is particularly suitable for simulating partial load conditions during full load tests. Here, three important scenarios are used: conditioning mode; simulation of data centre cooling with supplementary cold or hot aisle enclosure; and environmental simulation mode that tests entire air conditioning systems with indoor and outdoor units.
As all scenarios permit the variable setting of heat and air volumes, air humidity and return air temperature, a customer’s specific local requirements can be simulated with great precision. Let’s look at each one in more detail:
Scenario 1 – Conditioning mode: Conditioning mode is a common standard test that simulates conventional closed circuit air conditioning, with or without a raised floor. The almost unlimited choice of operating conditions enables the performance and energy efficiency of the test object to be measured for a great variety of applications, as shown in figure 2. The test records and documents all the important performance data and measured values of the air conditioning system.
Scenario 2: Simulation of a side cooler system with cold aisle enclosure: The second test scenario simulates the row cooling equipment with a cold aisle enclosure that is commonly used today in small and medium-sized data centres. By separating the cold supply air and hot server air with partition walls, this method effectively prevents chaotic air conduction and ensures that airflows at different temperatures cannot mix. This means that the required cooling capacity and, consequently, the energy consumption are considerably reduced, as shown in figure 3. A further benefit of separation is that the return air temperature can be controlled, so that the air conditioning units can be kept at an ideal operating temperature from an energy efficiency perspective – ASHRAE recommends a server inlet temperature of up to 27°C. This temperature enables the especially efficient operation of cooling system components, such as compressors, while simultaneously protecting the sensitive IT equipment.
Scenario 3: Testing entire air conditioning systems with indoor and outdoor units: If a facility has two separate climatic chambers, it is possible for entire cooling systems to be tested in combination, along with their indoor and outdoor units. In the first climatic chamber, conditioning systems generate the desired heat load, which equates to the data centre’s expected IT load, and the indoor unit under test then cools this air down. The second climatic chamber simulates the country’s specific environmental conditions, in order to reflect heat removal via air cooled condensers, air cooled heat exchangers or chillers.
These individual test scenarios enable precisely the right dimensioning of air conditioning systems in data centres, under consideration of local environmental influences. In this way, users have full cost control as regards investment sums and expected future running and energy costs. The use of test centres offers a further advantage, however, as the extensive accompanying documentation can be used not only to verify the performance of air conditioning systems, but also as a basis for configuring the data centre cooling. It is therefore worthwhile for all customers of air conditioning solutions to take up services of this kind that are offered by manufacturers.