S 1.27 Air conditioning of the technology / in technology rooms

Initiation responsibility: Building Services Manager, Head of IT

Implementation responsibility: Building Services

In order to reliably operate IT equipment on a sustained basis, it must be ensured that the environmental conditions are within the limits specified by the manufacturer. The term "air conditioning", when used in this context, always consists of regulating the following four air conditioning parameters:

Maintaining the temperature within prescribed limits is the most important task of an air conditioning system. Almost all of the electrical energy provided to the IT will be dissipated in the form of heat and needs to be removed from the area. If the normal exchange of heat and air in a room is inadequate, it will be necessary to install additional cooling.

In addition to regulating the temperature, the humidity will also need to be kept within certain limits in many cases in order to avoid electrostatic discharges (when the humidity is too low), oxidation, and the formation of mould and mildew (when the humidity is too high).

The suspended particle content of the air is usually already low enough due to the filters normally used in air conditioning systems. Additional filtering is usually only required when special hardware is used or when the air in the environment has an especially high particle content. In order to guarantee that the required flow rate is provided, the air conditioning system filters must be checked regularly and replaced promptly whenever necessary.

The fourth air conditioning parameter, the amount of fresh air mixed in, is irrelevant for actual IT operations. However, if the air conditioned spaces are used as workplaces, fresh air will need to be mixed-in in accordance with the relevant workplaces ordinances.

An air conditioning system must have adequate capacity to serve its primary purpose. If certain differences in the power consumption of all the IT systems are taken into account, it can be assumed in an initial approximation that every kilovolt-ampere (kVA) of electrical energy will generate 0.8kW to 1kW of heat.

The cooling capacity should be based on an exact thermal load calculation and a generous reserve capacity added to this value and it should be easily expandable. The actual thermal load placed on the areas cooled must be checked by recalculation or by taking measurements at regular intervals (about every 12 to 24 months), as well as after any major changes to the IT hardware. Measurements should be taken at different times of the day to determine whether the air needs to be humidified or dehumidified.

When performing the calculation, maximum outdoor temperatures of up to 40°C should be assumed due to the increasingly high temperatures in the summer, which can in turn lead to a higher cooling requirement. On the other hand, modern IT devices allow the air temperature to be 30°C or even higher, which may then reduce the total cooling capacity required.

The range of reasonably usable technologies covers simple split units (a cooling unit in the IT room and a heat exchanger located outside) to highly complex air conditioning systems, depending on the amount of heat to be removed. Every solution should be examined to determine how it will react to a short interruption of its power supply. While simple split units will just shut off and then turn back on again once power has been restored without any problems, this is usually not the case for large air conditioning systems.

Air conditioning systems, due to their enormous electrical power consumption, are almost never supplied with power by a UPS. For this reason, even minor interruptions in the power supply will cause them to malfunction and shut down. Even if they are supplied with power by an EPS that starts up quickly, the cooling capacity available before the interruption may not be available immediately. For technical reasons (to protect against icing, for example), air conditioning systems are started up in several stages. This means it can easily take from 10 to 15 minutes until restoration of the full cooling capacity.

During this time, the IT is usually supplied with power by an UPS or EPS and continues to generate heat. If this heat is not removed or not enough heat is removed, massive damage due to overheating or even total failures may result from an interruption of the air conditioning system.

When the air conditioning fails and the air is not cooled any more, the room temperature can climb to significantly over 60°C after just 3 minutes! This short amount of time is usually not even long enough to properly shut down the IT systems. Therefore, it should be examined in every case how long an interruption of the cooling system could last, what the consequences of such an interruption could be, and which countermeasures should be taken.

The usual method used in this case is to install a cold storage system. This may include an ice storage unit installed specifically for this purpose, but it is also possible to use an existing extinguishing water tank as well. The superfluous cooling capacity produced in normal operations is used to cool the cold storage and then use this cold air or water when needed. When the air conditioning system is designed accordingly, it is possible to provide cooling capacity almost immediately after power has been restored (regardless of whether the power company restored power or the EPS was activated).

The energy density in modern computer centres is constantly increasing. While an energy density of 500 W/m² (low energy density) was common even in the 1980s, energy densities anywhere from 5 to 10 kW/m² and higher (high energy density) are not unusual at all today.

For high energy densities, the conventional method of cooling the air under the raised floor and allowing it to circulate through the racks in the room is not adequate any more. Nowadays, there are rack-based high performance cooling systems available on the market that meet current requirements.

An air conditioning system is connected to one refrigerant and one condensate pipe at a minimum, and if the air is humidified, then also to a water pipe. Safeguard S 1.24 Avoidance of water pipes therefore needs to be taken into account in each case. This safeguard can be applied to ensure that a leak in a heat exchanger or the failure of the cooling system will not result in any damage due to moisture.

The air conditioning system must be serviced regularly in order to maintain its protective effect. It is recommended to install an additional monitoring unit for the air conditioning system.

Occasionally, the control behaviour of an air conditioning system for temperature and humidity reaches limit areas not yet to be considered errors, but with the power of causing unexplainable malfunctions of the IT. Such shifts regarding the control behaviour are often accompanied by changes regarding the IT utilisation, the external temperature, or other time-variable parameters. In order to be able to establish clarity regarding the relationships and therefore regarding possible causes, it is recommendable to record the two parameters temperature and humidity at least for a week in increments of 15 minutes in the event of suspicions. If performing this fully electronically is not possible, at least a conventional thermo hygrograph with 7-day drum should be kept at hand in order to be used at any time.

Heat exchangers of an air conditioning system must be protected against lightning when installed outdoors. If there are high or very high availability requirements, the heat exchangers should not be generally accessible and should at least be physically protected against sabotage, if necessary.

The air conditioning technology must be taken into account during contingency planning (see module S 1.3 Business continuity management).

Review questions:

© Federal Office for Information Security. All rights reserved.