Mobile internet, cloud computing and ever more indus­trial networking have led to a massive increase in the volume of data being processed in data centers. Cooling is the key to effi­cient oper­a­tion. Energy consump­tion is the main cost factor – and the cooling system is a crucial aspect. At present it accounts for around 37 per cent of the energy costs on average – and the figure is even higher with older systems.

Energy costs: the key to econom­ical oper­a­tion

Figure 1a: FanGrid made up of back­ward-curved RadiPac centrifugal fans.

All efforts to cut oper­ating costs there­fore hinge on this item. Effi­cient cooling concepts these days are often based on free cooling and many do not require the use of compres­sion refrig­er­a­tion systems. Modern instal­la­tions use the outside air, often in combi­na­tion with adia­batic cooling (evap­o­ra­tive cooling), to provide appro­priate air condi­tioning for data centers. Such cooling concepts do however require a large volume of air. Use is increas­ingly being made of so-called FanGrids – a system of centrifugal or axial fans oper­ating in parallel – to satisfy the demand for a higher air perfor­mance (Figures 1a and 1b).

Air flow routing

Figure 1b: FanGrid with axial fans.

A distinc­tion is made between indi­rect and direct free cooling systems. Indi­rect free cooling employs two sepa­rate cooling circuits. The cool outside air is not routed directly into the data center but is rather used to cool the circu­lating air flow in the data center by way of a heat exchanger. By contrast, a direct free cooling system draws in the cold outside air, filters it and routes it directly into the data center. Addi­tional outside air filters are required with this method to ensure air quality and purity. The prin­ciple employed ulti­mately depends on the require­ments, loca­tion and size of the data center concerned.

Customized FanGrids

ebm-papst helps customers to design an ideal FanGrid with the support of the company’s Product Selector soft­ware FanScout (patent pending). The most econom­ical system is worked out on the basis of para­me­ters such as the instal­la­tion space avail­able, the required oper­ating points and the desired level of redun­dancy. The soft­ware also takes the life­cycle costs into account, in other words the purchase price and oper­ating costs over a defined period.

Figure 2: The blue dots symbolize various oper­ating points. The size of the dot reflects the number of oper­ating hours at this oper­ating point.

In the past, the oper­ating point with the highest air flow (maximum oper­ating point) often formed the basis for FanGrid design. This is however seldom attained – usually only if the data center is oper­ating to full capacity with high outside temper­a­tures in summer. Most of the time a data center cooling system runs at part load. For this reason the design soft­ware from ebm-papst allows up to five different oper­ating statuses (oper­ating points) to be spec­i­fied. The applic­able oper­ating time in hours per year is stored for each of these oper­ating statuses. This produces weighted oper­ating points which reflect oper­a­tion for the year as a whole.

Figure 3: If the annual energy consump­tion of the FanGrid is calcu­lated on the basis of weighted oper­ating points as opposed to the maximum oper­ating point, the real energy consump­tion will be 20,000 kWh lower, as shown by the calcu­la­tion example for a FanGrid with four RadiPac centrifugal fans.

An example is shown in Figure 2. Real­istic figures for the expected oper­ating costs can be calcu­lated from these points. For this purpose, the soft­ware config­ures all the possible FanGrid combi­na­tions (type, size and number of fans) and works out the most energy-effi­cient alter­na­tive. When viewed over the course of the year it is quite possible that the combi­na­tion with the greatest effi­ciency at the maximum oper­ating point does not neces­sarily produce the best consump­tion figures on the basis of the weighted oper­ating points.
The weighted oper­ating points enable energy consump­tion to be calcu­lated far more accu­rately. By way of example, Figure 3 shows the energy consump­tion calcu­la­tion for a FanGrid with four RadiPac fans. The left bar repre­sents energy consump­tion calcu­lated on the basis of the maximum oper­ating point (approx. 70,000 kWh). The right bar (approx. 50,000 kWh) shows the actual annual energy consump­tion of the FanGrid calcu­lated using real­is­ti­cally weighted oper­ating points.

All the latest tech­nology

Figure 4: With the new RadiPac, partic­ular atten­tion was paid to the air intake in the impeller, the posi­tioning of the external rotor motor in the impeller and the blade profile of the impellers.

Such savings can be achieved through the use of EC fans in FanGrids. These are highly effi­cient and can always be regu­lated to the required oper­ating point. The new RadiPac EC centrifugal fan for venti­lation tech­nology has been on the market since October 2015 (Figure 4). These fans are not just 13 per cent more effi­cient than their prede­ces­sors, the noise level has also been cut by more than 3 dB (A). Opti­mized outflow char­ac­ter­is­tics ensure the best possible flow control even when there is not much room avail­able (Figure 5, see page 34). The new RadiPac in FanGrids is thus the ideal space-saving solu­tion for effi­cient oper­a­tion in data centers.

Figure 5: The new RadiPac EC centrifugal fan is not just 13 per cent more effi­cient than its prede­cessor, but also 3 dB (A) quieter. Opti­mized outflow char­ac­ter­is­tics ensure the best possible flow control even when there is not much room avail­able.

Instal­la­tion losses are another factor which tends to be over­looked in prac­tice. If fans are installed too close together, they will influ­ence one another. As a general rule, the greater the volume of air to be conveyed, the further apart the fans should be. ebm-papst’s design soft­ware auto­mat­i­cally allows for possible instal­la­tion losses. FanGrids often feature built-in redun­dancy. If one fan fails, the speed of the others is auto­mat­i­cally increased to provide the best possible compen­sa­tion for the loss of air perfor­mance. This does however have the following conse­quence: On account of the fan failure, some of the air produces so-called back­flow. The asso­ci­ated losses depend on the oper­ating point and must be taken into account at the design stage.

To sum up

ebm-papst FanScout: Completely reli­able and, above all, precise data as the soft­ware is based on true measured values. Not only is the perfor­mance of the indi­vidual fan compo­nents measured but also that of the fan as a complete system – as confirmed by TÜV SÜD.

ebm-papst starts by taking a look at the specific situ­a­tion, including the instal­la­tion space avail­able, the oper­ating points and the required level of redun­dancy, to ensure that the free cooling concept works with an optimum FanGrid design. Bearing in mind the life­cycle costs, the most appro­priate system is then defined in terms of the type, size and number of fans to be used. Spacing and arrange­ment are further impor­tant factors when designing FanGrids. As a rule of thumb: The greater the air volume, the larger the space between the fans. Only then will the desired perfor­mance and effi­ciency be obtained.

The Product Selector soft­ware FanScout permits the weighting of different oper­ating points. Conse­quently the system design is not based on the maximum oper­ating point as is often the case, but rather geared to indi­vidual customer require­ments. In connec­tion with the weighted oper­ating points it is also possible to simu­late various oper­ating scenarios such as a constant air flow or constant pres­sure – making FanGrid design more effi­cient and cutting oper­ating costs. Offering sophis­ti­cated EC tech­nology and a wealth of exper­tise, ebm-papst can help customers find the optimum solu­tion for their cooling concept.