Keeping IT cool

Howard Aiken, one of the pre-war pioneers of modern computing, once said: "Don't worry about people stealing an idea. If it's original, you will have to ram it down their throats."

Aiken, who died in 1973, would have been surprised by developments in IT and comms room technology. A man of insight but possibly not foresight, he may have regretted his comment that: "Only six electronic digital computers will be required to satisfy the computing needs of the entire United States".

His Mark I computer was 17 m long and 3 m high – just about the size of your average modern comms room. Silicon technology and the subsequent boom in global telecomms have changed all that. Both voice and data transmissions, and the growth of the Internet, have resulted in the development of new types of electronic equipment (blade servers, routers, etc) designed to carry the heavy levels of traffic associated with such operations.

Nearly every new office building currently under construction incorporates somewhere within its remit a designated area for a comms room. Likely as not, it's not as big as the office space. Likely as not it's tucked away in a corner somewhere. There's probably no clear client brief and it may just be separated from the main office area by the simplest of partitions. In the majority of cases the solution seems to be to treat the comms room as an add-on to the main offices and design accordingly, simply adding a few more comfort cooling units whether vrf or one-to-one, onto the existing air conditioning schedule.

Splits in comms rooms however, are not the most effective solution. Instead, this article makes the case for specialist air conditioning in the form of sensible cooling units. There are a number of specialist sensible cooler products on the market specifically designed for this type of application, so why isn't it standard practice to use them?

The common themes
Anecdotal evidence and a straw poll of consultancy practices large and small, suggest some answers. A number of common themes seem to run through the reasoning.

The main ones, and indeed the order of importance in which they were placed, can be summarised as:

• Cost: the average split unit is thought to be far cheaper than a sensible cooler.
  
• Technical sufficiency: a split is thought good enough to do the job.
  
• Space: a split is thought to be smaller.
  
• Ease of design: some of the larger split manufacturers offer consultants a design service for their air conditioning solution.

So do the arguments stack up in the face of detailed analysis?
A quick glance through a typical split manufacturer's data shows it's not easy to make direct comparisons between splits and sensible coolers, model for model. The majority of split manufacturers use Asia-Pacific conditions as the basis for their nominal figures. However, a 27ºC room temperature, on which nominal figures are based, is too high for comms rooms which normally operate at between 21ºC and 23ºC. Equally crucial, the comparison should be between the units' sensible cooling capacities (see table 1) as latent cooling is of no value in this type of environment.

A number of components make up the life-time costs for any air conditioning system:

• cost of supply;
  
• cost of installation;
  
• energy consumption costs;
  
• operation and maintenance costs.

The tabulated comparative data presented in this article is based on a range of wall mounted split systems from one of the market leaders in split systems and a comparable vertical upflow range from a leading sensible cooling unit supplier.

Costs and savings
Table 1 indicates the importance of comparing sensible cooling capacities as, in the case of splits, these are 30% to 45% lower than the nominal figures. Note the differences in sensible heat ratios between sensible coolers and splits.

Installation costs vary enormously from project to project subject to pipework lengths, pipe routing, ease of access etc. Table 2 provides a simple comparison between some typical installations. The figures allow for the fact that splits are usually supplied with a significant refrigerant charge. In mitigation, sensible coolers operate with discharge gas rather than suction lines for which pipes are smaller, require no insulation and are therefore cheaper.

New ISO 14001 standards include the need for sites to document and make available to the public a clear environmental policy. In addition, procedures must be established for ongoing review of the environmental aspects and impacts of all products, activities and services.

In the light of ISO 14001, table 3 details probably the most surprising and least known data in this field, ie that sensible cooler units offer major energy savings in comparison to splits. Similar conclusions have been drawn by independent analyses of these issues. Dr Colin Hawthorne concluded in his article 'Design issues for air conditioning split systems' (BSj, 03/99) that "specifiers should be very wary when selecting comfort cooling systems". The coefficient of performance (cop) figures require close inspection.

Normally, one might compare the total cooling output to the power input. However, in this application it is the sensible cooling capacity that is relevant. The effective cop is therefore the ratio between the sensible cooling capacity and the total absorbed power. It is notable that the cops for splits are significantly lower than for the equivalent sensible cooler. Convert these figures into energy consumption and some fairly staggering results appear: the annual energy costs for splits are, in many cases, comparable to the capital expenditure for the cost of supply.

Notoriously difficult to measure, operation and maintenance costs for individual items of equipment are rarely monitored by the majority of users. However, one sensible cooling unit manufacturer has produced some long term costs based on analyses by a number of international users (figure 1).

Detailed analysis of data detailed in tables 1-3 indicates that the additional capital expenditure for a sensible cooler (of the order of £800 per system) is recovered within 2-3 years by the cost benefits of more efficient operation. This analysis is further supported by user data (see figure 1) from real working environments. If the long term data between 6-10 years is as consistent with analysis as the short term, the benefits of sensible coolers become even greater.

It is difficult from this data to dispute the view that, in comms rooms, sensible cooling not splits is the lower cost option.

Technical sufficiency 20 years ago, any room that contained technological equipment was subject to reasonably stringent temperature and humidity controls (see figure 2, area 1). By the turn of the millennium, the majority of comms equipment including computer servers could operate satisfactorily under widely varying humidity conditions if somewhat restricted temperature constraints (see figure 2, area 2).

Current practice for cooling these environments is as diverse as the comms rooms themselves. Design parameters reflect more on the IT manager's perceptions and operators' comfort than on the equipment's particular requirements. However, while design parameters may vary, the fundamental physics remain.

Despite the IT industry's well known affection for hype, ask any IT manager and he or she will tell you that current heat load densities really are averaging 750 W/m² and rising. You'll also be told that 750 W/m² gives very little room for manoeuvre in the event of a crisis. For example, switch off the air conditioning in the comms room and tests show that the temperature will go critical within 15 minutes. All IT equipment will crash and five seconds later business, all business, will shut down. Then wait for the call from the finance director.

The right type of cooling
Cooling of comms rooms has therefore two basic requirements: enough cooling and enough of the right type of cooling ie sensible, not latent, cooling.

Table 1 details the sensible cooling capacities in comparison to the headline, nominal capacities. Calculations should only be based on the former.

It is also worthwhile considering the effects of the low sensible heat ratios inherent in splits. The net effect is the wrong type of cooling ie latent cooling, and dehumidification of the space. If the comms room has no vapour barrier, and this is quite likely, moisture will be drawn out of the surrounding offices creating a drier and drier environment, or a need for humidification. Alternatively, a vapour barrier is inserted and dehumidification of the comms room continues until such time as an equilibrium humidity level is established, in some cases as low as 35% rh. Such levels lead directly to static build-up and potentially cataclysmic static discharge. A problem that 1980s computer room designers overcame by using anti-static measures inherent to raised access floors – unfortunately not available in many comms rooms where equipment is mounted directly onto the slab.

Consider also a largely unknown effect of high sensible loads, unique to Splits. Analysed in detail by Dr Hawthorne (BSj, 03/99), the depression of the refrigerant evaporating temperature in splits is a perverse phenomenon. The less latent heat load, the lower this temperature falls. At a certain point, sensors in the split designed to protect the compressor will shut the system down. The result is a temporary loss of cooling. Not significant in lowly occupied environments; highly problematic as the loads pick up.

Some of the many split ranges offered to the market have strong reputations for reliability. However, the basis of their design is to cool office environments whose daily working time is between seven and 10 hours. For much of this period the fans and compressors are silent as the occupation levels, solar gains etc rise and fall. Mostly, units cool in summer and heat in winter. There is little evidence to support their use in continuous cooling 24/7 operation, 365 days of the year.

Sensible coolers on the other hand are generally fitted with fan motors with mean time between failures of around 100 000 hours (in excess of 10 years continuous operation). Still, splits remain the air conditioning of choice. As real heat loads rise, expect to see the chickens coming home to roost.

Insects and engineers' fingers
Walk out of an Underground station, blow your nose and you'll not only see how effective a filter you have fitted to your face, but also why you need one - little wonder the escalators are always breaking down. Ask an air conditioning engineer to look at the air filter on a sensible cooler and they'll see the same thing. On the other hand, split unit filters are just about sufficient to prevent access to large objects such as engineers' fingers and insects - an exaggeration, but not much of one.

The optimum solution is a raised access floor, floor grilles and a downflow type, vertical sensible cooler. As this kind of luxury is not always available, analysis must be made on the basis of solid floors and upflow or wall mounted units. In a small room with no vertical obstructions, wall mounted splits do have sufficient air throw to provide a cool environment. However, with air throws limited to 5-6 m, no possibility to take advantage of ceiling Coanda effects (the inlet air grille is above the outlet) and limited directivity, air distribution in the majority of comms rooms is bound to be poor.

Better distribution is available from ceiling mounted cassette units but the implications of a condensate spill do not bear thinking about. (This hasn't prevented their use in the field, but this author would question not only the technical but also the commercial validity of this solution). Note that ceiling mounted cassettes are significantly more expensive than their wall mounted counterparts.

The sensible cooler solution is to use more powerful fans and air grilles capable of deflecting air towards the ceiling and above vertical obstructions.

The issues of cooling capacity, reliability, static electricity, air quality and air distribution are key to the long term operation of comms rooms. As occupancy levels of comms rooms grow, the technical competence of the air conditioning solution will become more and more important. Sensible coolers seem to address these issues to a far greater extent than splits.

Space
It's not the space that's filled, it's the space occupied that counts.

Very often design engineers don't have sufficient space allocated to them for services, especially in comms rooms. Naturally, they look for the smallest air conditioning solution; but do wall mounted splits provide it?

Table 4 shows dimensional data comparisons between sensible coolers and splits. There is an impression that wall mounted units are inherently smaller than their clunky upflow unit counterparts. But look at any wall mounted installation and you'll see that the space actually occupied extends from the top of the unit down to the floor. No one should consider using the space below the split for siting equipment as a condensate overflow would be catastrophic. No operator should consider sitting underneath one, not without several jumpers and a thermos flask. Any other use would block air flow or restrict access for a service engineer. So the only dimension that's worth comparing in practical terms is the width of the unit. Note that the width of the split is almost double that of the equivalent sensible cooler.

The larger split manufacturers offer a design service which significantly simplifies the life of the time-pressed design engineer. The challenge for sensible cooler manufacturers is to provide a similar level of service.

Take note
Recent studies have detailed the proportion of costs for a typical building over a forty year life span. Most notable is that operational costs make up almost exactly 50% of the total, and that a large part of these costs are related to energy consumption.

Consider, for a moment, a typical office building. Excluding solar gains, which vary significantly from site to site, office loads are approximately 50-60 W/m². Somewhere on each floor there will be a little hot spot (usually titled comms room) which will constitute anything up to 10% of the office space and have a heat load density of 750 W/m² (solar gains also excluded).

It may well be, therefore, that during office working hours the energy consumption of the comms room cooling is greater than that of the whole of the office space cooling.

But that's not the whole story. When the workforce has long since gone home and set-back functions rule, the comms room churns away deep into the night; updating, down-loading, backing-up and billing. The cooling rolls on and the energy bill mounts. 16 long hours pass before the office cooling rejoins the comms room cooling, and starts to do its work again.

During the next few years more column inches will be written on the subject of air conditioning buildings than almost any other area of m&e services. Whole office buildings are being designed with one aim in mind: to reduce energy consumption. And yet, the comms room remains mainly under-considered.

The government has set a target to reduce emissions of greenhouse gases by 20% before 2010. While ambitious, this target challenges engineers to evaluate current practices and where necessary introduce significant change.

When linked to the requirements of the Energy Performance of Buildings Directive (see BSj 05/03) and ISO 14001, even greater attention will be focused on the overall environmental performance of buildings, both new and old.

Design implications
Specifiers should, therefore, question all of the following items before specifying a particular type of comms room cooling:

• the room heat load density;
  
• the required sensible heat load;
  
• the manufacturers' technical data;
  
• the room temperature conditions on which the technical data is based;
  
• the cop of the systems under UK conditions.

This author believes that such analysis will demonstrate that sensible coolers, not splits, are the sustainable solution to comms room cooling and that they provide clients with massive whole-life cost savings.