Airflow, in the context of a data centre environment, is a measurement of the amount of air per unit of time that flows through server cabinets.
Typically, it’s expressed in cubic metres per hour or cubic feet per minute.
Each ICT device inside the data centre requires a certain amount of airflow to work within its design’s temperature range.
Any deviation from the recommended temperature range results in incremental increases in power usage as the fan speed of devices would likely go up to battle the temperature irregularities. That leads to shorter life spans and, subsequently, sudden failure of these devices.
Naturally, none of us would accept irregularities in a business-critical environment. Any failure of the ICT devices can potentially result in revenue losses which we don’t want to be responsible for.
With the constant increase of utility cost and migration, and the consolidation of legacy servers to a virtual environment, the power usage of individual server cabinets is increasing exponentially.
Nevertheless, the power efficiency of individual servers is improving and the spatial requirement for each is far less than it used to be. More servers can be housed in server cabinets than was possible a few years back.
As a result, the amount of heat dissipated by each server cabinet is significantly higher than legacy server cabinets. Considering this trend, it’s becoming vital to have an efficient cooling environment to cope with these changes.
Across the globe, most key players in the data centre space have set carbon pollution reduction targets to reduce the amount of greenhouse gases emitted due to energy usage in their data centres.
Cooling equipment, being the biggest energy consumer in data centres after ICT equipment, is being targeted by most large organisations to ensure it’s operating as efficiently as possible.
Fortunately, most ICT equipment can now operate at a higher temperature bracket as per ASHRAE guidelines.
About data centre cooling parameters
In order to substantially reduce energy consumption of cooling units, the return-air temperature should be increased to the maximum acceptable amount. One can say that efficiency and power consumption of cooling units have a direct correlation with the supply and return-air temperature difference.
Generally, the higher the difference, the higher the efficiency of units.
This differential is also referred to as ∆t.
There are several aspects that lead to lower temperature difference: excessive cooling air, cooling-air bypass, poor airflow distribution, and the improper layout of ICT cabinets to name a few.
In most existing data centres, as per the legacy approach, the cooling air is usually supplied at low temperatures (for example, 13 °C) as a precautionary measure against hot spots. As mentioned earlier, based on today’s server technology, the supply air temperature can be increased without risking the operation and/or life span of the servers.
So now that we have a reasonable understanding of the importance of cooling parameters in data centres, we can take a look at cooling layouts in the environment.
There are a number of methods to ensure that cooling equipment is installed to fit the purpose and not be over-engineered.
However, most of these methods are practised by data centre/facility managers for existing data centres, who predominately use a trial-and-error approach.
So it’s more of an after-the-fact exercise.
If we’re building a new data centre, how can we ensure the cooling system chosen will match the demand and still keep the energy consumption under control?
The answer lies with predictive computer modelling.
The benefits of using airflow modelling software for a data centre environment
Let’s look at five benefits of using predictive computer modelling to control data centre airflow:
1. Optimising airflow distribution and cooling resources
In the conventional trial-and-error approach, the optimal airflow distribution is achieved through a number of iterations. This is time-consuming and quite often ends up in an inefficient design.
It means that you supply an excessive amount of cooling which is more than the servers need. If you are thinking about reducing your electricity bills and carbon footprint, this wouldn’t work out for you.
By using computer software, different cooling and ICT equipment layouts can be modelled to achieve the best outcome without providing excessive cooling and at minimum cost. Once you work out the optimum configuration and layout, the other design parameters can be planned so that double-handling — reworking or replanning the ICT and cooling equipment layout — can be avoided during or post construction.
2. Reducing energy consumption and increasing energy efficiency
One of the key criteria that drives the efficiency of a cooling unit is the return-air temperature. In principle, as the return-air temperature rises, the cooling unit becomes more efficient.
For instance, increasing the return-air temperature from 23.9 °C to 26.7 °C can increase the efficiency of a cooling unit by approximately 15-20% (subject to type and size of the unit).
Modelling software assists you to verify that an increase in temperature range won’t have an impact on the cooling demand of the ICT equipment and, therefore, enable energy consumption to be reduced.
3. Increasing system cooling capacity
Invariably, organisations don’t have sufficient funds or the luxury of time to overhaul the cooling environment of their facilities as this will mean outages.
So how can they gradually improve the environment based on what they have?
The first step is to understand if there’s sufficient cooling available to the data centre based on the heat load and airflow demand of the equipment.
Subsequently, you plan to ensure the cooling units are configured to deliver the maximum amount of cooling to the room without overloading them.
Modelling software assists you to optimise the airflow distribution and rack layout, and increase the air temperatures at the inlets of the cooling units (that is, supply-air temperature).
In this way, you can accommodate more servers without purchasing new cooling units and going through the complexity of their installation.
4. Reducing design and construction expenses
At the beginning of design, you’ll need to answer several questions:
- What is the best layout for my new data centre?
- Should I go with a hot-aisle containment or cold-aisle containment design?
- Raised floor or no raised floor?
When it comes to the best layout for a data centre, the vital factor to determine is the efficiency and effectiveness of a cooling solution for that layout.
Any of the former approaches can be used for your environment, however, if the cooling system that is designed for them isn’t verified, relocating the equipment or changing the layout will become expensive and time-consuming.
For expanding an existing data centre, you need to ensure that the modifications don’t have an impact on the cooling of the existing equipment.
Modelling software assists you to predict the airflow and temperature distribution and study the changes in layout to achieve the desired distribution of cooling air and temperature throughout the data centre.
For a data centre that isn’t fully populated on day one, determining the quantity of required cooling units to match the heat load, as well as pinpointing their placement for best airflow distribution, can be achieved simply by utilising the modelling software.
5. Improving reliability and reducing downtime
You can use modelling software to study failure scenarios for a data centre and ensure that the operation of the business-critical computer equipment is not negatively affected by failure of the cooling units.
More importantly, the results of these failure scenarios can be used to identify what areas are affected by the failure of each cooling unit.
Having this information on hand can save you a significant amount of time and expenditure to find the necessary precautionary measures to address each case. This can be as simple as having spot cooling units on site to condition the air around the affected sections or turning off the servers before the high temperatures can have an impact on the internal electronics.
Take advantage of the benefits of airflow modelling software
Data centres require a consistent and acceptable temperature range so that no ICT equipment will fail due to environmental irregularities.
Without predictive thermal and airflow modelling, optimal cooling design can’t be guaranteed.
And that means energy consumption targets set by organisations won’t be met.