By Paul Johnson, UK Data Centre Segment Leader, ABB www.new.abb.com
Adapt to demand by adopting power distribution systems based on standard modules to deliver an elastic approach to critical infrastructure
On the one hand, data centre operators experience variable demand on servers and therefore variable power load and deployment. But on the other hand, they face fixed costs from the fabric of data centre buildings and the grid connections that feed them.
When surveyed, 64% of data centre experts identified scalability as their number one issue as it sets them the challenge of meeting growing demand at the same time as maintaining uptime and operational efficiency. Having power infrastructure that is flexible and scalable will help them to better meet the needs of their servers – and ultimately their customers.
Elastic critical infrastructure
The concept of Elastic Critical Infrastructure (ECI) has developed to support this through the use of standard products for medium voltage switchgear, UPS systems and also low voltage switchgear. These are configured in a smart way so that data centre operators can use it to power a pay as you grow approach.
The concept enables data centre operators to use many small design blocks to build a large-scale system over time to meet the evolving needs of a growing data centre. The key is to use standard products that are readily available on short lead times, and that can be deployed at scale. Smart control and communication capabilities help manage these systems and leverage operational efficiency gains.
Improving asset utilisation
In a basic 2N system architecture operators use two parallel systems, each of which is capable of meeting the data centre’s entire load. The benefit is the certainty that the infrastructure will supply loads. However, it comes with the drawback that operators can never achieve more than 50% utilisation from infrastructure.
By using the ECI approach with many small modules, operators no longer need to rely on one of two systems operating. The approach is perhaps most valuable for UPS systems, which are relatively costly per kW delivered in comparison with medium and low voltage switchgear and transformers. For example, with four 500 kW UPS modules to support a 1.5 MW IT load, a malfunction of one module won’t affect overall power availability and the remaining three modules can be used as a group to meet the entire power demand.
By building in smaller design blocks, operators are able to deploy various architectures, even changing between architectures as the facility is fitted out. Depending on the size of the block that’s used, a facility could start out with two 1MW power streams in a 2N architecture, and as the facility increases in capacity, an additional power stream could be included to reconfigure the system to be a distributed redundant – three to make two architecture.
In general, the flexibility of a system grows as an operator installs more modules, offering more options to create load groups. In addition, the impact of any single module reduces – with the important implication that operators can downsize overall capacity and increase the utilisation factor (UF) of each module and of the system as a whole.
Reduce stranded capacity in UPS
Traditional monolithic UPS systems contain multiple UPS modules and shared components such as switchgear. This provides a straightforward approach but has the drawback that capacity can become stranded as the UPS module ratings are sometimes quite large. In addition, a failure on any single UPS module can impact the availability of the UPS system as a whole.
With the ECI approach, modular UPS systems are employed and include multiple power modules. For example, some modular UPS systems have their own dedicated control logic, static bypass, user interface and switchgear that enables each power module to act as a complete UPS. The result is greater levels of flexibility to group power modules to feed loads and less stranded capacity and higher availability of backup power.
How ECI influences transformers & circuit breakers
The laws of physics mean that a fault on a large single transformer can give rise to high fault levels. As an alternative, using multiple smaller transformers will meet the same capacity but with a significantly reduced fault level. In turn, operators can downsize their protective devices (ACBs and MCCBs), with the potential to find significant cost savings and also provide a safer system because the lower fault levels mean reduced incident energy in case of an arc flash event.
Communication and control for ECI
Switching to the ECI approach offers huge opportunities for capital savings. However, the smaller blocks do have more components so implementing digital metering of current and voltage, as well as monitoring of temperature and component wear, will provide accurate and timely data for data centre infrastructure management (DCIM) systems. This will optimise operations by making decisions based on availability, status and condition of critical power.