While UPSs have been an integral part of data centre installations for many years, expectations of their performance have increased steadily – and current trends in the data centre arena are set to increase this pressure further.
As customers expect on-demand access to services in a 24/7-transaction environment, businesses are looking for UPS systems, which counter the falling tolerance of data loss due to power interruptions. High UPS availability therefore becomes essential, yet it must be achieved with high electrical efficiency, as a number of factors will potentially affect the UK’s energy capacity, whilst driving prices up.
In this article, Ian Jackson, General Manager at Kohler Uninterruptible Power, a Kohler company, discusses the latest in modular UPS topology and shows how data centre managers can use it to achieve the power availability and efficiency vital to today’s business climate.
If you’re responsible for a data centre or other IT installations of any significance, the drive to secure the maximum possible availability from your UPS system will no doubt have intensified steadily over recent years. Organisations increasingly conduct their transactions online, while their customers rely on real-time responses on demand, 24/7.
In this environment, being forced offline through a system failure could have severe commercial and reputational consequences, yet this can happen if the susceptible electronic equipment is exposed to even the briefest of power failures or disturbances. So a UPS’s highest priority is to secure near-perfect power availability for itself, the power infrastructure and its supported load.
But it’s not the only priority. Although no one knows exactly what will happen to energy prices, they are high, with many factors liable to drive them higher in the future. According to a Telegraph report, figures from the National Grid at the beginning of 2017 suggested that there would not be enough power plants operating in the UK to keep the lights on for most of December, January and February. The supply gap emerged because a series of old, polluting power stations had been shut down, with insufficient new plants being built to replace them. This is the first time since the published data system began in 2001 that a shortfall rather than a surplus of spare UK power plants has been forecast.
National Grid can and will solve these problems by bringing in backup plants and by buying in more power from the Continent, but these measures are likely to push prices up. Other factors could also increase demand: we could, for example, have a cold winter, compared with the average conditions assumed in the forecast. The forecasts also assume a UK wind generation of over 3 GW, while wind farms produce almost nothing on a still day. Therefore, levels of wind power cannot be assumed as available when needed.
All of this means that facility managers are seeking UPSs that offer the most efficient possible operation along with the highest availability that can be practically achieved. Fortunately, evolution in UPS technology, particularly modular solutions, enables implementations that can satisfy both these objectives.
Modular UPS Systems
The most important step in this evolution has been the transition to transformerless technology. This above all has made modular UPS topology possible, simply due to the substantial reductions in size and weight achieved, not only by eliminating the transformer itself but also by removing the need for the 12-pulse rectifier previously required to improve input THDi performance. These reductions allow the implementation of a complete UPS as a transportable rack-mounting module rather than a monolithic floor-standing installation.
Accordingly, a modern modular UPS system comprises a number of modules mounted into a 19” racking unit. With a single module typically offering 100 kW capacity, large-scale installations for major data centres can be configured; a rack can accept up to five modules to offer vertical scalability to 500 kW, with further, horizontal scalability for example to 3 MW achieved by paralleling up to six racks, as shown in Fig.1 below.
While catering for installations of all sizes in this way, modular technology has several aspects that contribute to the very high power availability that it can provide. As we shall see later, it can deliver this while maintaining uncompromised levels of energy efficiency.
Ways to Improve Availability
Availability can be improved both by increasing mean time between failures (MTBF) and by reducing mean time to repair (MTTR). Modularity facilitates both these approaches, by using redundant configurations, by hot swapping and by its decentralised parallel architecture.
Redundancy is achieved easily because of the modules’ small size relative to the load. For example, a 400 kW load can be supported in a N+1 redundant configuration using a single rack populated with five 100 kW modules. The 100 kW of redundant capacity is an efficient deployment of UPS capital equipment compared with the minimum 400 kW redundant capacity that would be needed with a monolithic installation, yet it provides full redundancy. The MTBF and availability of the UPS system are effectively increased because if one module fails, the others can fully support the load until it can be replaced. If more than one module fails, the critical load can be transferred to the incoming mains.
Decentralised Parallel Architecture or DPA is another way to improve availability, as it removes single points of failure from the system. Each UPS module contains all hardware and software required for full system operation, with no shared or common components. In particular, each UPS module has its own static switch as shown in Fig. 2. This is rated for the full load capacity, plus overload, of the UPS module.
As mentioned, availability can also be improved by reducing MTTR – and this becomes possible if the modular system has a ‘hot-swap’ capability. True hot-swap capability in a redundant configuration allows a faulty module to be safely removed from the UPS rack without powering down, transferring to mains or even interrupting power to the critical load. This minimises repair times and correspondingly maximises availability.
Modern modular systems with hot-swap capability, DPA and a redundant configuration can achieve up to 99.9999% availability, sometimes referred to as ‘six nines’ availability.
High Energy Efficiency
The transformerless technology’s underlying modular topology is also energy-efficient. Unlike the earlier transformer-based implementations, transformerless solutions can achieve electrical efficiencies of 96% or better, and maintain these over the entire load spectrum. This reduces not only the direct energy losses in the UPS, but also cuts energy losses associated with cooling. Modern UPS technology’s high input power factor can sometimes additionally reduce energy costs.
As Fig. 3 shows, further savings can be made by running the UPS in Eco Mode, if site conditions make this acceptable. Eco Mode allows UPS efficiency up to 99% and is an energy saving recommendation within the Green Grid’s Data Center Maturity Model. However, it can carry risk. By default, a UPS in Eco Mode is drawing power from the raw mains. Accordingly, Green Grid recommends that users contemplating this mode should consider factors including transfer time between supplies, design ride-through in IT power supplies, availability of surge protection in Eco Mode, quality of internal site power, static transfer switch compatibility, reliability and quality of utility power.
While data centre and ICT facility operators remain under ever-increasing pressure to achieve near-perfect power availability and efficiency, modern UPS technology offers many tools to help them if it is deployed correctly. Availability can be maximised by using hot-swap modular topology in a redundant configuration. Newer technologies have raised UPS energy efficiency, and this efficiency can be further improved by using Eco Mode if site conditions and ICT equipment allow.