Skip to content Skip to footer

Reducing critical power spend

Electrical Review Logo

Electricity is a top agenda item for datacentre operators. The UK grid continues to demonstrate its inability to self-heal and keep running without incident. There is simply not enough energy generation or storage capacity built-in to the grid. Robin Koffler, MBA CDCEP, Thamesgate director and CDCEP: Certified Data Centre Energy Professional, explains

Energy costs continue to rise and while investments in de-carbonised electricity generation systems appear attractive, solutions can be expensive and not scalable to the levels required to supply the total power demanded even for small datacentre operation sites.

The operational design, construction and characteristics of datacentres are also changing. Scalability and modular responsive strategies are in demand as these reduce initial investments, optimise efficiency and allow timely expansion to meet what can sometimes be instantaneous demand growth.

Given these factors, it is still possible for datacentre operators to find ways to reduce their operating budgets when it comes to critical power protection, without impacting the resilience of their operations. In the short-to-medium term, electricity spends can be reduced by looking for the ‘low hanging fruit’ as a facility-wide exercise. Even small-scale wins from this exercise can lead to incremental improvements the sum of which can be quite surprising. 

De-energising and amalgamating under-utilised systems

There is sometimes a boundary between facilities and IT management teams within datacentres: each has individual remits and key performance indicators (KPIs) to hit but joint energy saving initiatives can yield some interesting results. 

One of the most productive areas to concentrate on is the datacentre’s critical power path, running from the building incomer(s) to server connection and power distribution points. 

The Conservation of Energy Law states that ‘energy can be neither created nor destroyed but can change form’. Heat or noise within the critical power path is a sign of inefficiency and waste power. The actual aim is to review the entire critical power path and identify the loading, heat output, operating efficiency, age and service status of each piece of equipment. 

‘Hot spots’ can be identified using hand-held thermal imagers as part of this energy efficiency audit. Analysis of these areas can reveal further power quality and infrastructure problems including harmonics, overloading and undersized cables. 

Older, less efficient systems can be identified for retirement and/or upgrade to more energy efficient ones. Underutilised systems can also be reviewed, as historical kit tends to be less efficient when under-loaded. The loads themselves could be amalgamated and the number of support systems reduced; thereby saving energy.

Right-sizing with fixed capacity or modular UPS systems

Of course, uninterruptible power supplies are a case in point here. Older UPS systems (transformer-based and early generation transformerless) typically operate at around 85-90% efficiency. They are easily outperformed by modern IGBT-based transformerless UPS systems. These can reach on-line operating efficiencies of 96% (or greater in Eco Mode – 99%). They also have the ability to do this over a far wider load range.

Traditional UPS systems peaked their operational efficiency curves around 80-90% loading. This meant that a UPS rated at 100kVA would be most efficient when run at 80-90kVA. Modern systems have far flatter efficiency curves and the most efficient can run at 96.5% on-line efficiency down to 20% loading. 

If the energy savings and cost-savings look attractive enough to warrant upgrade to the latest UPS technologies, the question that follows is whether to install a mono-block fixed capacity or modular UPS system.

Fixed capacity UPS systems, as the name implies, have fixed capacity ratings. A 500kVA UPS system will deliver 500kVA and most installations would look to load this no more than 80-90% (i.e. 400kVA) to provide an operational safety margin. The UPS could be run at a lower load but the energy losses may be greater dependent on its efficiency curve.

Modular UPS systems allow the datacentre operator to mirror the same infrastructure expansion strategy adopted throughout the installation. As additional server racks become operational, the modular UPS can be expanded vertically through the addition of slot-in UPS modules or horizontally through the addition of another UPS cabinet. 

In terms of reliability, modular UPS and fixed capacity systems are inherently reliable with the modular UPS able to offer higher levels of availability more easily. As a UPS module is slid into a UPS cabinet, it energises and goes through several hundred self-tests before system connection is allowed. Any test failure results in an alarm condition and rejection by the system controller. Once operational, the system controller monitors the overall and module loads and can automatically decide which modules to put into ‘sleep mode’; thereby saving further energy. Though idle, these ‘sleeping’ modules can instantaneously take up load within the datacentres as server workload increases.

The efficiency range of the modular UPS system is exceptional with the best reaching 96.5% from 20-100% load. This really comes into play when operating a modular UPS system with parallel redundancy. 

Most modern UPS (fixed capacity or modular) tend to be Unity Power Factor rated where their kW rating is equal to their kVA. Installing a 120kW system, using three 40kW UPS modules to power a 40-80kW load profile, could see two modules at 50% load (20kW) and one 40kW module in redundancy. System efficiency would be around 96.5% even with the modules running at 50% or less load. The availability would be high (99.9999%).

Upgrades can also be attractive for UPS included on the Energy Technology List (ETL) – Carbon Trust, allowing organisations to benefit from Enhanced Capital Allowances and reduced tax bills. Adopting ‘Critical Power as a Service’ (C-PaaS) can also be an economical and flexible way to upgrade. As well as enjoying a rebate for the removal and disposal of the existing UPS system, the datacentre operator can move onto an operating lease that includes the UPS, batteries, annual maintenance and future battery replacement.

Extending UPS battery life

The batteries within a UPS are the primary part of its energy storage assembly. The installation may feature a standby power generator but the batteries have to be able to deliver all or part of their expected runtime until the generating set takes over.

A UPS battery delivers energy through a chemical interaction process and its performance will degrade over time naturally as a result of the number of charge/discharge cycles to which it is subjected. Connected to a UPS system the batteries will automatically be tested (every 24 hours) by firmware algorithms.

While the UPS may have a one, two or five year warranty, the battery set will normally only be supplied with the battery manufacturer’s warranty (typically one year). Some suppliers extend the battery warranty to three or five years if the battery blocks are fitted with thermal/usage monitors. This is because batteries require a 20-25°C ambient and will degrade with use. 

The batteries in a typical UPS system will require replacement around years 3-4 for a five-year design life battery and years 7-8 for a ten-year design. During these periods individual battery blocks can begin to show signs of early failure, which will normally be picked up during annual inspections, thermal imaging and battery block testing.

Adopting annual battery service visits may seem like an additional cost but most battery testing services will only cost around £200 per 40-piece block set (normal working hours). Using a hand-held tester an engineer can identify under-performing battery blocks before they degrade the battery string. This allows individual replacement or battery string reconfiguration; some UPS can operate from a wide-ranging DC supply, allowing 32-40 battery blocks in a single battery string configuration.

Select an open protocol UPS system

As with any piece of hardware shipped into a datacentre environment, the manufacturer will look to secure additional revenue through after-sales maintenance, service and the supply of consumables and spares. 

In order to guarantee its operation, a UPS system should also be checked annually during a preventative maintenance visit and backed-up with an emergency response plan (4 clock hours, 8 or 12 working hours).

The problem for the datacentre operator is that most UPS are shipped with closed-protocol communications. This locks the buyer into the manufacturer and their service team and parts logistics. Only they have the licensed software and encrypted security required to allow the UPS to be interrogated and parameters set at the service engineer level. This creates a closed club and the buyer may well be locked-in for as long as the equipment is installed.

With an open-protocol UPS, the buyer can choose from a wider range of service partners. They can seek the most competitive bids and pricing for maintenance, service and manufacturer made or recommended consumables.

Manufacturer or independent 

The final decision is whether to work with a manufacturer direct or an independent UPS systems specialist. Both offer merits and often work closely with one another – depending upon the size of UPS installation. This relationship is often reinforced through certified UPS service engineer schemes and approved reseller partnerships. These give the UPS client peace of mind, with the backing of a manufacturer and the consultative expertise of an independent supplier.

{Jathumbnail off}

Top Stories

Stay In The Know

Sign up to receive the Electrical Review Newsletter, the Digital Issue of the Electrical Review Magazine and be the first to hear about the latest events from us.