The Importance Of Demand Response Within The Data Centre


Dr Alastair Martin, founder and chief strategy officer at Flexitricity, discusses the importance of demand response within the data centre, the benefits of getting it right and the consequences of getting it wrong.

For operators of mission critical sites such as data centres, the number one priority will always be service continuity for their customers. Anyone proposing ways to reduce data centre energy bills should be ready to explain, above all, how their ideas affect the security of the local electricity supply.

Dr Alastair Martin - founder and chief strategy officer, Flexitricity.
Dr Alastair Martin – founder and chief strategy officer, Flexitricity.

When a proposal claims to enhance site supply security, yet involves using uninterruptible power supply (UPS) systems to help an external party like National Grid, many data centre managers don’t know which way to jump. This explains the mixed progress demand response has made in this sector.

Demand response means using the capabilities of business energy users to generate revenue by helping to secure electricity transmission and distribution networks. It’s an activity for which many data centres are already well equipped. If done right, it doubles as a best-practice asset test and exercise regime, which is important to the health of critical generators in particular.

Where data centres stand out

Critical power supplies at data centres have one crucial advantage over other sources of customer-side flexibility – speed. For example, a diesel rotary UPS (DRUPS) can act within a second of a failure at a major power station, just as it will when there’s a local power cut. However, data centre owners do not want to carry National Grid through all of the minor issues that occur every day. This places their capability firmly in one category: Frequency response.

When a power station (or one of the cross-channel interconnectors) fails, the frequency of the mains supply falls very quickly. This happens around 10 times each year. When it does, Flexitricity’s critical customer sites immediately take their demand off the mains, and stay off for 30 minutes.

That adds up to around five annual hours of service delivery. But as these are some of National Grid’s toughest hours in the year, and as these sites are among the first to step in, the service is well paid.

Setting up a data centre to deliver frequency response is normally low cost, because most of the necessary capabilities are already there. It’s often the case that a national frequency disturbance would be picked up by the site’s DRUPS anyway, so they have minimal additional capital costs, and frequency response delivery, in its various forms, tends to happen when there is a high chance that the site may be about to switch onto DRUPS power anyway. This means they can participate with little or no impact on day to day operations.

Setting up a data centre to deliver frequency response is normally low cost, because most of the necessary capabilities are already there.

As well as receiving income for participating, there are a number of other benefits. The cost of keeping Britain’s lights on whilst new forms of generation are developed represents a significant burden for industry. Reforms to the energy market mean that business electricity users could face a hike in bills in the coming years of around £75 per MWh during working day winter peaks – roughly doubling the electricity price during these periods. This represents an average of £20,000 per megawatt of average consumption.

So, by turning down energy consumption for short periods when networks are at their most stressed, businesses reduce their exposure to these charges. In addition, all emergency generators – including DRUPS – need occasional on-load running in order to provide reliable emergency power. That works well with demand response – the trick is to ensure that testing is done when the electricity is most needed.

Demand response in context

Demand response is by no means new; we pioneered it over 12 years ago. It has, however, become increasingly important for a number of reasons.

Margins are tight, older coal power stations are coming off the system, new plant build has been delayed and the contribution of varying renewable resources is increasing. This reduces inertia, which in turn increases the risk of system failures and blackouts. This is why National Grid is looking for more flexible capacity from business energy users who can turn systems off and on as required. That’s another reason for critical power users to get involved in grid services – they’re keeping their own lights on.

Our capacity comes from a connected grid of companies, who use our fully automated system to either reduce electricity use or operate small, on site generators, whenever the national electricity system is under stress.

This resource is now recognised as integral to the future of the Grid, as it means we can provide extra capacity and help keep the lights on when demand for electricity is at its greatest.

Getting it right on site

Implementing demand response systems and contracts can take time and consideration.
When first engaging in demand response, customers need both realism and persistence. They also need to spend time finding the right aggregation partner by being ready to ask difficult questions. Track record, experience of working with similar customers, and how the core business operations of participating sites are protected are key considerations.

Data centres are rightly cautious of proposals coming from outside their core area, so as an aggregator we need to spend time getting to know a business and understanding their infrastructure before a proposition can go forward. There is an essential due diligence process for every site.

Anyone proposing ways to reduce data centre energy bills should be ready to explain how their ideas affect the security of the local electricity supply.

As part of this appraisal we need to ensure that the defensive engineering in place to ensure all demand response participation is respectful of the customer’s core business. The customer always retains ultimate control anyway, as we always ensure that every one of our partner sites has a lockout mechanism so that they have the choice not to participate on any particular day, whatever the reason.

We also have to set out exactly how a site is going to measure and respond to mains frequency changes. There’s a lot of confusion in this area which we have to work hard to dispel. All DRUPS systems already measure frequency, so it might seem easiest to simply use that. But proving those internal systems to National Grid’s standards can actually be quite invasive. That’s why we prefer to keep our gear outside the box, leaving the critical DRUPS perimeter undisturbed.

Ultimately, when it comes to delivering megawatts and generating revenue for customers, slow and steady wins the race. By doing the groundwork and getting it right from the start, we can ensure that everyone wins.