Is it ‘back to the future’ for data centres – and should they be built in Ireland?
20 September 2016
Marty McFly had his own unique approach to problem solving in the Back to the Future films, but there is, in practical terms, an awful lot of merit in applying what has successfully worked in the past to the somewhat analogous challenges we might face in the future.
Problem solving is by definition: ‘the act of defining a problem; determining the cause of the problem; identifying, prioritising and selecting alternatives for a solution; and implementing a solution.’ What, then, is the problem with data centres?
After all, Ireland, like other regions with a temperate climate and a favourable environment for industrial development, has been building lots of them; they are a mature technology so surely there is no problem to solve?
The problem is electrical energy, particularly, as both the number of data centres and their size rapidly increases. Until there is a radical breakthrough in how computer processors work or the public completely changes its rapidly growing affinity for electronic data, this problem is not going to go away.
So, what are the options? Well the ‘zero option’ has always to be considered: do not build them or have them built elsewhere. With regard to the latter, while Europe has unfortunately been exporting its heavy process industry to other countries with lower energy costs, effective internet commerce is about the speed at which the website or other data loads.
This can be influenced by a number of factors, but the travel distance between the server and user is obviously critical. Europe is a big consumer market for data, so data centres will inevitably be built there.
Should data centres be built in Ireland?
Interestingly, The Irish Times’ technology correspondent Karlin Lillington raised this very issue in a recent article entitled, ‘Data centre growth comes at price for Irish infrastructure’.
Essentially, the gloss has come off the technology race and the fundamental questions are now being asked by the media, not least ‘cui bono?’ – who benefits? Lillington writes: “New centres offer relatively few jobs given the demands they make on Ireland’s power grid.”
In simple terms, the Republic of Ireland’s power demand curve has traditionally fluctuated from a summer valley of circa 2,000 MWe to a winter peak of circa 5,000 MWe. According to EirGrid’s ‘Generation Capacity Statement 2016-2025’, there are about 250 MVA of installed data centres with connection offers for approximately a further 600 MVA.
“At present,” according to the EirGrid statement, “there are enquires for another 1,100 MVA. This possibility of an additional 1,700 MVA of demand is significant in the context of a system with a peak demand in 2014/15 of about 4,700 MWe (where it would add 35%).”
Thankfully, Ireland has an installed dispatchable generation capacity of just over 7,000 MWe plus a 500 MWe interconnector to the UK and considerable intermittent renewable generators. The legacy of the Celtic Tiger years left us with, unlike many of our neighbours, an excess of dispatchable generators.
However, it is not surprising, given the energy requirements of future potential investments, that the article in The Irish Times went on to question the societal cost, not least the billions of euro potentially required for additional infrastructure investment.
Lillington’s piece then concluded: “Perhaps we have enough of them, at least until better and greener energy options are available, and data centres are required to utilise them.”
Clearly, data centres are part of the future landscape and it is therefore a question of managing their footprint as efficiently as possible. Data-centre projects in the past are different to those now planned.
Managing the footprint of data centres
Previously, projects were rolled out rapidly to meet a dynamic market for data services: a high tech warehouse with servers, cooling systems and a limited number of emergency generators. This size of these data centres, of about 20-30 MWe, had a very low ‘regulatory footprint’, essentially limited to a building permit.
As data-centre projects can now be a factor of ten larger in size, the whole dynamics are changing, starting from the position that they cannot be rolled out rapidly, as the regulatory footprint is completely different.
Grid connections required are now high voltage and not medium voltage. For overhead 220 kV lines an Environmental Impact Assessment (EIA) procedure is mandatory, while also in most jurisdictions a requirement for 110 kV lines.
A project to construct a combustion installation with a combined heat output of 300 MWth or more also requires an EIA procedure; that one would never normally actually operate it (i.e. it was for emergency generation) was something the EU legislator never foresaw or specifically prescribed for.
Neither did the legislator consider this point, when prescribing a requirement for an industrial emissions permit for combustion installations with a thermal input of more than 50 MWth.
Regardless of what actual permitting strategy the authorities decide is appropriate, the days of a simple building permit procedure are in the past. Furthermore, the impact of these projects is now so great that cumulative assessment applies. In particular, in the EU since 2004 there has been a requirement for Strategic Environmental Assessment (SEA) of plans and programmes in the field of energy, which lead to future development consent of projects requiring an individual EIA procedure, such as those above.
As United Nations Economic Commission for Europe (UNECE) states with regard to SEA: “SEA facilitates consideration of the environment in relation to fundamental issues (why, where and what form of development) rather than addressing only how an individual project should be developed. The potential for environmental gain is much higher with SEA than with EIA.”
There are also technical reasons as to why ‘business as usual’, but only larger, is reaching its limitations. Instantaneously starting and synchronise a limited number stand-by diesel generators is well established practice, but as the number of generators rapidly increases, so too does the technical challenge.
PM Group has recognised that things would have to change and has completed its own internal design study. With regard to the Irish Times comment, “until better and greener energy options are available”, what is traditionally perceived as ‘green-energy options’ remains a challenge.
For example, intermittent wind generation, emerging storage technologies, novel generation technologies etc have yet to prove themselves on a large scale, while Ireland’s available hydro resources are limited by geography.
However, there is a very viable option with a far-reduced environmental and regulatory footprint, plus economically it is very attractive. After all, Europe has had heavy energy users in the past, so what did they do?
It was the then norm for on-site generation or to co-locate in industrial parks and share utilities. Even for an integrated iron and steel works, where a failure of electrical power would rapidly lead to furnace damage costing hundreds of millions, the necessary on-site reliability could be built in.
Clearly, on-site generation leads to major cost-saving opportunities in distribution and transmission, whose combined costs are some 24% of the electricity price cost breakdown in Ireland, while transmission and distribution losses are over 8% of generation output. If you are smart as to where you build your generation, you do not need a grid with its associated financial and environmental costs.
Design concept for futute data centres
PM Group’s energy and data-centre experts have developed a design concept for future data centres based on a phased roll out of such units and their associated capital costs up to a load of about 360 MWe. This could occur through either a single company investment or an infrastructure park arrangement.
There would be no emergency generators, associated buildings, switchgear and fuel storage. Instead, there would be a phased development of a dedicated power plant, located within a reasonable distance and connected by underground AC cables. This would initially use a number of open-cycle gas turbines and then be reconfigured into combined-cycle gas turbines (CCGTs), the latter utilising gas turbines, heat recovery and steam turbines to provide the most efficient (55%) form of generation with the lowest emissions.
Reliability is ensured by designing enough units to allow for planned and unplanned outages, plus operation at 75% load; if one trips, the others rapidly ramp up.
For sure, some additional capital investment is required over that for standard emergency generators. However, overall operating costs are lower, grid infrastructure requirements are now significantly reduced and such a CCGT is compact and visually unobtrusive.
When both capital and operating expenditure are compared over a reasonable timeframe the CCGT demonstrates merit, as was the case in the past with such on-site generation for large energy users.
Furthermore, given that diesel tanks and generators were never ‘core business’ for internet companies, it is likely that this dedicated power plant would be a specialist design, build and operate arrangement.
While there are so many advantages to going this route, the lack by all parties of a longer-term strategic outlook, has to date left a missed opportunity behind. If the industry sector is to go forward, ‘strategic’ will be key: the public will demand it and the law will require it.
This proposition should be attractive to enterprise data-centre users and/or local authorities with the strategic vision to provide the electrical infrastructure resilience for their data centre at or before ‘facility ready’ and prior to integrated systems testing.
More importantly, it should be seen by society as a sustainable solution for the medium-to-long term, where data centres continue to flourish without having major adverse infrastructural impacts.
Colm O’Mahony BSc CEng MIEI, energy and mission-critical sector director, PM Group, Dublin
Pat Swords BE CEng FIChemE CEnv, principal process and EH&S consultant, PM Group, Dublin