These are my notes and speech on the topic of SMR Nuclear - A future energy source for data centres held at the Data Centre World - London on the 7th March 2024.
SMR Nuclear – A future energy source for data centres?
This house believes that nuclear is the only way forward to power this industry moving forward!
I oppose this motion, as a Sustainable ICT consultant, doing the majority of our work in data centres, we are somewhat bemused by the hysteria over the projected energy consumption and expansion plans for the industry over the next 5 years, driven by AI which isn’t really AI but ML or machine learning. Recent news reports suggest that the use of SMRs located on and with data centres and inside their system boundaries is not only desirable but necessary as exisitng grid power will not be sufficent.
There are a number of reasons why this, in my opinion, will never become mainstream, at least not in the UK or in Europe.
The first one is Cost, we simply do not know how much an SMR will cost, and judging by real life nuclear projects, whatever price is set at the beginining will over run considerably by the end, for example Hinkley Point C was granted final approval in 2016 at a cost of £18 Billion and would be generating by 2025, the most recent update in January 24 suggests that the cost will now be in the region of between £31-34 Billion or nearly double and it wont generating electricity until 2029, so 4 years late.
Other similar new-build nuclear projects in Flamanville, France, and Olkiluoto, Finland have been delayed and faced big increases in costs.
The only SMR project that was on the table the CFPP in Utah was cancelled, despite over 10 years of preparatory work and with US Government subsidy.
The second is location, many countries and world cities have “nuclear free” zones this prohibits both nuclear weapons and nuclear power plants, an SMR is a small nuclear power plant, these countries include:
Austraila, Austria, Canada, Italy, & New Zealand
For cities, notably London, Manchester and 198 other UK cities have declared themsleves to be nuclear free.
Even in the US, cities such as Missoula in Montana, Garrett Park and Takoma Park in Maryland, Berkeley, California, Oakland California.
I ask you this, would you be happy to have an SMR nuclear power station in your neighbourhood?
The third is Cost, all nuclear reactors, indeed all power plants have maintenance windows, all fossil fuel power plants and I include nuclear in that description use steam to run turbines, the transfer of the heat from the boilers to the turbines use plumbing and this needs to be checked on a fairly frequent basis, for nuclear this problem is magnified by the need to separate the water used in the reactor away from the water used by the turbines, a heat exchanger is required. Of our current operational nuclear fleet, Heysham 1 and 2 are offline for the inspection of steam valves and are coming back into service today for reactor 1 and next week for reactor 2, the Hartlepool 1 and 2 units were offline for a similar reason for over 6 weeks only coming back into service last week
What this means for DC SMRs is that now you have to have two of them on your site for resilience purposes, adding cost.
The fourth is access to fuels, access to nuclear material is quite rightly, highly regulated, it is not a case of popping down to Tesco and picking up a bunch of fuel rods during your weekly shop, in the UK most fuel deliveries are via the railway system and a lot of planning goes into the re-fueling process and these are planned years in advance. So now, your SMR needs to include a spur to the main railway system, which adds additonal cost to the project.
The fifth is security, we are dealing with a highly regulated material which will require a considerable enhancement of security measures on site, The nuclear police force, the Civil Nuclear constabulary website states “The Civil Nuclear Constabulary (CNC) is a unique armed police force, trained to College of Policing standards. Our core focus is the security of the nation’s licenced civil nuclear sites and civil nuclear material in transit in England and Scotland. Failure to deliver this is simply not an option.
Having visited Dungeness in a past life I can assure you that they take their role very seriously, I was greeted personally at the gate, “Good morning Mr Booth”, Oh good morning sir, how did you know it was me?
Ah he said you didn't see the hidden cameras at the entrance then?
Although I was expected, they had seen my car reg plate upon entering the road 5 miles away from the plant.
So far we have the cost of the construction which will always be significantly higher than first expected, you will have to plan for at least two SMRs and now enhanced security, the costs are mounting.
The sixth is timescales, even the most accelerated SMR project will not be generating electricity until realistically, the mid 2030’s and I cover a little bit more on that in my closing address.
The seventh is regulations, up until quite recently the data centre sector enjoyed a realitively benign regulatory environment, we mostly got caught under legislation designed for power plants hmmm, do you see where I’m going here? Now, of course, our European cousins are facing a raft of environment legislation specifically targeted directly at data centres, the TCDA, CSRD and EED, of which at least 2 of the sessions at this very event have been about and it is highly likely that similar legislation will be introduced in the UK that will closely align.
And we know that DC operators have been whining about these regulations
But, I can tell you that the regulatory regime for nuclear power stations is on another level, in effect you become a utlity, and they are heavily regulated, the whining will reach fever pitch.
So, my penultimate point is cost, have I mentioned the costs?
We have the cost of construction, plus the added bits of infrastructure, we have the costs of having to build not one, but two independent SMRs to cater for maintenance windows, we have the costs for enhanced security and finally the hugely enhanced regulatory regime.
Closing Statement
I’d like to end with some extracts from a UK Parliamentary report I found dated July 23, there were a number of interesting comments, and I quote...
The UK’s current nuclear power capacity is around 5.5–6 gigawatts (GW), enough to generate 15% of our electricity needs. Yet all but one of the UK’s nine currently operational civil nuclear reactors are scheduled to be shut down by 2028. Predicted increases in electricity requirements due the decarbonisation of sectors such as transport, and the commitments of the Net Zero Strategy, require the UK to significantly increase its capacity for low-carbon electricity generation.
It goes on...
We heard in our inquiry that a range of nuclear technologies, currently at various stages of technical readiness, could contribute to the UK’s nuclear capacity targets. In addition to well established gigawatt-scale reactor technologies (of which the European Pressurised Water Reactors (EPR) being built at Hinkley Point C is an example), new technologies such as small and advanced modular reactors (SMRs and AMRs), are being developed.
However, there remains uncertainty on the precise mix of technologies and deployment timescales that will make up the future UK nuclear new build programme.
The National Audit Office report on ‘Decarbonising the power sector’ found that the Government had set “stretching ambitions” for delivering new nuclear power, that would need multiple new reactors to be deployed quicker than has previously been achieved in the UK.
However, our central message to the committee is that ambitious targets must be backed up by equally ambitious strategies to achieve them. We are concerned that the government lacks a comprehensive plan to deliver the promised new generation of nuclear power at the speed and scale required.
On 18 July 2023, the Government made a long-awaited announcement on the role that GBN would take. The announcement included some further details about the SMR selection process that GBN would run.
It said:
There is a process currently underway to identify the best, most appropriate, SMR technologies:
1) market intelligence gathering, which concluded in June 2023
2) technology initial down selection, launched in July, concluding in Autumn 2023 with the next phase to launch as quickly as possible after that
3) successful technologies will be supported to be ready to enable a Final Investment Decision (FID) by 2029. This will entail funding to support technology development and site-specific design; a close partnership with GBN, which will be ready and able to provide developer capability; and support in accessing sites.
Ladies and Gentlemen, the government will not be making its Final Investment Decision until 2029, 2029!
So, I’ve mentioned the cost four times, I’ve mentioned the problems with location, i’ve mentioned access to fuels, security and a highly regulated regime, i put it to you that all of these mean that we are going to have to find another solution, SMR will simply not work in a purely specific data centre environment, but as my colleague will cover in a moment it may be a wider existing nuclear play to supplement the existing nuclear fleet but only on their sites.
Now, as the opposition frequently reminds me, we lost this debate, but it was the last session of a packed programme and held just before the event ended and the audience were all nuclear supporters!
It is clear that the expansion plans of the data centre sector gloabally are driven mostly by the creation of cloud availability zones and the expansion of footprint in countries that do not have a large footprint, I get that, it is also driven by the expanding use of digital products by us, the users and consumers of those digital products, the last growth point is the insatiable needs of Artifical Intelligence (AI) I must admit I'm not sure about the need for AI for everything (which is the way it appears to be marketed, AI this, AI that, AI the other etc). Even things like AI will help the planet and assist in addressing climate goals!!
But, and it is a huge '*BUT*' should we be doing it?, I am sure that the squillions of £/$/€ being thrown around for AI Infrastructure would be better spent on actually adapting and mitigating the effects of climate change.
I'm sure that this will not be the last post I write on this topic, so I'll leave it there for now...until next time.