Reaching net zero: the scourge of the scenario and the myth of the model
Let me first start by saying scenarios are, of course, hugely valuable. They allow us to think about different futures, test different assumptions and plan for unexpected outcomes. So why am I suggesting scenarios are a “thing that causes great trouble or suffering” to quote a definition of scourge?
Because they also allow us to procrastinate and dither; there is always another scenario we could consider, some assumptions we could tweak and the possibility (nay reality) that we haven’t considered every eventuality. Many of us, and I include myself, love models, especially an all-singing and dancing whole system energy model; we can run innumerable scenarios and tweak those assumptions forever. We prevaricate whilst waiting for a better model or improved assumptions and hide behind the outputs; the model says ‘x’, but we need to do more work on ‘y’ to be sure and round we go again.
But any model forecasting 25 years into the future cannot possibly give us an accurate answer; think back 25 years to Y2K and ask yourself if you could have predicted the world of energy today?
We have run out of time to go round in circles with models, we can’t be forever exploring different scenarios and assumptions. Nor can we wait to see which scenario will emerge, because no net-zero scenario will happen without policy to make it happen. Which of course means that policy will be a key determinant of the scenario we end up in. So, what should we be doing?
Critical Thinking
Fundamentally, I believe we should be doing more critical thinking and then testing where necessary using those models and scenarios.
Critical thinking is about making clear and reasoned judgement based on interpreting and understanding the world we live in, the physical constraints on technologies and the risks and opportunities in different pathways.
We need a plan to get to 2050 – a pathway that we have confidence will deliver net zero at a reasonable cost. One that doesn’t rely on non-existent technologies, but which is poised to take advantage of the evolution and cost reductions in those available today, including nascent technologies. We should remember that in the world of energy, which has huge scale, but also long-duration investments, new technologies are slow to reach market maturity. The South Australia example shows what can happen over the 25 year timescale we are interested in to net-zero, but remember that the first commercial PV cells arrived in the 1950’s and Lithium Ion batteries powered the first consumer goods in 1991. Equally, we need to be nimble and quick to change direction, taking advantage of unexpected developments and recognising the cul-de-sacs.
An Example: The Nuclear Quandary
Let’s take nuclear power as an example. I’m going to use nuclear because it’s a technology I have shifted my opinion on as new technologies and understanding has emerged; I strongly suspect it’s an area where politicians need to be rather more nimble. As a young engineer who grew up close to Berkeley and Oldbury nuclear facilities in the UK, I was quite a fan of nuclear. Events like Chernobyl and, much later, Fukushima, made me more circumspect. Around the time of Fukushima, UK renewables were only contributing around 9% of electricity, Offshore wind was £136/MWh, and the idea of a cost effective 100% renewable energy system still seemed implausible to me, I begrudgingly accepted we would need nuclear to decarbonise. But that no longer feels to be the case; in January this year, wind reached 69% of total generation for a short period.
But what about when the wind isn’t blowing people will say. Well critical thinking tells us that nuclear isn’t the answer. Flexible generation is what's needed to balance renewables; nuclear is inflexible. We could doubtless build it to be more flexible, but then it would get even more expensive, both due to the extra engineering required and running at a lower capacity factor. My thinking brain tells me that if we asked it to run like we are now running our gas CCGT plants (35-40% capacity factor), it would cost around double the current unit cost. Exactly how much that is, nobody knows of course. Hinkley C strike price is £137/MWh in today’s money, but it’s nearly 100% over budget, so perhaps the real price is closer to £200/MWh at its design capacity factor. Operating like a CCGT could make that power £400/MWh. The nuclear industry has recognised that nuclear isn’t really compatible with renewables and so have started promoting ‘pink hydrogen’ – nuclear electrolysis – so when the wind is blowing hard, the nuclear fleet switches to hydrogen production. But let’s put our critical thinking hats on. Offshore wind, even with a fair bit of curtailment, will be cheaper than nuclear, so that suggests flexible nuclear will only be generating electricity (rather than hydrogen) for a small proportion of the time. Our whole system models are useful here, because they can tell us roughly what the capacity factor is likely to be – it turns out it could be single digit. So how would we split nuclear costs between hydrogen production and electricity? It’s true electricity will be very valuable when the wind isn’t blowing, but my own simple modelling puts 5% capacity factor hydrogen CCGT at around £350/MWh (using DESNZ plant capex and opex + dedicated wind hydrogen[1]), and our nuclear is going to cost much more than that at those same capacity factors. So that means the hydrogen produced will need to incur a lot of the cost of running a nuclear power plant; perhaps we should call it Gold or Platinum hydrogen at circa £150/MWh or 7.5 times the cost of natural gas today. Who’s buying that?
And of course, the more astute amongst readers will have noticed that the nuclear industry solution to making nuclear flexible could also firm renewables. So the question is whether it costs more to build wind or nuclear – and I think we already know the answer to that.
But what about SMRs (small modular reactors) I hear you cry…well frankly, because of the huge engineering challenges in any new nuclear plant I would say they fit into the category of non-existent technologies today. Even if they were successful and we could roll them out at pace in the late 2030’s and 40’s, what can critical thinking tell us about the risks? To make it cheap we will need to build lots of identical units – that’s the SMR principle. But what if one of them develops a fault and it presents a real safety risk? Presumably, like the French in 2022, we’ll have to shutdown all the reactors with the same design features, i.e. all of those identical SMRs because that’s how we made them cheap. So will we need to back-up our SMRs? Multiple wind units are not susceptible to that problem – even if you identify a blade failure risk, the consequences of such a failure 20+ miles offshore is insignificant and you can manage the risk locally.
None of this means we shouldn’t invest in R&D on SMRs, they may well have a role in some situations, they could be useful post 2050 and there may be more strategic reasons to include nuclear in the mix. It's just that the logical conclusion of the critical thinking is that nuclear shouldn't be a core component of our pathway to net zero in 2050.
Let's keep thinking...
We can think through the impacts of lots of other pathways and technology choice options in a similar ‘critical analysis’ fashion and learn what makes most sense. We can identify the risks and opportunities and then, yes, we should use models to test the impacts of our assumptions, identify features common to different pathways (to help us avoid creating future stranded assets) and estimate the likely magnitude of investments needed, but let's not get hung up on the odd £1billion, it's insignficant, and it'll be wrong.
But not for too long...
We need to make a quick decision on which way we are heading, put policies in place to drive in that direction and be prepared to alter course if something new and unexpected emerges at a pace that will have a significant impact in the time frame of interest.
Finally...
As my Chemistry teacher used to say, Engage Brains everyone and let’s limit the scenarios and stop hiding behind models that will turn out to be wrong.
Energy journalism and writing workshops through writefirstdraft.co.uk
3moBrilliant piece, thank you Rachel.
Gas ¦ hydrogen ¦ electricity ¦ network planning ¦ decarbonisation
4mo“Models don’t tell you the answers, they hey tell you the questions you need to ask.”
Innovative Solutions Architect - Flexibility
5moIf you (not you specifically:) ) are using models to replace thinking then you're using models incorrectly. They are tools to support decision making. When their purpose is understood, used in the right way and interpreted alongside a range of other information they will support making decisions that provide better outcomes (ultimately to consumers). From my perspective, they allow you to challenge the prevailing narrative, allow you to rationalise and prioritise the areas of highest importance and allow you to avoid unintended consequences. When it comes to scenarios, there's a developing thought to "pick one" because more than one creates stagnated decision making. Personally, my view is that sticking to one deterministic (or even worse, forced) scenario is dangerous. I would be on the side of using Monte Carlo, not as a "let's look at all of these scenarios in their own right" but to come to a confident single view of how to move forward with confidence. What I would say is that there are loads of models about and there should be a high bar for making more - that takes too long. Use the ones that exist, use it in the right way, understand what its saying and gather information from other sources to make decisions
Academic research and consultancy on long duration energy storage (particularly flow batteries) and net-zero techno-economics.
5moMark Workman
Academic research and consultancy on long duration energy storage (particularly flow batteries) and net-zero techno-economics.
5moGreat post Rachel! We are definitely guilty of spending huge numbers of hours making whole system models higher resolution, while ignoring uncertainty in our inputs, particularly where that uncertainty can't be approximated with a +/- 10% or gaussian distribution. Scalability and deep exogenous uncertainty fall under this category. On scalability, in your preferred scenario above I don't think it's controverisal to say that the weak point is hydrogen production and storage. Is there enough iridium for PEM scale-up? Could alkali electrolysis work dynamically enough? While vast storage capacity exists in principle in the North Sea, how do we go about piloting this, and how much hydrogen do we have to spend up front to get them running? On deep exogenous uncertainty, will offshore wind still be a viable energy source after 20 years of climate change? Wind turbines are short lived assets so there is little lock-in risk here, but what if we needed to scale up nuclear from scratch? Coming back to the optimisation theme, the UK approach with CfDs has been a bit like a greedy algorithm - it's worked so far, but eventually we might come unstuck by not considering the risks sufficiently!