Saturday, August 26, 2017

Addressing Anti-Nuclear Myths: Part 1a

In this post I deal with LCoE (Levelised Cost of Electricity) reports.

As I stated in Part 1 of this series, LCoE reports are used extensively as ‘evidence’ of why a particular electricity generator type is an economically sound investment. It seems there’s an LCoE for each and every type that compares it favourably against all others.

But as we’ll see below, certain generator types are better candidates for accurate LCoE reports that others.

Part 1a: To LCoE or not to LCoE - that is the question


Firstly, every LCoE contains a list of assumptions and constraints. Because we live a dynamic, complex world things are rarely simple or apparent to everyone, and the people preparing the LCoE have to make assumptions to justify the data inputs. In a well-written and properly researched LCoE, these are based on peer-reviewed scientific papers and other real-world verifiable sources, and guesswork is kept to a minimum.

Even then, there’s no guarantee those assumptions will become reality over time. That’s simply the nature of modelling future events and conditions. It gets even more nebulous when you consider the importance of selecting the right peer-reviewed literature to use.

It’s a bit like nailing jelly (jello) to a tree.

Another key component in all LCoE calculations are projections. There are projections made to satisfy almost all aspects of cost calculations, such as fuel, compliance with regulation, maintenance, etc. These should be based on the assumptions and constraints already defined, and often bring rise to ever more assumptions and constraints based on those projections.

It’s no wonder they can get caught up in a constraint loop and lose objectivity.

Then along comes the media and other vested interests who analyse these reports and make further assertions on the effect of their findings on ordinary folk. This results in yet more layers of abstraction, confusion and misdirection.

However, it is possible to make a determination of the veracity of an LCoE based on the nature of the generator being evaluated. Current Renewables (with the possible exception of geothermal) all have one glaring feature that renders any Renewables LCoE unacceptably inaccurate from the get-go - they rely on and are dependent on the WEATHER.

LCoEs need to make projections of the yield of the generator they're evaluating, and this is where Renewables LCoEs inevitably FAIL.

There is no way to predict the weather in any given location with the accuracy needed for costing a Renewables generator plant. Period. As a result, there is no way to accurately predict the yield of electricity from that plant. Local weather conditions are notoriously unpredictable at any single point in time, which means energy yield is subsequently notoriously variable. This includes solar which is at the mercy of something as innocuous as a cloud, regardless of solar plant type.

Of course for every minute of lost yield there needs to be electricity available from another source to handle demand. If that energy source isn’t another Renewables plant with excess available yield, then it has to come from another source. Batteries are touted as a way to guarantee this availability. This is a fatally flawed option because batteries are NOT generators and so need to be charged by a generator themselves. If that generator is intermittent there is no way to predict the availability of sufficient battery charge. So the 'real' backup generator of choice is Gas. Any backup Gas usage profile would be as notoriously variable as the Renewables generator it’s backing up. As there is no way of accurately predetermining this profile, there go your LCoE calculations!

Every time I see a report of a town or city claiming they have or will soon 'go 100% Renewable for their electricity' I cringe. I cringe because I know they really mean they're sourcing their electricity from Renewables generators backed up by a dispatch-able source such as Gas. It's deception at its most damaging, and people need to read the fine print (often not present in media reports or industry spin).

The energy sources with the best operating profiles for LCoEs as accurate as required are the dispatch-able ones: Coal, Gas and Nuclear. Hydro and Geothermal are intermediate in their yield profiles and so are also better than other Renewables in this respect.

Coal, Gas and Nuclear have predictable fuel requirements the cost of which, with the probable exception of Gas right now, are relatively stable. Stable costs are supposed to mean better predictability and so more accurate LCoEs, but even here sometimes they don't.

This is painfully borne out by real-world experience. In the case of nuclear in USA and Europe, another set of assumptions have proven to be as unreliable as the weather. Man-made obstructions such as incompetence, public opinion, political will and market interference have taken their toll on new plant builds.

So even when the technology and yield curves are predictable, there are other assumptions that can derail LCoE reports. So why would you put any faith in an LCoE (or an electricity generator) with the additional uncontrollable variable of weather listed as an operational yield constraint?

Sunday, August 20, 2017

Addressing Anti-Nuclear myths: Part 1

I’ve seen a lot of anti-nuclear sentiment over the years from people on FaceBook, Twitter and other social media forums, and they all seem to follow the same tired by-lines.

So I thought I’d begin a series of posts addressing these myths.

UPDATE:
I have been criticised by people both for and against nuclear energy for the bluntness of these calculations. Please note this is not an attempt at an LCoE (Levelized Cost of Equity/Energy/Electricity - see? Even the acronym is nebulous), but rather a reflection on the kind of assessment most people make when they see headlines like 'Nuclear costs blowout to $25Billion'.

Even though a properly researched and detailed LCoE report is the best way to fully assess the cost of an energy source, people don't immediately consult their nearest LCoE report do they? No, they make a direct basic calculation in their head and conclude that nuclear is too expensive.

This post is designed to show that even at billion dollar levels the basic calculations they make don't necessarily prove nuclear is any more expensive than renewables.

I'm sure hardened critics will immediately Google for an LCoE that matches their basic assessment, and supporters will do the same. In my view, we need to get back to basics, or at least to a set of assumptions we can all agree on. Frankly, I'm sceptical of the accuracy of 'weighted average cost of capital' over a 90 year timeframe. Too many things can happen in 90 years. Even 30 years.

But if this article prompts you to deep-dive into the relative costs of energy sources, all power to you, but take note LCoEs are a dime-a-dozen and I haven't seen one yet that's truly independent, peer reviewed, and without unverifiable long-term assumptions or wide margins of error.

Part 1. Nuclear is too Expensive

Compared to what? Modern nuclear power plants are built to last - their expected lifespans are 80-100 years. Anti-nuclear people triumphantly tout costs like $24 Billion for Hinkeley C in the UK, and more than $25 Billion for Vogtle in the USA. These are very large numbers, true, but let’s break it down.

Hinkeley C’s NET capacity when completed is 3,200MW of electricity [1].
Vogtle’s additional NET capacity when completed is 2,234MW of electricity [2].

Energy yields from these plants is expected to be around 95% of that capacity 24 hours/day, year round and unaffected by weather.
No backup power sources or battery storage required.

Compare that with solar plants Ivanpah CSP and the proposed South Australian CSP plant ‘Aurora’.

Ivanpah’s NET capacity is just 277MW but can only actually produce 20.5% of that capacity [3]. Yet it cost a staggering $2.2 Billion to build.
Aurora’s NET capacity is a mere 150MW. It is expected to produce at 56% capacity, but its sister plant (Crescent Dunes in the USA - 110MW) has only been able to produce 16% of capacity[4]. The price tag for Aurora? $650 Million!

Both these plants are heavily reliant on backup gas generators (NOT included in the costs above) and their yield entirely depends on weather conditions and seasonal fluctuations. As a result they don’t (and can’t) live up to the hype. They also last around 25-30 years[5].

So, let’s do some maths:
There are 24 * 365.25 = 8766 hours per year where electricity is needed.
So to calculate MWh per year generation, we multiply capacity by hours by % capacity factor:
Hinkely C = 3,200 * 8766 * 95% = 26,648,640 MWh per year
Vogtle = 2234 * 8766 * 95% = 18,604,081.8 MWh per year
Ivanpah = 277 * 8766 * 20.5% = 497,777.31 MWh per year
Aurora = 150 * 8766 * 56% = 736,344 MWh per year

As you can see, I’ve taken an optimistic view of capacity factor in the case of Aurora, where we have no actual data.

So, let’s extrapolate that to a per MWh cost, which is $/(MWh per year * lifetime (years)). I’m taking the lifetime of a nuclear plant as 90 years (the median) and solar CSP as the maximum of 30 years:
Hinkley C = 24,000,000,000/(26,648,640 * 90) = $10 per MWh
Vogtle = 25,000,000,000/(18,604,081.8 * 90) = $14.93 per MWh
Ivanpah = 2,200,000,000/497,777.31 * 30) = $147.32 per MWh
Aurora = 650,000,000/(736,344 * 30) = $29.42 per MWh

Also note that the limited hours for actual generation for CSP (daylight hours) means that significant gas is needed to make up that capacity factor, and the costs here for CSP do NOT include that extra cost! Due to the relative small amount of fuel needed for the nuclear plants, for now I’m going to suggest fuel costs between the systems cancel each other out.

And remember, CSP plants will need to be completely replaced 3 times over the nuclear plant lifetimes.

So is nuclear more expensive? Absolutely not. In fact it’s far cheaper.