I fear that my analysis of gas power plants will be less exhaustive than my work on offshore wind. The Companies House data, which I found all-encompassing for the offshore wind industry, is a little more scarce with respect to gas power stations. With my apologies out of the way, I am confident that there is still plenty for me to learn – starting with some insights about the capital costs of historic gas power plants.

I wrote a brief introduction to gas power stations in an earlier post, which should provide some helpful context. In particular it explains the difference between open cycle and combined cycle gas turbines - the former combusts natural gas through a turbine to spin a rotary shaft and generate electricity. The combined cycle version adds an extra step that converts some of the turbine’s hot exhaust gases into electricity, improving its overall efficiency.

I am trying to analyse energy in the UK to help improve policy. None of the content should be construed as investment advice. I have done my best to ensure that the content below is accurate – but I am human and will make mistakes – if you spot any, please let me know and I shall update as appropriate.

Data Sources & Method (In brief)

My starting point was a list of all major power stations in the UK that were operational as of May 2024 – a useful data set, provided by the Department for Energy Security & Net Zero (DESNZ) as part of their DUKES report series.1

The data allows you to identify 35 combined cycle gas turbine power stations (CCGT’s) totalling 30.5 GW. The spreadsheet also includes 31 “single cycle” stations that total 2.4 GW. Not all of these are listed as having natural gas as their primary fuel, and one open cycle gas turbine (OCGT) appears to be incorrectly listed as a CCGT.

To get some data on capital costs, I needed Companies House accounts that were specific to a single power station – which ruled out any sets of accounts covering multiple power plants. Typically, this occurred when a large utility owned several plants or when a number of co-located power stations were held in a single corporate entity – for example a gas power station built on the site of an old coal plant.

I ended up with 13 CCGT’s totalling 11.4 GW and 7 OCGT’s totalling 740 MW. It was then a case of adding up additions to relevant tangible assets and then adjusting them for inflation to 2023 values.

I will add more detail in the appendix, but I would highlight that this work is a sample of a sample – my starting data set exhibits survivorship bias (a list of stations that still existed in May 2024) and then I had to select for companies where I could get cogent data from Companies House.

CCGT Capital Costs:

One of the differences with my work on wind farms is that some of the gas plants tended to have some fairly sizeable ongoing capital additions after construction, perhaps related to turbine maintenance and upgrades. As a result, I had to think carefully about when to draw the cut off for capital additions - I wanted to avoid the logic of Trigger’s broom or the Ship of Theseus.

The most recently commissioned CCGT in my data set was from 2016 and my accounting data ran until the end of 2023, so I had 7 years’ worth of data post the commissioning year. Thus, I decided to define my capital cost as the combined capital additions starting with corporate inception and ending with the 7^th year post commissioning. Commissioning year was taken from the DUKES data set.

Whilst 7 years might be a bit arbitrary, it allowed me to treat each gas power station on a like for like basis and avoid biasing against older projects. Each year’s capital additions were inflation adjusted to 2023 values using UK CPI before being summed together and expressed on a per MW basis.

I’m pleased with the CCGT data – apart from one outlier data point, it looks a consistent picture. The earliest four stations average about £1m per MW, with the rest of the data set (excluding the outlier) averaging a little over £0.7m per MW.

These capital costs are low compared to my work on offshore wind – where recent projects came in at £2m - £3m per MW. To be clear, this doesn’t say anything meaningful about the overall costs of the two different technologies – gas power stations will have much higher variable costs of generation such as fuel and carbon costs.

OCGT Capital Costs:

My data from the seven OCTG’s was much patchier. I managed to find two reasonably sizeable stations (300MW and 140MW) and then five smaller projects that were commissioned in the late 90’s/early 2000’s. The group of five were all under one holding company, though each had their own subsidiary accounts. The five projects appear to use a similar turbine size and are based in relatively urbanised areas. I have shaded these five with vertical lines.

I have used the same approach of adding the capital additions from inception to 7 years post commissioning. This doesn’t apply to the most recently commissioned project on the bottom right of the chart, which only uses 4 years of data post commissioning (but shouldn’t materially bias the data)

The data is slightly unsatisfactory. I had expected OCTG capital costs to come in lower than their more complex CCTG counterparts. Whilst this is certainly true of the two larger OCTG plants which averaged c. £0.45m per MW, the smaller and more urban group averaged around £0.9m per MW.

I think it’s reasonable to conclude that an open cycle gas power station can be built appreciably more cheaply than a combined cycle, though I would like more than two data points to prove this! More OCTG’s are being built in the UK, so hopefully I will get to update this chart in time.

I have combined both the CCGT and OCGT data sets into a single graph below:

Capacity factors and the future

Whilst I was researching capital costs, I came across an interesting data set from DESNZ, that showed the capacity factor of CCGT’s, nuclear and coal fired power stations since the late 1990’s.2

The capacity factor of the CCGT fleet has trended down over time, with a brief resurgence in 2016 associated with lower coal load factors.

Faced with this trend of declining capacity factors, it would make logical sense for anyone considering investing in a new gas plant to give serious consideration to an OCTG over a CCGT, despite their lower efficiencies.

Lower up front capital costs should become more advantageous in a world of lower capacity factors and uncertain duration, as the extra capital costs of a CCGT would be amortised over a declining number of operating hours. It also helps that open cycle gas turbines and reciprocating engines appear to have greater flexibility and lower start costs, which will be advantageous on a grid with more intermittent generation.3

I found it slightly ironic that adding more renewables to the grid might skew the remaining gas fleet towards a less efficient and more carbon intensive generation technology. This isn’t necessarily a problem if the overall result trends towards lower emissions – after all I am writing on a sunny day in March with high solar and wind output!

I’m reminded of a lesson that appears to have cropped up a few times during my 6 months of learning about the UK energy system.

Whether its decarbonisation, costs or energy security - It’s the properties of the collective system that matter and not just each individual technology.

Appendix: Capital Cost Methodology

For each gas power station, I went to the tangible asset disclosures and noted the additions to relevant assets for each year. Relevant is a slightly vague term but more specifically I mean anything that is related to the construction of the power plant. Occasionally this would sit in an intangible asset line item such as “project development costs” but these examples generally converted the balance to property plant and equipment once construction was completed. I excluded any additions relating to decommissioning assets of IFRS 16 leases.

For each power station, I converted every year’s capital additions to 2023 values using UK CPI. To get a total capital cost, I summed up all inflation adjusted annual capital additions from company conception to seven full years after the year of commissioning. The only exception was one of the OCGT plants that was commissioned in 2019 - my accounting data only runs up to the end of 2023, so this data point only included four full years after the year of commissioning.