Thread by @nworbmot

🚨 New 24/7 report for Europe! 🚨

Companies moving from *annual* to *hourly* clean power matching get:

- lower emissions for them *and* system

- reduced backup needs for system

- only small cost premium for 90-95% hourly matching

- stimulation of new tech for final 5-10%

🧵

Today we published a report on the first results from an ongoing project between @TUBerlin and @Google:

doi.org/10.5281/zenodo…

Sign up for our webinar on the report in two weeks on 25th October at 2pm:

tu-berlin.zoom.us/webinar/regist…

Easy reading blog post:

blog.google/around-the-glo…

Below: some more technical details 🤓

What's the issue?

Many companies match their electricity demand with 100% renewable power, often wind and solar. Great!

BUT They only match demand on an annual basis, not necessarily in same grid, and sometimes use credits from old existing plants.

E.g. this can happen:

Here a company has matched demand (grey) with wind (blue) over two weeks.

The problem?

When wind power is low, the company still relies on fossil-fuel power from the grid.

=> Emissions & higher costs for the grid.

(But at least demand and generation are in same bidding zone.)

This example shows just how hard hourly matching is if you rely on existing tech like solar, wind and lithium ion batteries alone.

Bridging these multi-day periods of low sun and wind, the dreaded Dunkelflauten, needs something more.

So how to fix this?

@Google has introduced some metrics that measure exactly how much demand is really being met by clean energy on an hourly basis, without counting generation in excess of demand towards the total.

We investigated the impacts of hourly matching for 4 countries in model simulations of the European power system for years 2025 and 2030.

By "we" I mean my colleague Dr. Iegor Riepin and me.

We use our open source model PyPSA-Eur-Sec at bidding-zone level for IE, DK, DE, NL.

We follow the methodology introduced by the Princeton group (@qingyu_xu7, Aneesha Manocha, @NehaSPatankar, @JesseJenkins) for their study of 24/7 impacts in US.

(Spoiler alert: our conclusions align pretty well with that study.)

bit.ly/24-7clean

acee.princeton.edu/24-7/

We modelled 3 types of procurement

- a reference case taking power from the grid
- 100% annual renewables matching (100% RES)
- hourly-matched carbon-free energy for various targets (80-100% CFE)

Here are some results for a company procuring electricity in Ireland in 2025.

The ref case, grid energy, has just 61% carbon-free energy (CFE) on hourly basis.

Matching 100% of annual demand with wind and solar brings this up to 85% (80% from directly procured energy, 5% from grid that happens to be clean).

Higher CFE fractions need dedicated targets.

How does this affect emissions, including power from grid?

100% RES matching significantly reduces direct generation emissions versus the grid average, but still has emissions from fossil-fuelled backup generation.

Hourly CFE targets get us all the way down to zero.

What about the required technology mix?

If we just rely on existing tech like onshore wind, utility-scale PV and lithium ion batteries, we can reach 90-95% CFE without excessive capacity.

Above 95% leads to escalating capacity as we cover those multi-day dark wind lulls.

If we add up all costs and divide by total generation, we can get an average cost of meeting the targets. This can be broken down by component.

Here we see escalating costs above 95% CFE.

[NB: Costs include grid purchases and are netted by revenues from selling to grid.]

That's with *existing* technologies.

If we allow new long-duration storage with low €/kWh cost (represented by hydrogen stored in 2.5 €/kWh underground caverns here), a 100% CFE target costs less than 50% more than an annually matched 100% RES one.

Throw in dispatchable generators like a natural gas Allam cycle plant capturing almost all CO2 for sequestration, or a firm generator like advanced build-anywhere closed loop geothermal or advanced nuclear, then, depending on costs and availability, cost increases drop further.

How about the system impacts if many companies pursue hourly CFE?

If we look at total emissions in Ireland, 10% of companies following 100% CFE procurement leads to lower emissions in the system compared to 100% RES, here by just under 0.2 million tonnes of CO2 per year.

This is due to two effects

- volume effect: higher overall generation, since for 100% CFE it's efficient to over-generate and sell excess to grid

- profile effect: generation at times when the rest of the grid is dirty, thereby displacing dirty generation

In dirty grids the volume effect dominates (e.g. Ireland in 2025), since there's still lots of dirty generation to displace.

But in cleaner grids the profile effect dominates (e.g. Denmark in 2025 and 2030), since adding RES to a high-RES system just crowds out clean generation.

If we look at backup capacity in the rest of the system, we can see that 100% hourly CFE also reduces the need for dispatchable fossil capacity too.

Here for Germany in 2030, we see a reduction in the need for open cycle gas turbines with unabated emissions with 100% CFE.

Finally, one of the biggest benefits of hourly clean procurement:

By pursuing hourly clean energy targets, companies create early markets for technologies the rest of the system will need later (i.e. as it decarbonises towards 2030-5). This will spur innovation and lower costs.

Just as a demand-pull for solar PV stimulated deployment leading to improved manufacturing and cost reductions, so companies pursuing 24/7 carbon-free energy can spur development of long-duration storage and clean dispatchable generators.

(graphic from @Breakthrough)

The project has only just begun and will continue until March 2024.

As well as refining our initial results, we will be including the possibility to shift demand in datacenters in time and between different locations.

Here are the links to the report, webinar on Tue 25th October and blog again:

doi.org/10.5281/zenodo…

tu-berlin.zoom.us/webinar/regist…

blog.google/around-the-glo…

The code is open source and available in the PyPSA modelling framework here:

github.com/PyPSA/247-cfe

zenodo.org/record/7181235

#freethemodels
@openmod

Many thanks to our project partners @Google for all their help and feedback!

@DevonSwezey Hallie Cramer @denvirb @CarolineBGolin @Marc_Oman @jareinhardt1 @markeliotcaine