What is the closest thing to an modelled open source version of modelling to support this idea in 2023?

Hey gang,

I have a sort of nerd snipey question, and I don’t know whether it could be answered with modelling in PyPSA, to create an update on these figures:

RMI and Carbon Tracker estimate that for $116 billion we could buy and retire every coal plant around the world. That’s a quarter of the cost of a cash-for-clunkers car program Senate Minority Leader Chuck Schumer has floated in the U.S., except for that price we could retire every coal plant globally, drive significantly more emissions reductions, and eliminate the top source of carbon emissions globally.

Source: The World Needs a Cash-for-Coal-Clunkers Program by @WoodMackenzie

These pieces were published a while back, before interest rates raised massively, and before we saw a temporary resurgence in coal.

Any one know is there are any more up to date figures for this otherwise extremely attractive idea?

As no one else has yet bit, I’ll make a couple of peripheral comments.

Formulating a reference scenario, then arbitrarily removing a class of technology — in this case, coal‑powered thermal generation — and rerunning the system should be straightforward for most energy system modeling frameworks (PyPSA included).

Given that both scenarios face the same carbon budget (provided as an external constraint), there may not be much difference in the two trajectories “identified” by that exercise. So the question migrates to one of policy design in relation to desired outcomes. Many if not most commentators would prefer the removal of explicit subsidies and the provision of suitably high carbon prices to drive this phase‑out on the ground — rather than the proposed purchase‑and‑closure scheme advocated by RMI and Carbon Tracker, whether based on willing sellers or mandated and compensated.

This then raises the question of whether public policy feasibilities and the ensuing dynamics — the wider “politics” if you like — can and should be studied with numerical models and perhaps even bundled within general energy system optimization frameworks in due course.

Recent work in the United Kingdom built a first‑stab system dynamics model using three decades of social and political data. The calibrated result is named TEMPEST. More here:

The Italian‑based CMCC ran a one hour webinar earlier this week on this clearly innovative work, now available on YouTube (00:59:34) as OSl2Vs6QHLs for those who want to follow these developments (skip ahead to 00:26:14 for the section by Rachel Freeman).

Notice that I have ducked the core question of whether a “cash‑for‑clunkers” scheme for coal plant is a sensible idea or not. Those kind of dilemmas are very country specific. Germany is completing its multi‑pronged coal phase‑out in the sense of projected dispatch as opposed to installed capacity — while other industrialized countries are building new higher‑efficiency coal plant in the global south under the guise of lowered carbon intensity. HTH, R

Here is the system dynamics causal loop diagram (figure 1) from Freeman and Pye (2022) for reference:

Thanks for this Robbie!

I have a question here:

Notice that I have ducked the core question of whether a “cash‑for‑clunkers” scheme for coal plant is a sensible idea or not. Those kind of dilemmas are very country specific. Germany is completing its multi‑pronged coal phase‑out in the sense of projected dispatch as opposed to installed capacity

My understanding was that the capacity was being replaced by renewables, which individually provide variable power, but in aggregate offer some smoothed out power of a wide enough area, and might have a lower capacity factor - so you’d likely need higher renewable capacity to deliver the same amount of megawatt hours provided by a coal plant with a lower capacity over a given year, simply because you can run the coal plant 24 / 7 if you wanted to, and you can’t do that without some form of storage for renewables.

However, I’m not familiar with the term projected dispatch here - I’ve heard dispatch mainly used in reference to being able to quickly ramp generation up and down in response to conditions on the grid (i.e. a dispatchable gas generator), but it sounds like you’re using it in a to refer to generation a given amount of megawatt hours, more or less inline with demand over a given period of time.

I get that you might also consider that to be “dispatching power in response to conditions on the grid”, just over a different timescale to a peaker spinning up for a few hours.

Is there a glossary or similar resource you’d recommend for understanding when and how to use these terms of art properly? I’ve only picked them up through seeing them used, rather than doing any formal education in the field.

On reflection, the term “projected dispatch” was a poor choice by me. What I was trying to allude to was the fact that some energy system models covering Germany — analyses which, admittedly, predate the 2022–2023 Russian gas crisis — show that the existing coal plant may remain on the system but do not get dispatched after 2030. At some point, the plant owners will elect to decommission those assets, but the models themselves do not normally include that divestment dynamic in the first instance.

This analysis takes place in “model land” of course and bridging to the “real world” is the subject of a webinar tomorrow. And the results I mentioned are not forecasts per se, but scenario‑based projections under given sets of assumptions and values.

And following on, the interplay between technical analysis and policy development offers another area of that puzzle to explore. A participant at a recent Climate Compatible Growth roundtable opined that the preferred or selected planning outcomes from energy system models need to reframed in terms of financial analysis so that investors can more easily take notice of the opportunities present.

Conventional generating units that provide those rapid responses you mention will often participate in secondary markets for those various ancillary services — related to the associated timeframes, ranging from sub‑seconds to hours. But that conventional world is changing markedly as the system transitions from linked synchronous machines to inverter‑based technologies — if you like, from centralized coal plant to distributed technologies such as rooftop photovoltaics. And as that world changes, utility‑scale battery banks, stronger interconnectors, and flexible loads all increase in significance. As you indicate.

Regarding a standardized terminology, I think Wikipedia is a good place to understand domain‑specific terminology and usage. And to mention that there is the Open Energy Ontology project — but that is more aimed at deriving a consistent set of concepts, with the labels themselves being quite secondary.  HTH R.

1 Like