Thread by @ThomasASpencer

1/n
There has been lots of discussion of IPCC IAM scenarios, particularly focusing on the supply side: eg assumptions on solar PV costs and coal consumption in #RCP85isBollox.
But what do these models and scenarios assume on the demand side? This is just as important.
Thread.

2/n
Firstly, for this I'm using the data so kindly provided here: data.ene.iiasa.ac.at/iamc-1.5c-expl…
Whatever you think about IAMS etc, this level of data provision is really welcome.

I'll focus on baseline and 2C scenarios

3/n
Today I'm going to focus on the issue of end-use transition. What do these models show?

The graph 👇 shows mean levels of H2 penetration in baseline / weak policy scenarios: penetrations in the order of 4-5% by end of the century, evenly spread between sectors.

4/n
👇 shows the penetration of elec in final energy which reaches >40% of world FEC by end of the century. Clearly these models assume a negligible level of transport transition by mid-century under BAU, while industry increasingly electrifies.

5/n
I find this a odd: techs for trans electrification are more advanced than process heat electrification; longer capital cycles in industry versus transport should surely mean that we should see, in a baseline, evidence of electrification in transport first.

6/n
In 2C scenarios the penetration of H2 in final energy consumption increases by only ~3 percentage points, to about 6%. This occurs largely in transport (14% of world FEC by 2100), and scarcely in industry (about 7% of FEC by 2100).
Electricity obliterates H2 in buildings.

7/n
In 2C scenarios, elec reaches 55% of world FEC by 2100, particularly in industry and buildings. Even in 2C scenarios, the elec transition in transport only really kicks in in 2050, and elec still only takes 16% of transport FEC in 2100.

8/n
Let's look at residual ff in these scenarios, starting with oil in transport.

In 2C scenarios, biofuels kick in earlier than elec, and take a larger share of 2100 transport FEC. Oil share only starts to decline around 2040. Still 2.8 Gt unabated CO2 by 2100.

9/n
Industry looks just plain odd. Coal is obliterated. Gas remains to 2100 (petrochemicals??). There is a large and early amount of CCS from industry (20% of energy system sequestration by 2050), and still ~1.1 Gt unabated CO2 from industry by 2100.

10/n
Buildings also look odd. There is still 1.1 Gt of buildings emissions in 2100, and gas still accounts for 7% of FEC, substantially more than hydrogen.
Surely, if we can't do elec heating here and have to do gas, we could do syngas or H2 by end of the century?

11/n
Most of these transitions will not be driven by explicit technology parameters but rather elasticities of substitution between end-use energy carriers.
But we can get a sense by looking at what the models assume for H2 capital costs

12/n
These seem to assume that electrolytic H2 is still more expensive than fossil H2 with CCS by 2040. Average world capital costs of electrolytic H2 are 1500 USD2010 by 2050. Implied learning rate is 4% per year between 2020-2050.

This seems conservative ...

13/n
We can see the effect of these kind of assumptions looking at the time profile of H2 production.

Electrolytic H2 only overtakes fossil H2 with CCS in 2085, and only ramps up from 2040, once its capital cost becomes lower than that of fossil H2 with CCS.

14/n
So to conclude:
1. We need to pay just as much to demand side in scenarios, as to supply. "It's the demand side, stupid", should be our moto.
2. Ins several respects the FEC transition in IAMs seems odd/conservative: big models don't seem to represent FEC transition well.

15/n
3. Many of the things that people question about IAMs (time profile of abatement, reliance on CCS, BECCS, residual fossil) stem as much from what happens on demand as the supply.
4. This is a twitter thread: someone should do a peer reviewed paper on this.