the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
MESSAGEix-Materials v1.0.0: Representation of Material Flows and Stocks in an Integrated Assessment Model
Abstract. Extracting and processing raw materials into products in industry is a substantial source of CO2 emissions, which currently lacks process detail in many integrated assessment models (IAMs). To broaden the space of climate change mitigation options and to include circular economy and material efficiency measures in IAM scenario analysis, we developed MESSAGEix-Materials module representing material flows and stocks within the MESSAGEix-GLOBIOM IAM framework. With the development of MESSAGEix-Materials, we provide a fully open-source model that can assess different industry decarbonization options under various climate targets for the most energy and emissions-intensive industries: Aluminium, iron and steel, cement, and petrochemicals. We illustrate the model’s operation with a baseline and mitigation (2 degrees) scenario setup and validate base year results for 2020 against historical datasets. We also discuss the industry decarbonization pathways and material stocks of the electricity generation technologies resulting from the new model features.
- Preprint
(1632 KB) - Metadata XML
-
Supplement
(706 KB) - BibTeX
- EndNote
Status: open (until 21 May 2024)
-
RC1: 'Comment on egusphere-2023-3035', Anonymous Referee #1, 16 Apr 2024
reply
This manuscript brings significant added value in enriching an integrated assessment model with physical flows and additional technologies, thereby making the model more relevant and pertinent to address the multi-dimensional questions arising from the challenge of addressing climate change. This work attempts to bridge the worlds of IAMs and industrial ecology, which is commendable.
The methodology is sound and the data collection appears to follow the state of the art of the literature. It is a paper that is useful for an IAM modeller and useful for a MESSAGEix modeller specifically, and indeed it serves its purpose as model documentation and validation of a first version of this module, awaiting future modelling improvements. However, it offers relatively few insights to the average reader. The paper's impact could be significantly enhanced if it had the ambition to bring more policy-relevant insights in terms of prioritizing actions to ensure a timely and cost-efficient decarbonization of the materials-industry nexus. This could be done by including more scenarios (for example with different climate policy intensities or other key drivers) and by including more detail on the mix of mitigation options by material sector.
For iron and steel, including DRI with hydrogen and CCS options would immensely increase the relevance of the results. This is being implemented in many IAMs and has become an essential part of steel decarbonization pathways. For example:
using MESSAGEix: https://doi.org/10.1016/j.jclepro.2022.130813
using IMACLIM: https://doi.org/10.1080/14693062.2023.2187750
using POLES: https://doi.org/10.1016/j.egycc.2023.100121For cement, the timing and rate of deployment of CCS is strikingly different from what is presented for chemicals. This requires some explanation.
There is a lot more detail in the petrochemicals sector compared to the other materials, which is also where this paper offers the most innovative work in how industry is represented in IAMs. The chemical industry is the most detailed sector, and most of the mitigation comes from that sector: these two things are inter-related, as the modelling choices drive mitigation potential. This has implications on the mitigation costs and potentials of other industrial sectors as well. More detailed work in each sector could provide options not previously considered and change the narrative on how hard to abate industrial emissions are.
Overall, this manuscript succeeds in making its case about represent more precise and explicit decarbonization pathways thanks to more sectoral detail. However, the mitigation potential appears broadly similar to before the model improvements (-85% in 2070). This might change if some of the questions above are addressed in the modelling (especially CCS coherence between cement and chemicals, and low-carbon steel processes). Interestingly, the mitigation potential in the new modelling is lower even compared to the old modelling in 2050, this could also be discussed in the final section.
Some additional points:
The rationale to choose to represent these materials is not clearly made explicit: is it on the size of direct/indirect CO2 emissions? on the energy use of production? on how their demand is expected to evolve in low-carbon scenarios? The mention of the direct CO2 emissions at the beginning of each sub-section in 2.2 points to this being the deciding factor, but it's not clear. On this topic, why choose to model aluminium instead of some other metal like copper?
The choice to explicitly model materials flows related to power generation technologies needs to be justified more, especially since this sector does not make up for a large share of demand for any of the materials detailed. Starting from a sector that makes up a larger share of materials demand (namely, buildings) would have made more intuitive sense.
The effort to be open source is appreciated -- although I did not see a link to the code and data. As such, the manuscript does not present several of its quantitative assumptions (especially regarding techno-economic parameters for production processes).
More detailed comments follow:
L168-172 description of P1 phase: as made clear only later in the manuscript, this has not been modelled for the materials, so these considerations are not needed here. There is no trade of raw material, so the assumption is that the raw material is present domestically to satisfy the needs for production: at what price?
L186 it's not clear how the scrap levels 1/2/3 are determined: are these qualities of scrap (in terms of purity of material, from complex alloys to pure metal) or classes of ease of recovery (from low to high cost of recovery)? Are they related to end-uses of the material, with particular equipment lifetimes? How are the Power Sector ans Other_EOL steps related to Scrap Recovery 1/2/3 in Figures 2-3-4?
L212, L232 refer to the relevant section here for clarity (2.3.2)
L271 surely there is a better source than Statista
L296 What are the drivers and cost components that are included for trade? (same question for aluminium, chemicals and N-fertilizer/ammonia)
L314 does this mean energy demand for refining is not included at all in MESSAGEix? Including this step to expand the model coverage does not seem to be overly complex.
L318 why were two technologies singled out? how do they differ?
L339 "There are" L341 "we also model": in some cases it is not clear whether the authors are describing real-world operations, or how they plan to represent them in an enhanced version of this model, or how they have actually modelled it for this work.
L364 toluene and mixed xylenes are not mentioned again in the manuscript, why is it necessary to list them here?
L371 what is the source and data for these rations and shares?
L399 "one of the renewable production options": are there other options?
L417 please provide more detail on how ammonia-as-a-fuel demand is calculated in the model
L423 how do the "hydrogen via electrolysis" and green hydrogen processes differ? Are there multiple hydrogen production pathways as input to ammonia production?
L446 "to allow endogenization": I understand that this is not modelled (yet)? If so, it should be clarified.
L448 please provide more detail on methanol use in "various" fuels and how the fuel/feedstock production technologies differ
L485 Presumably, the complexities of power sector planning and operation are modelled in the main MESSAGEix model, which provides new capacities to be installed to the Materials module. In which case, it would make sense to make reference to this instead of saying that the Materials module takes this into account.
L495 Some clarification would be useful here. Certain demands are exogenous (HVCs), certain are endogenously produced but by other parts of the larger model (oil for transport).
L503 Since you mentioned materials flows and stocks several times in the introduction, I expected some more discussion on the methodological choices to calculate materials demand. By using a single equation of demand per capita, you are making the choice to model annual flows for the total material demand, while also detailing flows and stocks for one of the uses (power capacities). Some further justification of the methods used, discussion of pros and cons, and, possibly, anticipation of future work, would improve the manuscript. Also, please specify if the equations are for the total material demand or for total net of what is calculated endogenously.
L516 The method to derive chemicals demand is not justified. You derive income elasticities from other projections, when these projections themselves used some other methodology that was not described here. It would be more correct to derive income elasticities from historical statistics, or to re-use the same methodology as the source you refer to (IEA).
L559 When discussing Figure 9, only materials demand for the construction phase were discussed. These other flows for operation, maintenance and decommissioning were not discussed. This is missing.
L567+ Some more discussion and critical review of the results would be welcome here. Why the differences in production levels with statistics, were not equations calibrated in order to reproduce these exact numbers? For steel, final energy is below statistics and emissions are above, while it's the other way round for chemicals, is there a reason for this? Final energy for aluminium and cement exactly matches statistics: is this because of rounding or is it by design?
L597 Your high value for hydro seems to stem from Arvesen et al 2018, and in this the high-range is due to the assumption of a plant in a remote location. Perhaps you can adopt a value closer to the majority of the literature. See for instance Chapter 12 in Ashby 2013: https://www.sciencedirect.com/book/9780123859716/materials-and-the-environment The studies referenced do specify which hydropower technology they assume, so you can make an informed choice on what technology or technology mix you want to represent. Arvesen et al 2018 is a reservoir and Kalt et al 2021 has a lower value and specifies that it is for a run-of-river type. (Also, Deetman et al 2021 is not actually listed in the manuscript references, presumably it is this: https://doi.org/10.1016/j.resconrec.2020.105200)
L615 Does your 2 degrees scenario need to reach net-zero CO2 emissions? What is the carbon budget assumption? This sounds rather like a 1.5C scenario.
L627 "may stem": You should be able to explain your modelled findings with certainty.
L637 & Figure 12: The different level in 2020 is surprising to me. By doing modelling enhancements, the model has deviated further from statistics? I would have expected that emissions not covered in the materials sub-sectors to be reported in the other industry sector, in order to have total industry emissions coincide with statistics. Also, given the large difference in 2020 emissions, it doesn't make much sense to compare cumulative emissions unless you harmonize the two series to a common starting level.
L642 From my understanding of earlier sections, the model does not account for the price elasticity in materials demand, i.e. demand is entirely inelastic, whereas you mention here that price inelasticity is not accounted for.
L651 Some introduction is needed here, to transition from the version comparison earlier to the discussion of results within the new version only.
L661 This paragraph would benefit from some discussion on what drives the relative flexibility of sectors' emissions.
L666 For aluminium, a mention of the indirect emissions of power production would flesh out the findings.
L669 As both materials industries and other sector are bundled together in Figure 13 panels b and c, it is impossible to disentangle the effects of the new modelling from what is happening in the other sector. Some results are likely largely driven from other sector (like the increase in electrification). Sectoral results could be included in the supplementary material.
L679 The description of the situation in 2020 calls for a sentence on how this changes in the projected years.
L708 As mentioned, not including CCS or hydrogen pathways for steel significantly reduces the mitigation potential of this sector, and the relevance of these findings.
L713 Where is the additional scrap coming from in the 2C scenario? Does the model use all scrap that is available or is there some scrap that is not used because it is uneconomic?
L722 Why does CCS only emerge after 2050 for cement but is an option from 2030 for ammonia and methanol? In general, the findings of much more CO2 captured from petrochemicals production compared to CO2 captured from cement production is surprising. Is this a finding driven by techno-economic parameters in your modelling or due to different sectoral availability assumptions for CCS?
L726 Figure 16: the increase in electricity share is considerable, what is driving this? Do you model an electric kiln option?
L729 As supply is driven by demand, consider switching the order of the sections by putting demand first.
L748 This finding is quite important, as it rebuts the story that the decarbonization transition will have a rebound effect that will result in more emissions.
L802, L825 No need to mention "will" here.
L845 What biophysical limitations on material demand are you envisaging to consider?
Citation: https://doi.org/10.5194/egusphere-2023-3035-RC1
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
214 | 133 | 18 | 365 | 29 | 5 | 6 |
- HTML: 214
- PDF: 133
- XML: 18
- Total: 365
- Supplement: 29
- BibTeX: 5
- EndNote: 6
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1