The value of spatially explicit energy systems optimisation on urban scale and for industrial decarbonisation
File(s)
Author(s)
Yliruka, Maria Isabella
Type
Thesis
Abstract
The increasing shares of renewable power generation have led to a paradigm shift towards higher temporal and spatial resolution in energy systems modelling. As the generation cost of renewables are location dependent, a trade-off between network reinforcements to access the least-cost sites and increased generation cost at well-connected sites arises. To account for this trade-off in the long-term investment strategy, spatially explicit energy system models are required.
Introducing spatial resolution in a model formulation alters the type and size of the technologies of the cost-optimal solution. However, similar changes in
the investment strategy were observed when including the uncertainty of input parameters. As increasing the level of detail and optimisation under uncertainty both increase the solving time, energy system modellers face a decision on how to prioritise the limited computational resources.
To help guide model development, the first part of this work presents a novel method to quantitatively compare the impact of detail and uncertainty. The
novelty lies in including the level of detail as an additional ’uncertain’ parameter in a global sensitivity analysis, making it therefore possible to determine its qualitative ranking against conventional input parameters. As a case study, the method is applied to a peer-reviewed heat decarbonisation model for the United Kingdom to assess the importance of spatial resolution on urban scale. Spatial resolution is amongst the most decisive parameters overall but its influence
is strongly dependent on the type of output variable. The level of spatial resolution is irrelevant for the total system cost, but an intermediate resolution is
needed to determine the technology mix and only the highest resolution gives an accurate design of the networks. Our results suggest, that neither a highly
detailed deterministic nor a coarse optimisation under uncertainty will allow energy system modellers to determine all output variables accurately.
New network capacities are required for the decarbonisation of energy-intensive industries when shifting from fossil-based energy generation on-site to distributed
renewable power generation across the country. Given the energy demand of industrial processes, the network capacities are expected to be significant. However,
they are commonly neglected when comparing direct and indirect electrification options. Using the decarbonisation of ethylene production via CO2 utilisation
as a case study, the second part of this work therefore explores the need for a systems perspective in the decarbonisation of energy-intensive industry.
Based on experimental data and heat-integrated process models reported in the literature, linear process models for the production of green ethylene from
renewable energy, air-captured CO2 and water are developed first. Seven different combinations of electro- and thermocatalytic subprocesses with at most one
intermediate are identified. Using a techno-economic assessment, the least-cost processes for the direct and indirect electrification of ethylene production are
determined and subsequently incorporated in a spatially-explicit energy system model. The mixed-integer linear optimisation model is formulated at an hourly resolution with a single year time horizon and comprises generation, transmission and end-use technologies. The model therefore captures differences in the operational flexibility and the transmission cost between the direct and indirect electrification approach. As a case study, the cost-optimal supply system for a
production site powered by offshore wind in the United Kingdom is designed.
Incorporating the emerging topic of offshore hydrogen production, the renewable energy can either be delivered to shore via submarine hydrogen pipelines or
conventional high-voltage direct current (HVDC) cables. Contrasting the costoptimal process selection on a process and a systems level, the need for a systems
perspective is assessed. The thermocatalytic conversion of CO2 to ethylene via methanol is the least-cost production process both on a process and a systems level. In the techno-economic assessment, its levelised cost of ethylene (LCOEt) amount to 2920 £/tC2H4 while the direct electroreduction process ranks second with 3210 £/tC2H4 . The ethylene production routes via methane, carbon monoxide or a Fischer-Tropsch process are uncompetitive. The cost of green ethylene production therefore exceeds fossil-based ethylene prices in Europe by at least a factor of three. Neglecting the market value of oxygen, the by-product of the direct electrification process, the cost advantage of the thermocatalytic process increases even further in the system analysis. The allocated system cost for the thermocatalytic process are about half of the cost for the electrocatalytic process. Both systems are dominated by the cost of offshore wind turbines and the air-captured CO2 feedstock, contributing up to 48% and 27% of the total system cost, respectively. In contrast, the cost of the submarine hydrogen pipeline and HVDC cables are negligible. Similarly, the difference in operational flexibility between the electro- and thermocatalytic process is irrelevant for the cost-optimal process selection if the water electrolysers can be operated flexibly instead and low-cost hydrogen storage in salt caverns is available. A systems perspective is therefore only found to be necessary when comparing the onshore and offshore hydrogen production from offshore wind. The reduced costs and losses associated with a submarine hydrogen pipeline save
6% of the total system cost compared to an onshore production system. Overall, the work herein goes to show that spatial resolution is especially relevant to determine the cost and capacity of networks. But as large network investments are often associated with large investments into generation and end-use technologies, the share of the network cost on the total system cost can be comparably small. Computational resources can therefore be saved by initially using a single-node model formulation and refining the results using a spatially-explicit model in a second step.
Introducing spatial resolution in a model formulation alters the type and size of the technologies of the cost-optimal solution. However, similar changes in
the investment strategy were observed when including the uncertainty of input parameters. As increasing the level of detail and optimisation under uncertainty both increase the solving time, energy system modellers face a decision on how to prioritise the limited computational resources.
To help guide model development, the first part of this work presents a novel method to quantitatively compare the impact of detail and uncertainty. The
novelty lies in including the level of detail as an additional ’uncertain’ parameter in a global sensitivity analysis, making it therefore possible to determine its qualitative ranking against conventional input parameters. As a case study, the method is applied to a peer-reviewed heat decarbonisation model for the United Kingdom to assess the importance of spatial resolution on urban scale. Spatial resolution is amongst the most decisive parameters overall but its influence
is strongly dependent on the type of output variable. The level of spatial resolution is irrelevant for the total system cost, but an intermediate resolution is
needed to determine the technology mix and only the highest resolution gives an accurate design of the networks. Our results suggest, that neither a highly
detailed deterministic nor a coarse optimisation under uncertainty will allow energy system modellers to determine all output variables accurately.
New network capacities are required for the decarbonisation of energy-intensive industries when shifting from fossil-based energy generation on-site to distributed
renewable power generation across the country. Given the energy demand of industrial processes, the network capacities are expected to be significant. However,
they are commonly neglected when comparing direct and indirect electrification options. Using the decarbonisation of ethylene production via CO2 utilisation
as a case study, the second part of this work therefore explores the need for a systems perspective in the decarbonisation of energy-intensive industry.
Based on experimental data and heat-integrated process models reported in the literature, linear process models for the production of green ethylene from
renewable energy, air-captured CO2 and water are developed first. Seven different combinations of electro- and thermocatalytic subprocesses with at most one
intermediate are identified. Using a techno-economic assessment, the least-cost processes for the direct and indirect electrification of ethylene production are
determined and subsequently incorporated in a spatially-explicit energy system model. The mixed-integer linear optimisation model is formulated at an hourly resolution with a single year time horizon and comprises generation, transmission and end-use technologies. The model therefore captures differences in the operational flexibility and the transmission cost between the direct and indirect electrification approach. As a case study, the cost-optimal supply system for a
production site powered by offshore wind in the United Kingdom is designed.
Incorporating the emerging topic of offshore hydrogen production, the renewable energy can either be delivered to shore via submarine hydrogen pipelines or
conventional high-voltage direct current (HVDC) cables. Contrasting the costoptimal process selection on a process and a systems level, the need for a systems
perspective is assessed. The thermocatalytic conversion of CO2 to ethylene via methanol is the least-cost production process both on a process and a systems level. In the techno-economic assessment, its levelised cost of ethylene (LCOEt) amount to 2920 £/tC2H4 while the direct electroreduction process ranks second with 3210 £/tC2H4 . The ethylene production routes via methane, carbon monoxide or a Fischer-Tropsch process are uncompetitive. The cost of green ethylene production therefore exceeds fossil-based ethylene prices in Europe by at least a factor of three. Neglecting the market value of oxygen, the by-product of the direct electrification process, the cost advantage of the thermocatalytic process increases even further in the system analysis. The allocated system cost for the thermocatalytic process are about half of the cost for the electrocatalytic process. Both systems are dominated by the cost of offshore wind turbines and the air-captured CO2 feedstock, contributing up to 48% and 27% of the total system cost, respectively. In contrast, the cost of the submarine hydrogen pipeline and HVDC cables are negligible. Similarly, the difference in operational flexibility between the electro- and thermocatalytic process is irrelevant for the cost-optimal process selection if the water electrolysers can be operated flexibly instead and low-cost hydrogen storage in salt caverns is available. A systems perspective is therefore only found to be necessary when comparing the onshore and offshore hydrogen production from offshore wind. The reduced costs and losses associated with a submarine hydrogen pipeline save
6% of the total system cost compared to an onshore production system. Overall, the work herein goes to show that spatial resolution is especially relevant to determine the cost and capacity of networks. But as large network investments are often associated with large investments into generation and end-use technologies, the share of the network cost on the total system cost can be comparably small. Computational resources can therefore be saved by initially using a single-node model formulation and refining the results using a spatially-explicit model in a second step.
Version
Open Access
Date Issued
2023-07
Date Awarded
2023-12
Copyright Statement
Creative Commons Attribution NonCommercial Licence
License URL
Advisor
Shah, Nilay
Sponsor
Engineering and Physical Sciences Research Council
Grant Number
EP/R045518/1
Publisher Department
Chemical Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)