Use this URL to cite or link to this record in EThOS:
Title: The value of learning about critical energy system uncertainties
Author: Usher, P. W.
ISNI:       0000 0004 8497 7029
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
In this thesis, a sensitivity analysis is used to systematically classify and rank parametric uncertainties in an energy system optimisation model of the United Kingdom, ETI-ESME. A subset of the most influential uncertainties are then evaluated in a model which investigates the process of resolving uncertainty over time - learning. The learning model identifies strategies and optimal pathways for staged investment in these critical uncertainties. By soft-linking the learning model to an energy system optimisation model, the strategies also take into account the system-wide trade-offs for investment across individual or portfolios of technologies. A global sensitivity analysis method, the Method of Morris, was used to efficiently analyse the model over the full range and combination of input parameter values covering technology costs and efficiencies, resource costs, and technology/infrastructure build-rate and resource-constraints. The results of the global sensitivity analysis show that very few parameters are responsible for the majority of variation in the outputs from the model. These critical uncertainties can be separated into two groups according to their suitability for learning. Some of the important uncertainties identified, such as the price of fossil fuel resources available to the UK, are not amenable to learning and must be managed through risk-based approaches. The parameters which are amenable to learning, the availability of domestic biomass, and the rate at which carbon capture and storage technologies can be deployed, are then investigated using the learning model. The learning model is formulated as a stochastic mixed-integer programme, and gives insights into the dynamic trade-offs between competing learning options within the context of the whole energy system. A UK case study shows that, if the resources are known to be available, total discounted net benefit of the availability of 150TWh/year of domestic biomass is £30bn, while the ability to build CCS plant at a rate of 2GW/year is worth up to £34bn. Together, the value increases non-linearly to a maximum of £59bn. This represents up to 17% of UK's discounted total energy system cost over the next four decades as quantified by the ETI-ESME model. The learning model quantifies the cost threshold below which investment in an uncertain learning project is optimal. The threshold is a proxy for maximum no-regret investment over the aggregate total of research, commercialisation and deployment and could be of use to research funding agencies. The results show that when the likelihood of success of the project is 20%, one-stage learning projects of £10bn or below are always undertaken. For the same likelihood of suc- cess, dividing a project into two-stages more than doubles the investment threshold to £22bn as it allows strategies in which investment switches away from a project if it fails. Dividing a project into multiple stages is particularly beneficial if most of the uncertainty is front-loaded, enabling switching to an alternative. The precise strategy to follow is a complex function of the cost, duration, net benefit and probability of success of each learning project, as well as the interac- tions between the project outcomes.
Supervisor: Strachan, N. ; Keppo, I. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available