Sustainability Assessment of Electricity Production and Consumption: DEA Systems Modelling of Complex Internal Structures

Sustainability and sustainable development research have become a very important part in the development of energy policy. This has resulted in growing research interest in the assessment of the sustainable performance of energy systems using methodological tools such as Data Envelopment Analysis (DEA). However, existing sustainability evaluation has tended to either focus on production and environmental impacts at the electricity generation phase or focussed on the use of electricity for some economic or social benefits given the environmental impacts. Assessing the production and use phases separately ignores the interlinkages between energy security, clean energy and energy equity policies, the so-called energy trilemma. This limitation with existing energy research is due to the differences in the electricity generation portfolios of different countries, the complex interrelationships between policy variables and the weaknesses of existing quantitative techniques to handle such complex sustainability problems.To address this shortcoming, this thesis draws on advances in network and non-homogenous DEA and Multi-Attribute Utility Theory to examine the application of DEA in sustainable energy research. European Union (EU) states are used as case studies to demonstrate the applications of these techniques. Specifically, DEA optimisation models are developed to allow for the integration of the generation and consumption phases of the energy system by adopting a systems approach to modelling while recognising the differences in the generation portfolios of the different EU states and allowing for cross-country comparison. Additionally, given the varying theoretical perspectives on sustainability in the literature, some of these divergent views are incorporated in the sustainable portfolio generation and energy consumption evaluations. Consequently, this thesis contributes to existing energy sustainability research by proposing novel approaches for handling non-homogeneity in electricity generation portfolios, developing models to provide holistic decision support and developing models to evaluate the incorporation of sustainability in electricity portfolio mix decisions.These are achieved by addressing three main research objectives. The first objective pertains to the generation phase of the energy system. Specifically, DEA optimisation models are developed to assess sustainability and resource efficiency in electricity production. In the production phase, since units have different portfolios of generation sources, there exist non-homogeneity in the sub-processes of units under investigation. This sub-process non-homogeneity problem has not yet been addressed in the literature. Consequently, two approaches for addressing these non-homogeneities are proposed. The distinction between the two proposed approaches is whether the non-existent sub-processes are included in the model to determine optimal multipliers or not. A comparison of the proposed approaches with a traditional network DEA reveals differences in the overall scores and rankings of units under investigation. This underscores the need to ensure appropriate model correction where there exists such non-homogeneity within sub-processes, especially in the presence of shared inputs/outputs in parallel network DEA. Empirical assessment of resource efficient and sustainable production of the European electricity generation systems revealed average scores of between 0.4 and 0.6, over the study period, without a clear pattern towards performance improvement.The second objective is to develop models for a holistic assessment of the energy system by integrating electricity production and consumption phases. While the production phase is composed of eleven parallel sub-processes, the consumption phase comprises three multi-stage serial sub-systems representing social development, economic development and environmental performance. Therefore, the integrated model is a mixed structure network problem that can provide decision support for both the generation and the consumption phases, as well as provide recommendations for the entire system. In addressing this objective, improvements to existing assessments are made. This includes incorporating electricity as a vital input and employment as a vital output in social development assessment; linking employment as an intermediate factor between social and economic development assessments; and determining environmental performance from emissions and bio-capacity. In the empirical application it was observed that, higher performers in the generation phase are not necessarily higher performers in the consumption phase. While higher performers in the generation phase include Estonia, Poland and Greece, higher performers in the consumption phase include Malta, Luxembourg and Cyprus. An integrated assessment, therefore, provides policy makers with a good understanding of the entire system which may not be evident in the phased assessments.The final objective examines the implication of sustainability in energy portfolio mix planning. Awareness of environmental and social impacts of energy generation has resulted in the incorporation of sustainability as a relevant consideration in energy mix planning decisions using portfolio optimisation. Current approaches for portfolio analyses has tended to add external (social) and environmental costs dimensions to operational (economic) costs to cater for sustainability in energy mix planning. It is shown that attempting to combine the various sustainability associated cost dimensions by adding the respective cost from the various dimensions has the potential to negate the relevance of some of the dimensions, thereby, making some dimensions more important than others. The result is optimal portfolios which may be inconsistent with expectations on sustainability. Relying on Multi-Attribute Utility Theory, the impact of the interactions and other relationships between the various components on technology ranking and optimal portfolios are explored. DEA is employed to examine the impact of the relationship between the cost dimensions on the technology ranking. The mean-variance framework is then used to construct optimal portfolios based on the different dimension relationships. It is found that portfolios constructed using multiplicative pooling of dimensions best conform to the expectations of sustainability while achieving lower emission potential and use of higher renewable energy sources given a cost minimisation objective. For example, additive pooling results in higher rankings for gas and coal powered power plants when compared to hydro, solar and biomass even though gas and coal under-perform renewable sources in two out of the three sustainability dimensions. It is, therefore, recommended that when constructing sustainable portfolio of generation sources, multiplicative pooling of cost dimensions along the sustainability objectives should be preferred.