Emerging catastrophic global challenges have stressed the need for implanting sustainability concept in every aspect of human activities. Chemical and process industry is not an exception to this trend as it is striving towards a more sustainable design and operation aiming at concurrently minimising relevant societal problems and environmental burdens whilst maximising economic benefits. Unfortunately, addressing sustainability problems is rather challenging, as trade-offs naturally arise among an array of conflicting sustainability objectives, thereby making it extremely difficult to select the final solutions to be implemented in practice. These challenges, therefore, call for advanced multi-criteria decision-support tools to assist in the assessment and optimisation of the sustainability level of industrial systems.
In this regard, this thesis aims to become a small part of research community contributing towards the adoption of more sustainable practices in process industry, through the development of systematic mathematical tools that can assist decision-making process in sustainability studies. The key contribution of this thesis is twofold: one involving methodological merits derived from the powerful integration of several mathematical programming techniques; the other covering the practical implementation of the tools developed to a wide variety of applications in chemical and process industry.
In summary, several key technical achievements have been recognised in this thesis. First, the thesis introduces potential implementations of Data Envelopment Analysis – a multi-criteria decision-making tool adopting linear programming to quantitatively measure the relative efficiency of a set of entities – to two case studies which differ in terms of spatial scales, namely products (i.e. screening of CO2 capture solvents) and processes (i.e. the assessment of food waste management options). We found that Data Envelopment Analysis can effectively identify efficient and inefficient entities, and for the latter it can pinpoint sources of inefficiency and establish improvement targets that if attained would make them efficient. Second, a combined approach of mixed-integer programming and Data Envelopment Analysis is proposed to address high dimensionality issues in sustainability studies. The method poses the task of identifying redundant metrics that can be omitted with minimum information loss as a bilevel problem. This approach for dimensionality reduction can greatly simplify the visualisation and interpretation of the Data Envelopment Analysis results in sustainability studies. Third, a tailored Data Envelopment Analysis-based hypervolume algorithm is put forward to perform the temporal sustainability analysis of systems. The approach can (i) identify the best (most efficient) systems; (ii) propose improvement targets for the inefficient ones; and (iii) understand how the performance of a group of systems evolved over time. Lastly, a novel method based on bilevel and multi-objective optimisation (MOO) is proposed to support decision-making in the post-optimal analysis of Pareto fronts. The approach allows to systematically select a final Pareto point to be implemented in practice without the need of controversial weighting scheme.
The capabilities of the multi-criteria decision-support tools developed in this thesis have been extensively explored through a series of case studies in response to the key transformations of TWI2050 that aim at achieving the United Nations Sustainable Development Goals. These successful implementations have demonstrated the feasibility and practicality of the newly developed methods and enlightened the potential synergy of several mathematical programming techniques to support decision-making process in sustainability studies, ultimately guiding policy-makers and engineers to make appropriate and practical decisions in the transition towards a more sustainable society.