Ever-increasing energy consumption and consequent extensive greenhouse gas (GHG)
emissions are two major urgent problems faced by all human beings in the 21st century.
As a major contributor, the energy production section appears to be the most suitable field
where further improvements could be explored to tackle these problems. Polygeneration
is a typical type of next generation energy production technology with higher energy
efficiency and lower/zero GHG emissions. However, methodologies guiding an efficient
and stable transition from our existing energy systems to more advanced ones are still
lacking.
The purpose of this thesis is to provide a generic modelling and optimization framework
to guide planning and design of energy systems. This framework of methodologies ad-
dresses the following issues arising in the planning and designing of energy systems: a)
decision making at both strategic planning level and process design level; b) selection of
roadmaps, technologies, and types of equipment from many available options; c) planning
or design according to both economic and environmental criteria; d) planning or design
under inevitable and unpredictable future uncertainty.
The thesis is organized as follows: first, a review of energy systems is presented, followed
by methodologies of energy systems engineering and their applications. Then a section
of polygeneration process modelling is provided, at both strategic planning and process
design levels, comprising superstructure representations of polygeneration energy systems
at different levels, implementations of the superstructure based modelling strategy using
mixed-integer programming, multi-objective optimization for the optimal process design
according to both economic and environmental criteria, and optimization under uncer-
tainty to account the impacts of future uncertainties at the planning/design stage and to
increase the flexibility and robustness of a process design. Finally, major achievements of
this work are summarised and future research directions are recommended.