Resumen de Contribution to the development of more sustainable process industries under uncertainty
Over the past decades, the challenges originated as a result of high energy prices and the growing pressure to reduce greenhouse gas emissions have fuelled a large interest in energy and process systems related research. On the one hand, process industries are faced with the need to cover the increasing demand for energy as developing nations grow and developed countries continue to progress in an increasingly uncertain marketplace, and on the other hand, the resources that have traditionally supported this continued progress begin to show environmental impacts that could threaten the sustainable development of species in the world. As a consequence, the present situation could be described as driven along three main edges: energy, sustainability and uncertainty.
Of particular relevance for these problems is research on computer-aided systems technology to develop strategies for investigating the impact of process industries on both, the system efficiency and its life cycle environmental impact. In this sense, Process Systems Engineering (PSE) offers a unique set of tools that are capable of applying traditional engineering and scientific knowledge to systemic problems, thereby enlarging the scope of traditional chemical engineering to larger system scales while allowing the application of robust and systematic tools to more complex systems problems. Hence, the new emphasis on energy and sustainability experienced in the area has been appended to its other more traditional computational and uncertainty related problems.
In this sense, the general goal of this thesis is to explicitly address these challenges by first making a step towards closing the gap between science-based and systems-based research in PSE. This problem is addressed through the integration of techniques and theories from different disciplines into advanced mathematical programming programming frameworks able to deal with both, the environmental and the uncertainty challenges in the design and planning of more sustainable process industries. For this purpose, multi-objective optimization is proposed as the core mathematical programming framework able to represent the effects of these, sometimes, conflicting criteria in the design of process systems.
In particular, a set of multi-objective optimization tools able to deal with both, uncertainty sources and the life cycle environmental impact, are proposed for two problems of process design. Thus, the first half of the thesis is devoted to the design and planning of hydrogen supply chains and the second half to the design of a large-scale complex industrial process plant. The problem of hydrogen infrastructure design has been mentioned in the scientific literature to be of paramount relevance for enabling the development of hydrogen as an energy vector with the potential to drive the transition towards a more sustainable energy system, whereas the problem of whole industrial process optimization has been a traditionally challenging one in PSE particularly due to the computational complexity involved in accurately representing process unit operations. In addition, a strategy for reducing the number of redundant objectives in cases where more than one environmental life cycle assessment (LCA) metrics need to be explored is also presented.
The energy challenge is addressed, first by providing two frameworks for designing hydrogen infrastructures, one capable of mitigating the effects of uncertainty in energy prices and another one able to optimize the economic performance and any life cycle environmental metric defined by standard LCA methodologies, identifying robust and non-redundant more sustainable hydrogen supply chain designs. Hydrogen presents several advantages as an energy vector, that are mainly given by its potential to become environmentally friendy and by its adaptability to the current energy system conditions, where it can play roles ranging from energy storage vector for the electricity system, to being used as a fuel or chemical agent in industrial operations.
With regards to sustainability, LCA has recently emerged as a key element for environmental impact assessment that allows to trace emissions and waste generated in industrial processing activities from ``cradle-to-grave'' in a holistic manner. Therefore, our approach is to append the LCA metrics as additional criteria to be optimized. Combining multi-objective optimization with LCA allows for the automation of the search for process design solutions that can be more environmentally friendy. By proposing a solution based on principal component analysis that can identify redundant environmental metrics, we ensure that all the metrics of interest can be explored, while decision-makers can be aware of particular interactions beteween them, therefore making the problem analysis and decision steps more tractable.
Uncertainty is addressed by formulating stochastic mathematical programming frameworks, that allow to quantify and evaluate the effects on parameter uncertainties at the design step. By appending risk metrics as additional criteria to be optimized, these models are able to represent different attitudes that decision makers can exhibit towards the risk associated to changing conditions. Thus, the mathematical models provided are able to account for the methodological uncertainty associated to classical deterministic frameworks. In addition, appending LCA metrics as additional criteria to be optimized allows to account for the methodological uncertainty associated to traditional frameworks that only dealt with economic performance in a similar way. Furthermore, by focusing on the uncertainty and evaluating the capabilities of the modelling frameworks in different conditions and with different structural characteristics, more robust models can be provided that are able to account for the many sources of uncertainty that usually affect modelling tasks and in turn the associated decision making.
