During the last decades, models have become widely used for supporting a broad range of chemical engineering activities, such as product and process design and development, process monitoring and control, real time optimization of plant operation or supply chain management. Although tremendous advancements continue to take place in the development of numerical techniques and the acceleration of the computing speed, these advancements have been outpaced by the tendency to make rigorous models of much more complicated and extensive systems. Such rigorous models cannot always be effectively used for design and optimisation. A reduction of the model size and complexity is required to make a model-based solution practical. Many current numerical approaches in systems engineering apply order-reduction to a model in its entirety, without preserving the underlying network structure of the process or its multi-scale decomposition. Retaining these meaningful structural features of a process in a reduced model is a necessity for numerous applications. This is the motivation for the research and the results presented in this thesis. The novelty of this thesis is in systematizing and exploiting the essential structural features of a process in model reduction. The model reduction approach aims first at simplifying the physical and the behavioural structure, as well as the systemic level of the chemical process in the model. Only then additional mathematical and numerical (scheme) reductions are selectively applied to individual compartments or units. In the following step, the reduced models of the individual units are connected at system level and the reduced model of the full process is obtained. In this way, the model reduction procedure is able to preserve the essential structural features of the process. Moreover, the physical meaning of the variables and equations is kept as much as possible. The feasibility and the advantages of the approach are presented for two types of applications: (1) the iso-butane alkylation process, an example of a complex process with relatively simple (one-phase) products; and (2) the freezing step in ice cream manufacture, an example of a single process unit with a complex product. The model reduction procedures works well for the cases considered. The resulting models are solved in acceptable amounts of time. Moreover, they are successfully used for applications such as assessment of the plantwide control structures and the dynamic optimization of the plant operation for the iso-butane alkylation process, and the sensitivity analysis of the model’s parameters in the case of the ice cream freezing process. However, the issue of the optimality with respect to the level of the multi-scale decomposition when developing the reduced model is still open.