Synthesis and global optimization of process networks involving natural resources

This dissertation focuses on the development of models and global optimization algorithms for the synthesis of industrial process networks involving natural resources. Water, crude oil, and bioethanol are the natural resources considered in this thesis. The dissertation begins with the optimal design of integrated water systems in Chapter 2 that are present in virtually every chemical and petrochemical industry and involve water using and treatment operations; the goal in designing such systems is to make efficient use of freshwater (which is a scarce resource in many parts of the world) and to reduce the cost of wastewater treatment. We propose a superstructure optimization approach for the design of these networks and propose a global optimization algorithm in this chapter for solving the design problem. The synthesis of integrated water networks serves as the underlying theme in this thesis and is considered at different places in the thesis.; In Chapter 3, we present the theory that serves as the backbone for the algorithmic work in this thesis. We develop a global optimization algorithm for solving nonconvex models that have decomposable structures. This algorithm is based on the use of cutting planes. The ideas proposed here are used in later chapters for particular applications. One such application is the design of globally optimal integrated water networks operating under uncertainty that is addressed in Chapter 4. The mathematical programming model corresponding to this system is quite large and complex to solve, but has a decomposable structure that can be exploited for its global optimization. The complexity in the design of the integrated water networks can also be increased by introducing detailed nonlinear models for the treatment units. In this context the design of a wastewater treatment network is considered in Chapter 5. Here we have a membrane based contaminant removal system that also recovers expensive metallic contaminants from a wastewater stream. Rigorous and simplified nonlinear models are formulated for the optimization of this wastewater treatment network where the solution of the simplified model serves a good starting point to solve the rigorous model. For solving the simplified model, the decomposition technique developed in Chapter 3 is used as part of a global optimization algorithm. The last application of the cutting plane approach developed in Chapter 3 is in obtaining the globally optimal schedules for the flow of crude oil at the front-end of a refinery in Chapter 6. The model used to represent the scheduling problem does not have an inherently decomposable structure, and so we use a spatial decomposition of the network to generate cutting planes that are embedded in an outer-approximation based algorithm for obtaining globally optimal schedules.; The final part of thesis deals with the design of energy efficient corn-based ethanol plants. Such plants are increasingly being built with the goal of producing bioethanol as an alternative energy source to gasoline. The focus in Chapter 7 is the development of a limited superstructure embedding different alternatives for the operations in the bioethanol plant, and representing its optimization as a nonlinear programming problem. The other important aspect considered in this chapter is the application of heat integration techniques to reduce the energy input in the bioethanol plant.