Thesis Archive
An Evaluation of Optimal and Near-Optimal Interplant Hydrogen Networks using Process-Graph Studio and Global Sensitivity Analysis
Chupeco, Angelo
Javier, Mona Lyndsay
Kua, Mariel
Abstract:
Hydrogen plays a vital role in the petroleum industry as it is needed in hydrocracking and hydrotreating, procedures in refining crude oil. Optimization of process networks is a key initiative to simultaneously reduce outsourcing fresh materials and maximize the utilization of resources before it is discarded. This research paper zeroes in on developing a methodology to optimize interplant hydrogen distribution systems considering the variations in parameters that occur in real-life settings. The P-Graph theory which was initially created to solve process network synthesis (PNS) problems and designed to optimize processes based on a cost function is utilized to optimize the hydrogen distribution network. Through the understanding and application of P-Graph algorithms and axioms, and the analogy between PNS problems and resource conservation networks (RCN), a novel methodology is proposed. A general structure, or a superstructure, is constructed to consist of all possible connections within a problem and is run to obtain all combinatorially feasible networks, both optimal and near-optimal with respect to the objective function. To further investigate on the structural integrity of these solution structures, Global Sensitivity Analysis (GSA) is conducted to identify which structures are less sensitive to uncertainties that may occur within the system parameters. Three case studies are presented to demonstrate the effectivity of the methodology. Through tabulating the number of occurrences a solution structure appeared, the most structurally sound ones can be identified. Furthermore, sensitivity analysis will show that there are times where the optimal or near-optimal solution structures will be more practical to implement, depending on what parameters one intends to prioritize. The results of these case studies assert that this methodology can generate thousands of optimal and near-optimal solutions and can also eliminate those solutions that have uncertainties, thus allowing the user to come up with structurally sound designs.
Adviser:
Aviso, Kathleen B
Tan, Raymond R