| dc.description.abstract | The supply of gas and liquids through energy supply chains is closely monitored via measurement stations. Allocating the share of supply from various sources requires understanding the uncertainty in measurements at each measurement station and, more importantly, understanding how uncertainty propagates/correlates in the system as a whole. We use mathematical models to describe properties such as energy to understand the system's supply fully.
An uncertainty analysis applied to those models alone does not offer a comprehensive understanding of the uncertain parameters' impact on the model output. Complementing the uncertainty analysis with sensitivity analysis is essential, as it quantifies the sensitivity of each uncertain parameter to the output.
This thesis highlights the importance of sensitivity analysis by examining a specific case study involving four instances of a synthesized problem, which investigates a system of hydrogen supply chains.
The system setup involves two gas flows, A and B, which originate from two different energy sources and merge in an unspecified process to form a single gas flow, C. The model quantifies the relative uncertainty in the allocated energy of flow A.
The cases differ in the magnitude of parameter uncertainty and flow rate of flow A.
The main objective is to evaluate the increased insights and advantages of performing a sensitivity analysis, comparing the two categories of analyses within the field: local and global. And to explore the methods within the latter category, global sensitivity analysis (GSA), such as the Sobol method, the Fourier amplitude sensitivity test (FAST) and the Delta moment-independent method. Local analysis is performed using the one-at-a-time method.
The findings favour GSA, specifically the Delta moment-independent method, as it provides a more thorough evaluation of uncertain parameters' influence on the output distribution. However, the Sobol method would be preferable if further investigation into parameter interactions is of interest, as it quantifies sensitivity from interactions.
The greatest impact on the output variability comes from the pressures in the source flow B and the combined flow C, for both low-flow rate cases and high-flow rate cases. Also, flow rate of flow B has a significant impact when flow rate A is low, and flow rate A has a significant impact when flow rate A is high. The purity of the hydrogen in gas B and the temperatures from each flow seem to have an insignificant impact on the model output.
The thesis provides insight into the applications of sensitivity analysis on mathematical models as an important tool to assess the impact of the model's uncertain parameters. It also includes a comprehensive review of the choice of method for a given problem. | |
