The two main purposes of this study are the performance of experimental tests on a pressure regulator to collect data and the model identification of the dynamical behaviour of a pressure regulator. Experimental tests are used to collect the dataset for building a dynamical model. Step changes are introduced by manipulating the position of a downstream valve in the test line. System Identification techniques have been used to build an empirical black-box model from collected data. The behaviour has been modelled with AutoRegressive eXogeneous model structure, with the aim of reproducing the measured response. Different transfer function models have been used, considering both MISO and SISO systems, attaining optimal results. The resulting transfer functions models are then introduced in the process simulator, namely Aspen® HYSYS, to compare the performances and assess model adequacy. The observed response during the dynamical simulations is not representative of the actual dynamic behaviour. The main issue concerns the impossibility of representing the stochastic part of the model in the process simulator. Moreover, closed-loop identification, the influence of upstream dynamics and the choice of non-independent inputs have reduced the possibility of a positive outcome. This suggests a high process-model mismatch as the deterministic part of the model is not sufficient to describe the input/output relation. As future perspectives, "fresh" datasets will be collected by reducing upstream dynamics influence on the system and considering only independent inputs.
Hydrogen blending in natural gas pipeline network: a dynamical model of a pressure regulator
SCHIAVO, ILARIA
2021/2022
Abstract
The two main purposes of this study are the performance of experimental tests on a pressure regulator to collect data and the model identification of the dynamical behaviour of a pressure regulator. Experimental tests are used to collect the dataset for building a dynamical model. Step changes are introduced by manipulating the position of a downstream valve in the test line. System Identification techniques have been used to build an empirical black-box model from collected data. The behaviour has been modelled with AutoRegressive eXogeneous model structure, with the aim of reproducing the measured response. Different transfer function models have been used, considering both MISO and SISO systems, attaining optimal results. The resulting transfer functions models are then introduced in the process simulator, namely Aspen® HYSYS, to compare the performances and assess model adequacy. The observed response during the dynamical simulations is not representative of the actual dynamic behaviour. The main issue concerns the impossibility of representing the stochastic part of the model in the process simulator. Moreover, closed-loop identification, the influence of upstream dynamics and the choice of non-independent inputs have reduced the possibility of a positive outcome. This suggests a high process-model mismatch as the deterministic part of the model is not sufficient to describe the input/output relation. As future perspectives, "fresh" datasets will be collected by reducing upstream dynamics influence on the system and considering only independent inputs.File | Dimensione | Formato | |
---|---|---|---|
Schiavo_Ilaria.pdf
embargo fino al 07/04/2025
Dimensione
31.89 MB
Formato
Adobe PDF
|
31.89 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/28981