Szczegóły publikacji
Opis bibliograficzny
A fast Gaussian process-based method to evaluate carbon deposition during hydrocarbons reforming / Wojciech Koncewicz, Marcin MOŹDZIERZ, Grzegorz BRUS // International Journal of Hydrogen Energy ; ISSN 0360-3199. — 2023 — vol. 48 iss. 31, s. 11666-11679. — Bibliogr. s. 11678-11679, Abstr. — Publikacja dostępna online od: 2021-08-26
Autorzy (3)
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 145957 |
|---|---|
| Data dodania do BaDAP | 2023-04-21 |
| Tekst źródłowy | URL |
| DOI | 10.1016/j.ijhydene.2021.07.213 |
| Rok publikacji | 2023 |
| Typ publikacji | artykuł w czasopiśmie |
| Otwarty dostęp |  |
| Creative Commons | |
| Czasopismo/seria | International Journal of Hydrogen Energy |
Abstract
Biogas, landfill gas, associated petroleum gas, and other tail gases accompanying various industrial processes are potential sources of hydrogen and carbon monoxide for solid oxide fuel cells via the reforming process. As these gases contain heavy hydrocarbons, fine-tuning of steam and carbon dioxide addition and specific temperature control are necessary to avoid carbon deposition during the reforming process. Numerical simulation plays a crucial role in designing miniaturized steam reforming reactors and optimal working conditions. All simulations of reforming processes must account for carbon deposition. The methods commonly seen in the open literature include Gibbs free energy minimization or parametric equations formalism. This paper utilizes Gaussian process regression as a tool for making predictions about which reforming parameters are suitable for carrying out this process without the danger of damaging the catalyst due to a carbon formation. Parametric equations formalism and Gibbs free energy minimization involve either the minimization of objective function or a search for roots of nonlinear functions. These tasks are sensitive to a choice of the starting points of the algorithm — wrong choice of starting points could lead to numerical instability. Unlike conventional methods, the Gaussian process regression approach bypasses the computation of equilibrium composition. It predicts carbon formation tendencies directly from the initial conditions which ensure stability.