Model-Based Design of Experiments for Estimating Heat-Transport Parameters in Tubular Reactors

Abstract

Heat-transport parameters in a two-dimensional heat-transport model are estimated from temperature data of a tubular reactor with a fixed catalyst bed. The reactor design is taken from Adler (Chem. Ing. Tech. 72:555–564, 2000). Using model-based design of experiment (DoE), two experimental control variables, the reactor wall temperature and the gas flow density, are optimized to yield minimal parameter uncertainties. Previously in the literature (Bauer, Theoretische und experimentelle Untersuchung zum Wärmetransport in gasdurchströmten Festbettreaktoren, Dissertation, Martin-Luther-Universität, Halle-Wittenberg, 2001; Grah, Entwicklung und Anwendung modularer Software zur Simulation und Parameterschätzung in gaskatalytischen Festbettreaktoren, Dissertation, Martin-Luther-Universität, Halle-Wittenberg, 2004), it was suggested that transient heating of the reactor wall (from ~30 to ~300∘C) yields characteristics in the temperature data that are relevant for estimating heat-transport parameters. It is shown in this work that temperature data from stationary heating at maximum temperature gives much lower parameter uncertainties as when compared to transient heating. This insight allows a significant reduction in the experimental effort. Three to four experiments were previously performed to gather information used to estimate the set of heat-transport parameters for a specific catalyst bed. The number can be reduced down to one experiment when the gas flow density is allowed to change over time. Also, the time for a single experiment can further be reduced when the transient heating period is omitted.