(672h) Investigating the Identifiability of Microkinetic Model Parameters with Transient Inputs

Microkinetic models are useful tools for catalyst design. When kinetic parameters for each elementary step are identified, rate limiting steps and critical intermediates can be confirmed. Steady-state kinetic studies are successful in parameter identification for traditional kinetic models, such as, Langmuir-Hinshelwood expressions. Identifying parameters in complex microkinetic models accurately with steady-state studies can be difficult due to limited information in available data. Transient kinetic studies providing information on reaction network and surface coverage are more conducive for parameter identification of the microkinetic model.

Therefore, parameter identifiability of a representative microkinetic model through a transient kinetic study is investigated and compared with the parameter identifiability from a steady-state kinetic study. A Langmuir-Hinshelwood mechanism is adopted for the proposed microkinetic model. In the identifiability investigation process, the microkinetic model is identified with synthetic reactor performances generated from the same model with preset kinetic parameters. If deviation between fitted and preset kinetic parameters’ values are small while confidence intervals of fitted parameters are narrow, parameters in the microkinetic model are practically identifiable. For clear evaluation, two criteria using the deviations and the confidence intervals are defined. A pseudo-random binary sequence (PRBS) is applied to the inlet concentration to induce transient behavior in a kinetic study for better practical identifiability. An appropriately tuned transient kinetic study outperforms a steady-state kinetic study in parameter identification of the microkinetic model. Parameters are more accurately identified with a PRBS inlet profile than the widely used single pulse. A good practical parameter identifiability is reconciled utilizing a PRBS inlet. The influence of rate-limiting steps, sampling time, and noise on practical parameter identifiability are also investigated. Collectively, this work demonstrates the potential of PRBS transient kinetics studies to identify elementary reaction kinetic parameters, and delineates requirements of sampling time and noise needed from experimental studies to provide meaningful data.