Computationally efficient and robust models of non-ideal thermodynamics, gas-phase kinetics and heterogeneous catalysis in chemical reactors

Abstract

The objective of this thesis is to develop the analysis, theory and implementation of algorithms to accelerate modeling of complex chemical processes, including non-ideal thermodynamics, gas-phase kinetics and heterogeneous catalysis. Computationally efficient models are written in object-oriented language C++ and implemented in the Cantera framework. Results are reported for packed-bed, catalytic membrane reactors, and high-pressure shock tubes. The models are written to accommodate multi-step, detailed chemical reaction mechanisms, such as for ammonia synthesis and decomposition, methane reforming, and high-pressure combustion. Detailed chemical reaction mechanisms are characterized by greatly disparate time-scales, potentially leading to substantial computational cost. The central theme of this thesis is to develop and implement computationally efficient algorithms.

The modeling applications focus on the catalytic membrane reactors. Two packed-bed reactor models are formulated. One considers high-speed flows, where axial diffusive transport is negligible. This model is mathematically posed as a differential-algebraic initial-value problem (IVP) that is solved using a DAE solver from the Sundials suite. The second formulation considers stream-wise diffusive transport, which is posed as a boundary-value problem (BVP), solved with a hybrid approach. This method combines the rapid convergence of the Newton's method with the robustness of the time-integration. The thesis also considers non-ideal combustion behavior in high-pressure shock tubes. All the models are implemented in the Cantera framework. Additionally, the Cantera software has been extended to include a multi-component Peng--Robinson equation of state with self-consistent thermodynamics.