Multiscale Model-Based Design for Inhibition of Thermal Runaway in Free-Radical Polymerization

Abstract

This dissertation work aims to develop a model-based design framework for the reaction inhibition technique as a promising emergency response measure against thermal runaway in process industry. This is achieved through employing and advancing multiscale modeling techniques that not only account for the complex kinetic mechanisms and flow physics in the stirred tank reactor, but also facilitate the rapid design screening of various failure scenarios to identify sufficient inhibition recipes.

At system level, the dynamic competition between inhibition and polymerization chemistry under non-isothermal condition is investigated. A well-mixed batch reactor model is developed for free-radical solution polymerization of MMA as an illustrative case to investigate three key design parameters of the “short-stopping” mitigation procedure, specifically the injection time, inhibitor quantity, and inhibitor kinetic strength. A general criterion for sufficient “short-stopping” recipe is subsequently identified as to ensure a negative time-derivative of local Semenov number upon inhibitor injection (i.e., dψ/dt|t = tinj < 0), as opposed to the far more conservative criteria in the current practice.

Built upon the system-level dynamic model, a generalized three-dimensional network-of-zones (NoZ) model is developed to capture the non-ideal convective mixing effect and spatial variations during inhibition by dividing the reactor volume into a network of zones with each modeled as a continuous stirred tank reactor (CSTR) and exchange flow rates formulated using an empirical correlation of the impeller design. The NoZ model is able to capture the hot spot evolution during thermal runaway and the local quenching effect of inhibition.

To advance the multiscale modeling techniques, a novel generalized zoning framework is developed for the automatic streamlined construction of CFD-based compartment models through an iterative zoning workflow with detailed characterization of the reduced-order representation of the flow physics. The proposed zoning algorithm not only reduces the computational load by several orders of magnitude, but also significantly lessens the trial-and-error efforts for model development. A test case of two miscible thermal fluids mixing demonstrates that the zoning framework exhibits an accurate representation of the hydrodynamics captured by the CFD simulation and is capable of capturing species transport and heat transfer in turbulent mixing systems with complex geometry.