Category
JFL, Terrace Conference Room (001)
Description
Using computational fluid dynamics (CFD) and statistical methods, the molecular weight distribution (MWD) of a low-density polyethylene (LDPE) autoclave reaction will be reconstructed to predict polymer properties and optimize reaction control. LDPE reactions are highly exothermic; poor control can lead to inefficiency, hot spots, and potentially explosive decomposition. A polymer's MWD significantly influences its properties, and the sensitivity of ethylene polymerization necessitates simulation techniques for safer reactor design. However, traditional polymer reactor software often assumes uniform reactant mixing, limiting its accuracy. CFD can capture spatiotemporal species gradients to create more accurate polymer reactor simulations, but it does not typically reproduce the polymer's MWD explicitly. Instead, the first few moments of the distribution are tracked to determine average properties and approximate the MWD if the shape of the distribution is assumed. While convenient, this approach struggles with multimodal distributions common in industrial polymers. To address this, the polymer MWD can be divided into classes based on degree of branching, assigning each class its own MWD. Using a weighted sum on these classes' MWDs will provide an overall distribution. This method will be validated in a simplified CFD reactor model and applied to a full-scale autoclave reactor, testing 5, 10, 20, and 80 polymer-class simulations. Ultimately, a multimodal MWD qualitatively consistent with plant data and literature will be achieved. This work establishes a plant-scale CFD model that incorporates detailed free-radical polymerization chemistry, offering a tool to optimize LDPE reactions efficiently and improve reactor safety.
Reconstructing a Reactor's Multimodal Polymer Molecular Weight Distribution
JFL, Terrace Conference Room (001)
Using computational fluid dynamics (CFD) and statistical methods, the molecular weight distribution (MWD) of a low-density polyethylene (LDPE) autoclave reaction will be reconstructed to predict polymer properties and optimize reaction control. LDPE reactions are highly exothermic; poor control can lead to inefficiency, hot spots, and potentially explosive decomposition. A polymer's MWD significantly influences its properties, and the sensitivity of ethylene polymerization necessitates simulation techniques for safer reactor design. However, traditional polymer reactor software often assumes uniform reactant mixing, limiting its accuracy. CFD can capture spatiotemporal species gradients to create more accurate polymer reactor simulations, but it does not typically reproduce the polymer's MWD explicitly. Instead, the first few moments of the distribution are tracked to determine average properties and approximate the MWD if the shape of the distribution is assumed. While convenient, this approach struggles with multimodal distributions common in industrial polymers. To address this, the polymer MWD can be divided into classes based on degree of branching, assigning each class its own MWD. Using a weighted sum on these classes' MWDs will provide an overall distribution. This method will be validated in a simplified CFD reactor model and applied to a full-scale autoclave reactor, testing 5, 10, 20, and 80 polymer-class simulations. Ultimately, a multimodal MWD qualitatively consistent with plant data and literature will be achieved. This work establishes a plant-scale CFD model that incorporates detailed free-radical polymerization chemistry, offering a tool to optimize LDPE reactions efficiently and improve reactor safety.
Comments
Doctorate