An integrated framework for the computer -aided formulation and part design of engineering rubbers using advanced knowledge engineering and artificial intelligence

Priyan Ramesh Patkar, Purdue University

Abstract

A formulated engineering rubber is a complex mixture involving several components that are involved in the vulcanization reactions. The life-performance of a rubber part is determined in a complex, coupled manner by the vulcanization kinetics, part shape, thermal histories during part manufacture and service, and mechanical loading. Due to this complexity, the current industrial design practice is a heuristic, trial-and-error procedure. Recently, Ghosh, et al. [1] presented a kinetic model to predict the temporal evolution of crosslink density, the key parameter connecting formulation chemistry to mechanical response. However, the model is applicable only to a limited range of formulations and does not describe all features of vulcanization. The primary objective of this thesis is to critically analyze vulcanization chemistry, which involves polysulfidic species, using fundamental quantum-chemical calculations. We have performed exhaustive Density Functional Theory calculations to determine the energies and equilibrium constants for the sulfur-sulfur bond dissociations in the various polysulfidic species. Molecular orbital analyses have been carried out to obtain qualitative insights into relative reactivities of the species. Our results show that the bond dissociation energies and equilibrium constants depend upon polysulfidic chain length, bond position and the type of terminal end group. Our analyses also suggest that some reactions in the Ghosh, et al. model may need to be modified or replaced by alternative mechanisms. Our secondary objective is to predict the life-performance of a formulated rubber part. We have developed a manufacturing model that involves transient heat conduction to describe the spatially heterogeneous temperature field in the part at the end of manufacture, which is then used with the kinetic model to predict the crosslink density profile in the part. We then employ a model by Sarkar and coworkers[2] to calculate the viscous dissipation during part service, and use the resulting temperature field with the kinetic model to describe the long-term mechanical performance of the part expressed as the evolution of the elastic modulus over long times. We also present a preliminary, conceptual framework for simultaneous formulation and part design, which is based on multi-scale, hierarchical forward models and knowledge-driven inverse search. [1] Ghosh, P., S. Katare, P. Patkar, J. M. Caruthers and V. Venkatasubramanian, Rubber Reviews, 2003, 76(3): p.592. [2] Patkar, P., A. Sarkar, S. Syal, J. M. Caruthers and V. Venkatasubramanian, in Fall Technical Meeting of the Rubber Division, American Chemical Society, 2003, Cleveland, OH.

Degree

Ph.D.

Advisors

Venkatasubramanian, Purdue University.

Subject Area

Chemical engineering|Artificial intelligence

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS