Predicting the mechanical strength of thermoplastic composites joined through fusion welding with computer simulation is challenging because the long timescales of entanglement formation are difficult to access with atomistically detailed models. Coarse-grained models provide one route to speeding up the sampling of entanglements, but because they abstract away atomistic details of the modeled molecules extra care must be taken to validate their structural and thermomechanical predictions across relevant state space. This work presents new results using a modeling framework for creating, validating, and using coarse-grained models to study fusion welding processes. We demonstrate our software on case studies of poly(phenylene sulfide) and poly(ether ketone ketone) (PPS and PEKK). We demonstrate coarse models of these two chemistries learned through multi-state iterative Boltzmann inversion. The models are validated against OPLS-UA models, and against experiments through agreement of density, structure, relaxation dynamics, and glass transition temperature. We demonstrate fusion welding process modeling with coarse polymers and conclude with a discussion of tradeoffs between performance and accuracy, and opportunities for future work.