Rigorous Dynamic Model-Based Economic Design of the Injected Pultrusion Process with Controllability Considerations

Abstract

The current industrial practice in design of the injected pultrusion (IP) process involves costly trial and error procedures, hitherto unavoidable in the absence of established process models. Even for well understood processes, economic design has traditionally used steady-state physical models, separately followed by control system synthesis using simpler dynamic models. This work describes the development of a rigorous model-based design algorithm for the IP process by integration of economics, operability, and environmental considerations. In our previous work, a 3D dynamic physical model of the IP process was experimentally validated on a bench-scale setup. Using this validated model, we can now reduce the time and expenses associated with a purely experimental approach. An understanding of processing economics leads to throughput rate maximization as the desired economic objective. We develop a multiobjective optimization algorithm that maximizes this objective function, using a response surface methodology with iterative meshing to determine optimal equipment specifications (die dimensions, puller ratings) and processing condition parameters (heating zone temperatures, resin injection pressure). Retrofit design of the bench-scale setup promised a tenfold increase in throughput rate. Calculation of theoretical limit on throughput rates from intrinsic resin cure kinetics supported this finding. Various metrics were proposed to evaluate quality of design in a distributed-parameter sense. Sensitivity studies verified that the economic optimum is also robust to a wide range of expected process disturbances. An analysis of process instrumentation revealed the need for additional measurements. We propose resin flow-rate and cutting force measurements for void and cure sensing respectively. This research is a first step to the integration of design and control for a complex multiphase process using a single dynamic physical model. The integration offers several benefits over current design practices, and highlights the potential for increased productivity in polymer composites manufacturing.