Host Institution: The University of Nottingham
Lead Investigator: Michael Tretyakov
Consistent and repeatable manufacturing process is seen by the composite industry to be the key to achieving repeatable mechanical properties of composite components. For the Resin Transfer Moulding (RTM) process, repeatability is achieved not only by the absence of incomplete mouldings, but also by repeatable cycle time or mould filling time. The possibility of getting a dry spot can prevent the RTM process to be used for high-value components. Consistent mould filling time becomes especially important in medium to high volume production (>30,000ppa), where waiting times between the operations should be minimal. Deviations from the designed filling time are the result of uncertainties in material or process parameters, which are almost impossible to eliminate. These deviations can indicate possible dry spots (when the time is longer than designed) or higher void content (when the time is shorter than designed). The injection moulding industry has successfully adopted active control (AC) systems to achieve high-quality, repeatable products. However, AC systems for composite manufacturing are still immature since the RTM process is more complex and involves more uncertainties.
The ideal cases of RTM process with no uncertainties have been considered in depth and are well understood [1, 2]. Recent advancement in the process simulation and uncertainty quantification research (see e.g. [3, 4] and references therein) provided insights into the effects of uncertainties on the RTM process, but the direct numerical simulations of preforms with uncertainties remain computationally expensive. The computational costs are the limiting factor in creating an AC system which can take into account the variabilities in the RTM process and materials. We will investigate the feasibility of novel mathematical and numerical techniques for uncertainty quantification in an AC system, where a very fast response is required. We aim to develop a system which will exploit the information collected during the mould filling to estimate the distribution of permeability online and perform predictions of the mould filling time and dry spot formation. This novel system will not only include uncertainties into the simulations but will also be aimed at reducing them by means of Bayesian inversion, thus improving the accuracy of predictions.
Hence, the objectives of this study are:
• to develop and test a Bayesian inverse algorithm for an online estimation of distribution of permeability using data from an online monitoring system;
• based on the estimated permeability distribution, to estimate the distribution of mould filling time and the likelihood of formation of dry spots;
• to implement and validate an AC system which can control mould filling time and dry spot formation under uncertainty.
The feasibility study will consider macroscopic voids; if this study is selected for a full project then we will consider formation of microscopic voids as well.