Composite materials are becoming increasingly important for light-weight solutions in the transport and energy sectors. Reduced structural weight, with improved mechanical performance is essential to achieve aerospace and automotive's sustainability objectives, through reduced fuel-burn, as well as facilitating new technologies such as electric and hydrogen fuels. The nature of fibre reinforced composite materials however makes them highly susceptible to variation during the different stages of their manufacture. This can result in significant reductions in their mechanical performance and design tolerances not being met, reducing their weight saving advantages through requiring "over design". Modelling methods able to simulate the different processes involved in composite manufacture offer a powerful tool to help mitigate these issues early in the design stage. A major challenge in achieving good simulations is to consider the variability, inherent to both the material and the manufacturing processes, so that the statistical spread of possible outcomes is considered rather than a single deterministic result. To achieve this, a probabilistic modelling framework is required, which necessitates rapid numerical tools for modelling each step in the composite manufacturing process. Focussing specifically on textile composites, this project will develop a new bespoke solver, with methods to simulate preform creation, preform deposition and finally, preform compaction, three key steps of the composite manufacturing process. Aided by new and developing processor architectures, this bespoke solver will deliver a uniquely fast, yet accurate simulation capability. The methods developed for each process will be interrogated through systematic probabilistic sensitivity analyses to reduce their complexity while retaining their predictive capability. The aim being to find a balance between predictive capability and run-time efficiency. This will ultimately provide a tool that is numerically efficient enough to run sufficient iterations to capture the significant stochastic variation present in each of the textile composite manufacturing processes, even at large, component scale. The framework will then be applied to industrially relevant problems. Accounting for real-world variability, the tools will be used to optimise the processes for use in design and to further to explore the optimising of manufacturing processes. Close collaboration with the project's industrial partners and access to their demonstrator and production manufacturing data will ensure that the tools created are industry relevant and can be integrated within current design processes to achieve immediate impact. This will enable a step change in manufacturing engineers' ability to reach an acceptable solution with significantly fewer trials, less waste and faster time to market, contributing to the digital revolution that is now taking place in industry.