The aerospace industry increasingly uses fibre-reinforced plastics that are manufactured using automated fibre placement (AFP). A common problem in AFP is the occurrence of gaps and overlaps that affect the structural perfromance of laminates. Tow width fluctuation is an important cause of gaps and overlaps. Two developments in manufacturing are combined to create a new mitigation method for gaps and overlaps. One development is fibre steering, which harnesses the flexibility of AFP to vary the fibre orientation within the ply. Another development is smart manufacturing which includes responding real-time to changes. The combination is local fibre steering which compensates for tow width fluctuation by placing tows adjacent to each other. In three steps it is assesed how effective local fibre steering is in increasing the structural performance of composite laminates by reducing gaps and overlaps through compensating for tow width fluctuation in automated fibre placement. Firstly, it is determined to what extent gaps and overlaps are reduced by constructing a conventional and a steered laminate. The tow width is modelled using a cubic spline. The steered laminate uses numerical approximations of the equations for parallel parametric curves. Two problems occurs that are: the accumulation of curvature leading to self-intersections and cusps, and large angle deviations. The curvature accumulation is tackled by using smoothing splines and the fibre angle deviation can be limited by taking a weighted average. Secondly, the tow width model is validated by measuring the tow width of two different materials. Thirdly, the effect of the reduction in gaps and overlaps and the fibre angle deviation on the structural performance is determined by using a finite element method based on the defect layer method that calculated the buckling load and the effective stiffness. Results show that gaps and overlaps cannot be completely eliminated but local fibre steering is able to increase the effective stiffness.

The versatility of automated composite technologies granted the possibility of manufacturing laminates with a continuous variation of fiber orientation, also known as Variable Stiffness Laminates (VSL). Despite offering enough dexterity to be adapted for composite deposition, industrial manipulators are nontrivial to integrate into other applications than those originally envisioned, especially when complex curvilinear cartesian paths are desired. Further improvements in layup control and performance are necessary to induce an accurate tracking of fiber tow courses. The goals of this thesis were to a) reduce variability of layup control, b) eliminate experimental iterative steps associated with programming tow courses and c) enforce a constant laydown speed that aids consolidation quality. To achieve such objectives, a framework entitled LayLa (Laying Laminates) was developed as an offline programming tool that automatically computes the series of continuous robot motions to perform a tow course at a desired laydown speed. Experiments with LayLa were conducted for Automated Fiber Placement (AFP) using a six degrees-of-freedom KUKA manipulator, without actual deposition. From a VSL design algorithm, two fiber tow courses with a reduced turning radius were selected to be performed at a laydown speed of 0.1m/s. The overall results revealed reasonably accurate trajectory tracking of both joint and operating space variables, with errors at the end-effector position around 0.05mm and at the laydown speed of 3.2mm/s, corroborating the use of LayLa to create external commands for composite deposition.

This master thesis focuses on developing an integrated system capable of performing ultrasonic phased array measurement on composite parts in an automated fashion. This could potentially be used in the manufacturing process of fiber laminate composite parts to obtain more accu- rate detection of manufacturing defects. Phased array ultrasonics makes use of an array of ultrasonic elements that will be used to send consecutive pulses with each of the elements which are received with the remaining elements in the array. By taking into account the signal strength of the pulses between each of the elements in the array and their respective timings a two-dimensional image of the area underneath the probe can be reconstructed into an image. This image contains information about the acoustic material properties through the depth of the measured object. By automating this method of inspection and taking consecutive measure- ments of this type it becomes possible to obtain volumetric data of the measured object. This research aims to investigate the feasibility and potential constraints of integrating this type of sensor system with a robotic arm. The benefit of combining this with a robot means it may be used in a repeated fashion, as well as to inspect a large number of possible geometric shapes. The final part of this thesis work will cover the processing of the data that is produced using this method.Since a large amount of data is produced, processing and analysis is needed in order to be able to visualise the data and to be able to draw meaningful conclusions from it.