Ultrasound testing is currently the only non-destructive inspection method fully accepted and certified for quality assurance in the aerospace industry. Typically it uses through-transmission or pulse-echo techniques, wet processes that are not suitable for pre-cure fibres. In-line NDE for completed components is still dominated by manual techniques, which, at speeds of only 1m2/h, add 10% to manufacturing time. Attempts to automate this have been with gantry systems, such as that developed by Orano (Areva) and in use by Airbus. These inspect large parts with simple geometry but are still slow, with automation only increasing speeds to 3.6m2/h.
Pre-cure inspection is most commonly microscopy or radiography - slow, expensive, highly-skilled manual processes, unsuitable for large areas or automation. Lufthansa Technik and others have focussed R&D in this area on thermography, a faster but low-resolution approach.
Eddy current is widely used for NDT in aluminium aerospace structures, and is popular as a fast and reliable method of detecting surface flaws. Requiring no contact, it is not affected by surface coatings and needs no couplant. However the signal can be hard to interpret in components of complex geometry, and it only works on conductive materials. Carbon fibre is 1000x less conductive than aluminium, so composite eddy current NDT has so far not been possible.
Research has taken place into the potential of magnetic NDT techniques in carbon fibre, but the low conductivity of the material means much of this work has focussed on superconducting quantum interference devices (SQUIDs). Results have been promising, but potential is severely limited due to their requirement for cryogenic cooling. Our innovation is to employ a far more practical and commercial approach to increasing sensor sensitivity.
We compensate for the low conductivity of carbon fibre by using fine-pitch Gallium Arsenide Hall Effect (GAHE) sensors with a dynamic range from nanotesla to tens of Tesla and 1000x more sensitivity than traditional sensors used in eddy current and MFL techniques, the precise sensitivity improvement required. The sensors measure less than 3mm2, and have sensing areas as small as 5µm2, so are ideal for use as linear or 2D arrays. Our novel probe head will take a linear array configuration, combining a transverse detecting field sensor and a longitudinal field detecting sensor. The solid-state sensors are unaffected by dirt and contaminants, giving them the resilience needed to operate reliably in the dusty environment of pre-cure layup. Our proprietary GaAs epitaxial structure has the sensitivity and bandwidth required for faster operation than conventional Hall sensors or coil technology, and the small size makes them ideal for use in robotic automated systems, allowing us to use robotic arms to overcome geometric limitations when scanning 3D components, reducing dead zones, and maximising the high speed potential of this technique. Our project aim is to mount our probe heads on robotic arms and we will demonstrate this potential using Kuka robots for which we have already developed path-planning capabilities including a software toolbox, RoboNDT, capable of supporting through-transmission data, and a software add-on called RoboTeam which supports synchronised teamwork of up to 15 robots.
We have conducted appropriate patent searches across our technical inputs and established that we have freedom to operate. Both Advanced Hall Sensors and TWI will bring considerable know-how into the project, which is currently trade-secret but not yet IP protected. All background IP will remain the property of the originating partners, with non-exclusive licence to partners for use within the project. We will use the Lambert Model D collaboration agreement for new IP arising from the project.