Co-authors​ :

 A SIMON (FRANCE), William LUCAS (FRANCE), Florence SAFFAR (FRANCE) 

Ultrasonic testing methods are widely used in Non-Destructive Testing field to find flaws in inspected materials. In particular, they show good efficiency in probing composite materials in depth in opposition to thermal methods. Most of the time, flaws to be detected are induced by use, such as delamination, impact or fatigue cracks. They can be easily detected due to their size and the mechanical impedance mismatch they introduce in the system. However, initial defects appearing during the manufacturing process can already be present in the material. These flaws often take the form of small porosities that agglomerate in high concentration in some of the interfolds of the composite. In particular, these clusters of porosities can appear during thermal welding of thermoplastic materials. Those porosities lead to a reduction in mechanical properties. It is therefore necessary to detect and quantify them before implementation. A classical tool to do so is to use X-ray tomography, but this method is restricted to small samples if high resolution is needed. Recent years have seen the emergence and democratization of the Automated Fiber Placement process to manufacture large-scale systems. X-ray tomography can therefore be difficult to implement. Ultrasonic methods, which have already shown good results in quantifying residual porosities distributed homogeneously in the material, and which can be robotized on the production line, appear to be good candidates to detect such flaws.
The objective of the proposed study is to investigate the sensitivity of the classical ultrasonic imaging method to agglomerated porosities located in specific interfolds. For this work, samples are extracted from a rejected industrial stiffened panel made of thermoplastic materials where flaws have already been located. Also, plates of thermoset material with calibrated porosity distributions are manufactured. The different samples are then scanned in various control configurations using conventional ultrasonic linear phased arrays in an immersion tank. In particular, the influence of the multi-element probe orientation in relation to the samples stacking is investigated. Next, the formed images are compared with X-ray tomography images. On the other side, numerical inspection simulations using CIVA are conducted to take a closer look at the influence of porosity sizes and concentrations on the ultrasonic signal. Finally, the capacity of focused ultrasonic imaging to quantify the shape and depth of porosity clusters is discussed.