Modeling and prediction of density distribution and microstructure in particleboards from acoustic properties by correlation of non-contact high-resolution pulsed air-coupled ultrasound and X-ray images

Non-destructive density and microstructure quality control testing in particleboards (PBs) is necessary in production lines. A pulsed air-coupled ultrasound (ACU) high-resolution normal transmission system, together with a first wave tracking algorithm, were developed to image amplitude transmission G p and velocity c p distributions at 120 kHz for PBs of specific nominal densities and five particle geometries, which were then correlated to X-ray in-plane density images ρ s. Test PBs with a homogeneous vertical density profile were manufactured in a laboratory environment and conditioned in a standard climate (T = 20 °C, RH = 65%) before the measurements. Continuous trends (R 2 \textgreater 0.97) were obtained by matching the lateral resolution of X-ray images with the ACU sound field radius (σwo=21mm) and by clustering the scatter plots. ρ s â�� c p was described with a three-parameter non-linear model for each particle geometry, allowing for ACU density prediction with 3% uncertainty and PB testing according to EN312. ρ s â�� G p was modeled by calculating ACU coupling gain and by fitting inverse power laws with offset of ρ s and c p to material attenuation, which scaled with particle volume. G p and c p variations with the frequency were examined, showing thickness resonances and scattering attenuation. The combination of ACU and X-ray data enabled successful particle geometry classification. The observed trends were interpreted in terms of multi-scale porosity and grain scattering with finite-difference time-domain simulations, which modeled arbitrarily complex stiffness and density distributions. The proposed method allows for non-contact determination of relations between acoustic properties and in-plane density distribution in plate materials. In future work, commercial PBs with non-uniform vertical density profiles should be investigated. © 2012 Elsevier B.V. All rights reserved.