In the wood industry, logs are commonly quality screened by discrete X-ray scans on a moving conveyor belt from a few source positions. Typically, the measurements are obtained in a single two-dimensional (2D) plane (a “slice”) by a sequential scanning geometry. The data from each slice alone does not carry sufficient information for a three-dimensional tomographic reconstruction in which biological features of interest in the log are well preserved. In the present work, we propose a learned iterative reconstruction method based on the Learned Primal-Dual neural network, suited for sequential scanning geometries. Our method accumulates information between neighbouring slices, instead of only accounting for single slices during reconstruction. Evaluations were performed by training U-Nets on segmentation of knots (branches), which are crucial features in wood processing. Our quantitative and qualitative evaluations show that with as few as five source positions our method yields reconstructions of logs that are sufficiently accurate to identify biological features like knots (branches), heartwood and sapwood. 

Densification, i.e. the transverse compression of sawn timber has been studied and commercialised for well over 100 years but remains an expensive niche product with low annual production volumes. One reason for this is the reliance on time-consuming batch processes in a hot press. To solve this, a continuous densification process using a belt press, capable of densifying full-sized sideboards was developed. However, there is insufficient knowledge about the effect of knots on the densification outcome. The objective of this study was to assess how different knot parameters affect the densified wood in terms of damage and deformation to the knot itself and the surrounding wood material. Multivariate data analysis methods were applied to a dataset of 171 knots, described by 23 variables. The data showed that it is possible to densify knots in a continuous process without causing damage. Especially sound knots are often unproblematic, even at relatively large sizes, while densifying dead knots often resulted in unacceptable damage to the knot or the surrounding wood. From a material selection standpoint, any knots bleeding into the board edge and dead knots greater than 20 mm in diameter should be avoided altogether.

Thermo-mechanical densification of sawn timber results in improved mechanical properties, but densified wood remains a fairly expensive niche product, partially because of high-cost batch processing. Densification in a continuous process could address this problem and was shown to be possible in previous research. The outcomes were limited to proofs-of-concept, partially due to insufficient cooling capabilities of the used densification equipment, resulting in high spring-back. Therefore, a novel continuous process using a belt press to densify full-sized sawn timber has been conceived. The press ensures almost constant contact between the wood and the heating and cooling elements. The study aimed to analyse how the processing temperature, speed and compression ratio affected the density profile, spring-back, set-recovery, Brinell hardness, and bending properties. Results showed a density increase concentrated close beneath the surface in contact with the heating element and approx. 20% spring-back. Belt speeds above 2 m min−1 caused higher spring-back due to the reduced contact time between the wood and the heating element. Brinell hardness increased by up to 250% at a compression ratio of 27% while bending properties were unaffected. Belt temperature and speed were shown to be critical factors to consider for the future optimisation of the belt-press process.

Wood under thermo-mechanical densification behaves differently depending on the cross-sectional growth ring orientation (GRO) relative to the direction of compression. This influences the degree of cell damage, but also the shape-memory effects occurring when the compression load is released (spring-back) and when the timber is re-moistened (set-recovery). To study how the GRO influences the shape-memory effects, Scots pine specimens were separated into three distinct groups of GRO (Flat, Inclined, Hybrid) and then thermo-mechanically surface-densified. Spring-back and set-recovery were determined by thickness measurements and by digital image correlation. A GRO parallel to the densified surface, resulted in a low spring-back and a high set-recovery which were uniform over the width of the specimen. Specimens with a GRO between 15 and 45° to the densified surface showed high spring-back and low set-recovery, indicating cell-wall damage. Spring-back mainly occurred in the non-plasticised region immediately below the heated surface region and elasto-plastic rolling-shear deformation along individual growth rings occurred. The GRO of softwood subjected to thermo-mechanical densification determines if an applied load results in rolling shear-deformation or radial compression. This in turn determines where in the cross-section and when in the process the cells deform and if this deformation occurs below or above the glass-transition temperature.

The densification, i.e., transverse compression of solid wood can lead to improvements in the mechanical properties, and this can expand the areas of application for low-density wood species. For the past one hundred years, many efforts have been made to mass-produce densified wood products, but despite being available on the market, they remain niche products with low annual production volumes. One of the main reasons for this is that all available densified wood products are produced in a batch-type process, which limits the achievable process speed and integration into the continuous wood processing chain. For this reason, we propose a continuous surface densification process using a bespoke belt press – similar in principle to those used to produce MDF panels. The belt-press is capable of densifying full-sized wood boards at processing speeds of up to 60 m min^-1. The primary belt for densification can be heated to temperatures above 160°C, while a subsequent belt functions as a cooling stage. During the densification process, the belt press can log the pressing forces, moments, and temperature. Preliminary tests with Scots pine specimens of 120 mm in width and 38 mm in thickness resulted in a twofold increase in peak density, after a pressing time of two minutes at 120°C. The resulting density profiles were similar to those obtained in studies using a static hot press. As the belt press can be fed with a continuous stream of boards, it has a higher net throughput than a static hot press. Further studies continue with the aim to evaluate different aspects relevant to the large-scale industrial production of densified wood products.

The established methods for testing the hardness of wood are of questionable value for assessing the performance of surface-densified wood, since the density profile beneath the densified surface is an important property that needs to be considered. The purpose of this study was to evaluate the influence of the density profile of surface-densified wood and the hardness test parameters, such as indenter geometry and applied load on the measured hardness. The influence of the density profile varied considerably depending on the hardness test parameters. This can make a comparison of hardness values of surface-densified wood prone to misinterpretation. The selection of hardness test parameters should either be product-specific, or the density profile itself should be used to evaluate the hardness of surface-densified wood. A strong influence of the density profile on the indentation depth development during the hardness tests indicates the possibility of predicting the density profile based on the hardness test methods.

The density profile of surface-densified wood has a major influence on the indentation resistance of the material. A method that can predict the density profile in surface-densified wood from measurements of the indentation in a hardness test was established. The combined information of hardness and density profile is expected to better assess the performance of surface-densified wood. Density profile and hardness test data for surface-densified Scots pine have been subjected to a partial least squares analysis to determine the relationship between the indentation depth measured during a hardness test and the density profile measured by X-ray densitometry. Among seven different hardness tests, which varied in test force and indenter geometry, the Brinell method according to the EN 1534 standard showed the highest correlation between the indentation-versus-time curve and the density profile. The mean absolute error for the prediction of density profiles in an external test set was 5–10%, indicating that the method proposed in this study can be used to replace X-ray densitometry in process control and process design.

Over the past decades, the surface densification of solid wood has received increased attention. However, the inhomogeneous density distribution in the densification direction might be a challenge with regard to process control within a large-scale production process, as the density profile governs many relevant properties of surface-densified wood. Currently, the measurement of density profiles relies on sensitive X-ray equipment and is difficult to integrate into an on-line process. Hence, in this study, three machine learning approaches were applied to predict the density profiles of surface-densified Scots pine specimens, only based on visual image acquisition—a technology that is ubiquitous in the wood industry: partial least squares (PLS) regression, artificial neural networks (ANN), and convolutional neural networks (CNN). The machine learning models were trained on images of the specimen cross-sections as input data, and X-ray density profiles as output data. There were 1850 observations, and the model performance was evaluated on external test sets. The models had mean absolute percentage errors of the predicted values between 9 and 18%; the CNN achieving the smallest error (9.24%). A deeper analysis of the data revealed that the ANN approach performed inconsistently between observations. PLS regression predicted the main density peak to a high accuracy but could not model other features. Only the CNN could reliably model the main density peak, wide growth rings, and the important region between the specimen surface and the main density peak. The ability of the models to generalise to untypical new data was improved by augmentation of the training data.