Mould and stain became troublesome problems on lumber manufactured at a coastal British Columbia sawmill, and the difficulties were suspected to be a fault of the mill's antisapstain treatment. Forintek technical staff visited the sawmill and planer mill and used two procedures to gather assessment. Firstly, boards were pulled from spray box chains after spray treatment and one-inch square samples were punched from four faces of these boards for chemical analysis of DDAC, one of the active ingredients in the formulation being used. With a large number of samples, this procedure quantitatively assessed DDAC retention levels at points along boards. Secondly, fluorescent dye was added to the spray solutions and special paper strips were stapled along the length of boards prior to treatment in the spray box. The paper strips were then retrieved for examination under ultraviolet light. This procedure is largely a qualitative one, showing the pattern of spray coverage over the whole board. Liquid samples of the treatment solutions were taken from both the sawmill and planer cross chain day tanks to determine their DDAC concentrations.
A pilot plant apparatus was built to electrostatically spray phenolic resin on strands. To evaluate the resin distribution on these strands, an image analysis method was developed. The experimental conditions in this study made it difficult to compare electrostatic treatments to control (non- electrostatic) treatments. Although not statistically significant, there were notable differences between these treatments which indicate that electrostatic resin application may improve panel properties and is therefore worth further investigation. First of all, the electrostatic treatment produced panels with a 10% higher internal bond than the control. Secondly, the resin distribution results show that the electrostatic spray, on average, covered a 30% greater area of the strands than the control even though both treatments applied resin at the 2% resin solids level. Further experiments using alternative test procedures are planned to compare electrostatic treatments to control treatments that simulate industrial conditions.
A prototype scanning system successfully demonstrated the capability of real-time, log pocket monitoring that can be used to alert operators of improper pocket filling practice.
Several previous FPInnovations studies have shown that the log pocket filling is often poorly controlled resulting in poor log alignment and/or incomplete filling that adversely affects both flake quality and strander productivity. Until now there have been no systems available to automatically measure the alignment of logs being fed into the strander or measurement of the actual filling of the strander pockets. An automated monitoring system is needed to signal operators when the log pocket is improperly loaded. This feedback can allow operators to maintain proper log filling procedure to maximize flaking quality.
In March 2013 a machine vision scanning system comprised of a camera, 2 line lasers and computer with image analysis software, was tested for two days at the Peace Valley OSB mill (PVOSB) in Fort Saint John, BC. The scanner’s camera and lasers were mounted 13 feet above the center of the log pocket base. Images of pocket filling were acquired and analyzed for pass/fail conditions. Two image groups were selected for analysis, one of full pocket, aligned logs (pass) and the other containing misaligned logs and/or insufficient pocket fill (fail). For all pocket scans, scanner measurements were compared to manual visual classification.
Approximately 50 different pockets were scanned with results and images saved for analysis. For fill height and vertical log alignment measurement, the scanner correctly identified >95% of all the pockets examined. However, log alignment measurement in the horizontal x-y plane did not function as intended due to poor image contrast that could not be resolved during the mill trial. This technique has been shown to work well in previous pilot plant tests (Groves, 2012) which confirms that the underlying measurement fundamentals are sound. It is recommended that only minor lighting adjustments are required for the scanning system to work well in a mill setting.
Implementing this technology in OSB mills should help to reduce the occurrence of poor pocket filling that can adversely affect strand quality. It should be noted that even small improvements to strand quality and productivity can yield significant cost benefits. It is estimated that reducing fines by a modest 1% and improving productivity by 1% can return in excess of $1 million/year based on an average size Canadian OSB mill.
In plywood mills finished panels are manually graded and sorted based on specific defects. The panel edges are especially difficult to grade by human visual inspection due to the small nature of different defects, especially at higher line speeds. This can result in misread errors that can be costly.
In this project a prototype scanner, based on 3D laser profilometry, was developed by FPInnovations and demonstrated in two Canadian plywood mills for automatic edge grading. At both mills, panels were scanned on the production line in real time, collecting full length, 3-dimensional edge profiles in the x, y and z coordinate fields that were then analyzed by computer software sub-routines to identify defects for each panel. The data was processed to categorize groups of data points, depending on the edge profile depth (z-axis) variation, length (x-axis) and height (y-axis) into the specific defect categories of core, top and bottom edge void, core gap and core overlap, based on the values of predetermined edge profile thresholds.
Results from the two mill tests showed that the scanner was effective with a correct identification rate greater than 80%. The lack of panel hold-downs at each mill resulted in extreme height variation of the panel edges and this limited the defect detection accuracy. Based on the tests, the technology for automated edge grading is feasible. Longer term mill evaluations are recommended with adequate panel hold-downs in place before confirming that this technology is ready for commercialization.
A new laser-based system was successfully developed in the pilot plant for veneer thickness and roughness measurements. This system was tested for both green and dry spruce veneer. The comparisons were made between the green veneer measurement and dry veneer measurement, and between the laser-based system measurement and actual digital measurement. A linear mixed effect model was used to estimate the within-sheet and between-sheet variations of veneer thickness and roughness and their causes.
A good correlation was found between the laser-measured thickness and caliper-measured thickness. The laser-measured average roughness could also capture the trend of veneer surface roughness determined by the visual classification. Thus, the new laser-based system can be a useful tool for measuring both veneer thickness and roughness.
For veneer thickness, the within-sheet variation seemed to be larger than the between-sheet variation, and the laser-based measure had a larger variation than the digital-based measure for both green and dry veneer sheets. With the green veneer, higher veneer moisture content and density would lead to a larger difference between the two measurements. The laser-based method tended to classify more "thick" sheets than the digital-based method, but this tendency was not obvious with the dry veneer. Such tendency also became negligible by factoring in either veneer moisture content or density. Thus, in the real applications, the accuracy of the new laser system can be improved for measuring green veneer thickness with a calibration of moisture content and/or density.
For veneer roughness, the within-sheet variation was again larger than the between-sheet variation, and the dry veneer had a larger variation than the green veneer. Further, the tight side variation was generally larger than the loose side variation. The above information is deemed useful for establishing an overall veneer quality criterion for industrial applications. Further work is scheduled to adopt the new laser-based system for real-time measurement of green veneer thickness and roughness.
This project was started on April 1, 1999 following recommendations at the fall 1998 Structural Board Association (SBA) Technical Committee meeting in Banff, Alberta to quantify the benefits of the GluScan resin distribution analyzer for measuring resin distribution in OSB.
Resin distribution refers to how resin is dispersed on strand surfaces after the blending process. The main attributes of distribution include resin coverage and resin spot size which both can change due to blending variables. To date, it has been unclear whether or not resin distribution is sufficiently important to require monitoring by the OSB industry. To answer this question, experiments have been designed in this project to quantify how selected panel properties including Internal Bond, Modulus of Rupture, Modulus of Elasticity and Thickness Swell are affected by different resin distributions from batches blended at the Alberta Research Council (ARC). The resin distributions studied in this project were intended to simulate actual mill conditions and were measured by optically scanning strand surfaces using an image analysis system (GluScan) developed by Forintek Canada Corp.
Finally, a mill case study was carried out to determine whether the laboratory results together with GluScan could be used to measure and characterize resin distribution from a mill production line.
Research was carried-out to investigate the feasibility of electrostatically applying PF powder resin to OSB furnish to improve resin distribution and thereby improve panel properties. Research elsewhere has shown that resin distribution is an important factor affecting panel properties including internal bond strength and thickness swell. It was theorized that electrostatic charging would distribute resin more uniformly on strands compared to conventional application techniques and would help to strengthen the bond between resin and strands thereby minimizing resin losses and allowing for greater levels of resin to be retained on OSB furnish. Small-scale tests were conducted in Forintek's Vancouver laboratory followed by pilot plant tests at the Alberta Research Council (ARC) which involved electrostatically charging resin and aspen furnish in a pilot plant blender. Lab results showed that the average amount of resin retained by strands (retention) increased with electrostatic application compared to control applications. During pilot plant tests, image analysis results showed that resin coverage increased using electrostatics. Also, the coefficient of variation in resin coverage was lower for electrostatic applications compared to the controls. However, contrary to previous research findings, test panels made from electrostatic and control batches in a pilot plant did not show any significant differences in internal bond and thickness swell properties. Electrostatic application of powdered PF resin does not appear to sufficiently improve resin distribution in aspen to affect panel properties. Although pilot plant results were not promising, lab tests have shown that electrostatic application techniques are sensitive to material properties and that other wood species may be more suitable for charging than aspen. Further research will be continued at Forintek's eastern lab in the 1998/99 year to investigate electrostatic application of resins to mixed hardwood species.
It is important to improve drying productivity since this process is the bottleneck in plywood production. To address this issue, we evaluated the veneer moisture content (MC) distribution and the accuracy of radio frequency (RF) sensors both in the laboratory and in mill trials in this study.
We analyzed the distribution of green veneer moisture content (MC) and density from sapwood to heartwood for lodgepole pine logs via veneer peeling in the Forintek’s composites pilot plant. The results show that the MC distribution appears to be a dual-peak pattern both for heartwood and for sapwood. The position of the first peak is more consistent whereas the second peak varies more among logs. This information is useful for determining the proper number of green veneer sorts and the optimum cut-off MC level for each sort.
Readings of the radio frequency (RF) sensor are affected by veneer species, veneer density, veneer thickness, temperature, green veneer MC, grain angle, and distance between the sensor source and veneer surface. Of these variables, veneer density, veneer species and the distance are found to be the three main factors affecting the readings of the sensor. Based on the measurement results from different species, we conclude that the RF sensor is only suitable for measuring green veneer MC below 80%.
The correlation between readings of the sensor and green veneer MC varies from species to species. In general, this correlation could be improved using an exponential or a power equation. The readings of mill sensor are more inconsistent than those of the lab sensor for the heart sort veneer. The lab sensor underestimates MC for the heart sort veneer and overestimates MC for the sap sort veneer. However, the mill sensor overestimates in both positive and negative ways for each sort. To improve the accuracy of the readings, the sensor needs to be calibrated based on the species and veneer thickness, and the distance between the sensor source and veneer surface has to be kept as small as possible. However, this is not practical for the spruce, pine and subalpine fir (SPF) group since these species are not separated on a commercial basis.
We conducted three mill visits to: 1) measure the distribution of green veneer MC for different veneer sorts, and 2) assess the accuracy of current green veneer sorting. The results demonstrate that the accuracy of moisture sorting differs among mills and species, and the species mix like SPF generates larger MC variation within each sort. In general, the heart veneer sort is well done, but there is a significant overlapping between light-sap and sap veneer sorts. This indicates that the current industrial RF sensors cannot ideally sort higher MC veneer. The results also show that there is potential for more accurate green veneer sorting which would result in a 5% increase in drying productivity. This improvement would generate more than $1 million in annual savings per mill.
Le bois exposé à des conditions extérieures sur lequel on a appliqué un traitement de finition devient, en quelques années seulement, inutilisable s’il n’est pas efficacement protégé des ultraviolets et de l’humidité. Lorsque le bois est exposé aux intempéries, la conformité du traitement de finition est particulièrement déterminante, ce traitement constituant la principale défense contre les conditions atmosphériques. Par ailleurs, on observe une croissance de la demande des consommateurs à l’égard de la durabilité, du faible coût de l’entretien et, dans certains cas, de la préservation de l’aspect naturel du bois.
Le bois traité adéquatement et exposé à des conditions extérieures peut durer fort longtemps. Les produits de finition au latex, entre autres, sont efficaces pendant dix ans ou plus. Par contre, s’il n’a pas été traité adéquatement, le bois s’écaille, se fissure, s’érode ou se décolore après seulement un an d’exposition. La durabilité du traitement de finition est tributaire de divers facteurs, les plus importants étant les conditions d’utilisation, les propriétés du substrat du bois et le type de traitement. Pour maximiser le rendement du bois utilisé à l’extérieur, il est essentiel de bien comprendre la nature de ces facteurs et leur interrelation.