Supervisor: Gary S. Schajer
Renewable Resources Laboratory
Close dimensional accuracy of the sawn wood is a major manufacturing objective when cutting lumber in a sawmill. Any dimension errors are a serious problem because they create low quality products and waste large quantities of a costly and environmentally sensitive raw material. Consequently, sawmills use quality control procedures to monitor dimension accuracy and, when necessary, to indicate the need for production machine repair. However, the traditional practice of manually selecting sample lumber pieces and measuring the thickness at a few points is only minimally effective. Such measurements can only reveal gross errors, often many hours after the start of the underlying machine problem. During that time, much defective material has been produced, causing substantial waste.
The use of in-line dimension sensors such as shown in the diagram significantly improves lumber quality control monitoring. Many thickness measurements can be made using laser triangulation sensors along the length of each board, and dimension errors can be identified promptly. However, such measurements can only determine material thickness, but cannot separate the surface height profiles on each side of the lumber pieces. Such separation would be very useful because each side of a lumber piece typically is cut using a different saw, possibly also on a different machine. Thus, if each side could be measured separately, the particular saw and machine needing repair could be identified and corrective action promptly undertaken, thereby greatly reducing material waste. By analyzing the characteristic shape of the individual surface profile, it is also possible to identify the specific repair needed.
A serious challenge to measuring height profiles on separate lumber surfaces is caused by the lateral motion of the lumber as it passes longitudinally along a practical production line. Ideally, the lumber would move in a perfect straight line so that the height measured by the sensor would directly indicate the surface height profile of the lumber. However, the lateral motions add to these measurements, thereby seriously contaminating them. (They are still useful for thickness measurements because the lateral motions equally affect the measurements on each side and cancel out during combination of the opposite readings.)
A sensor system has been developed that is able to separate the surface height profiles from the lateral motions. It uses multiple sensors to measure a surface, as shown in the example arrangement shown. As the lumber passes along the production line, the sensors measure the surface in delayed sequential order. However, any lateral motion is seen by the sensors simultaneously. Thus, the surface profile can be determined mathematically by separating the delayed sequential component of the measurements from the simultaneous component. This calculation is an “inverse problem”, and needs to be done with care to minimize noise sensitivity and computation time.
The sensor arrangement in the diagram measures surface height profile along one line within the surface of the lumber. Other sensor arrangements can be configured to make measurements along two or more lines. However, with discrete point sensors, the measurements are limited to being one-dimensional along specific lines. In many cases, the features of interest on lumber are spread over the entire surface, and thus are two-dimensional. Such surface features include wane, saw offset, and washboarding. Present work involves the use of line sensors to enable the identification of a 2-dimensional wood surface map. Compact mathematical techniques are being developed to enable the required calculations to be completed in real time.
In-line dimension monitoring provides a valuable method for real-time lumber dimension quality control. This ensures that the material is manufactured to a consistent high quality, and that if faults occur in the manufacturing process, they are quickly identified, diagnosed and repaired. The measurement system developed here allows accurate wood surface height measurements to be made, even when measurements are made with industrial conveyors that are not designed for accurate material movement.