Automated Hardwood Lumber Grading

Logan Wells, Purdue University

Abstract

Grading lumber is a process of assigning a quality measurement and thus dollar value to individual boards. Throughout the entire hardwood manufacturing process and individual board can be graded multiple times from when the board is first sawn from the log, before and after drying and when it reaches the end manufacturer. For hardwood lumber both appearance and structural soundness factor into the grade of a board. The National Hardwood Lumber Association (NHLA) was created to establish a uniform standard for grading hardwood lumber. Many companies still have proprietary grades specific to their business, but the vast majority of hardwood lumber in North America is bought and sold based on NHLA standard lumber grades or a very similar variation. Currently the industry practice is manually grading lumber by specially trained lumber inspectors. It is a complex job that requires great focus and critical thinking due to the amount of mental math and production speeds. It takes a lot of time and resources to train a good lumber inspector, but that knowledge base of lumber grade is extremely valuable to all areas of hardwood manufacturing. Quite simply a lumber grader can make or break a mill if they are good or not. Finding talented young people to fill this role is becoming more difficult with the shortage of new people entering the industry. The gap between the number of students going into the industry and what is needed to fill retirements yet alone allow for growth is alarming and a change must be made going forward. Technology and automation is vital in allowing any industry to stay competitive in the global markets of today. This thesis is about an automated lumber grading system and its performance. As a proof of concept study, over 1,000 kiln dried, rough surface boards from nine different commercial hardwood species- ash, basswood, cherry, hard maple, hickory, red oak, soft maple, white oak and yellow poplar- were scanned and graded. Approximately 300 to 350 boards from three different grade categories, Select and Better, 1 Common and 2 Common, were scanned. A 2014 Microtec Goldeneye300 Multi-Sensor Quality Scanner™ was used to scan boards and collect data on the board defects and other features needed for grading. There are 6 different types of sensors used by the scanner including color cameras, black and white cameras, profile cameras, line lasers, dot-grid lasers and an x-ray. Together the sensors map out the board and create a digital map that is then processed by the lumber grading computer program GradeView™ as developed by Dr. Rado Gazo and researchers at Purdue University. The results of this study show the accuracy of the automated grading system to be within industry acceptable standards. The NHLA Sales Code (2015) states two grading accuracy requirements for selling lumber. Both requirements are based off the entire volume, or board footage, of lumber sold instead of a board by board count of accuracy. The first requirement is at least 80% of the total board footage must be of the specified grade, or at least 80% on-grade. The second requirement is that the true value of the load of lumber must be within 4% of the invoice specified value. Over the entire study and across all samples, the automated lumber grading system was found to be 92.22 % on-grade accurate and 99.50% on-value accurate. This thesis is comprised of two chapters that go over the entire study and results but also the specific detection processes for different lumber defects.

Degree

M.S.

Advisors

Gazo, Purdue University.

Subject Area

Forestry

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS