Scientists & Staff

Ed Thomas

Research Computer Scientist
480 Cornbread Ridge Road
Princeton, WV, 24740
Phone: 304-431-2324

Contact Ed Thomas

Current Research

Log and tree quality: Surface defects on hardwood logs provide a map to predicting product and lumber quality. To this end, this scientist is developing automated methods using high-resolution laser scanning to locate and classify log surface defects. In addition, he has developed models capable of accurately predicted internal defect features based on external defect measurements. Thus, allowing a complete digital representation of a log, including: size, shape, external and internal defects to be generated. In a recent validation study on red oak, it was found that the scanning and defect modelling system could predict approximately 80% of all surface knots on the lumber sawn from the sample logs.

Process Simulation: The ROMI rough mill simulation software developed by Ed Thomas allows users to specify their rough mill configuration including: ripsaw, chopsaw, panel options, prioritization, part sizes, part quality options, and lumber grade mix to perform a complete rough mill cut-up simulation. Simulation results detail the volume and number of boards required to satisify the cutting bill requirements, as well as the number of strips, strip yield, part yield, and the number of ripping and chopping operations that were required. Small changes in grade mix, part sizes, and processing options can result in unpredictable changes to yield and processing results. Using simulation allows users a way to experiment with their rough mill operations to determine the most efficient processing methods before cutting a single board.

Research Interests

I have two primary research interests.  The first is automated defect detection on hardwood logs and developing models that predict internal defect attributes based on external defect indicators. Related to this I am also working on developing and refining sawmill simulation programs that use this "glass log" data to examine questions related to log quality and processing.

The second is the design and development of automated information location and classification systems that seek information related to specific areas of forest product markets and knowledge areas.  An example of this work can be found at the UMN CLT Knowledge Base Project.

My other interests involve secondary processing and the development and use of simulation programs to examine related research questions.


Why This Research is Important

The overall goal of this research is to determine the most efficient processing method: be it for sawing a log into boards, or sawing a board into moulding, flooring, or dimension parts. Poor and un-informed sawing decisions result in sawing mistakes and lower quality lumber. By knowing where the defects are on a log, the sawyer can position the log such that greater volumes of higher quality lumber are sawn. This scientist is working with WVU to develop grade-based optimizing log software which will determine the sawing strategy which will yield the highest valued NHLA lumber grade recovery. (Under NHLA rules, higher quality and larger boards have the highest value.) Thus, more of the log is converted to lumber, and in turn the higher quality lumber results in less waste when the board is sawn into parts. Using the ROMI process simulator, users can determine the most efficient processing method which has the highest yield and least waste. By reducing waste and in-efficiency in the sawmill and the rough mill, we can conserve the forest resource.

Featured Publications & Products

Publications & Products

National Research Highlights

Laser scanned image of a low grade, small diameter yellow-poplar log with large knots.

Investigating the potential of cross-laminated timber panels made from low-grade hardwoods for building construction

Year: 2017

The emergence of cross-laminated timber (CLT) for building construction in North America may provide an additional and valuable product market for low-grade hardwood logs.

High-resolution laser scan image of a log with detected defect areas highlighted and acoustic waves passing through. U.S. Department of Agriculture Forest Service.

Using New Technologies to Improve Log Defect Detection

Year: 2016

Scientists from the Forest Service’s Forest Products Laboratory and Northern Research Station cooperated with West Virginia University to enhance the competitiveness of the U.S. hardwood lumber industry by improving the accuracy of log internal defect detection and maximizing log value recovery.

High resolution laser scan of a red oak log. R. Edward Thomas, U.S. Department of Agriculture Forest Service.

Analyzing Internal Hardwood Log Defect Prediction Equation Accuracy

Year: 2016

The type, size, and location of internal defects dictate the grade and value of lumber sawn from hardwood logs. Significant correlations have been documented among external log defect indicators and internal defect features. Forest Service scientists developed a series of prediction models for four hardwood species based on these correlations.

High-resolution point cloud image of a scanned red oak log. Ed Thomas, USDA Forest Service

Improved Automated Detection of Surface Defects on Hardwood Logs

Year: 2014

In less than one second, a new parallel computer algorithm processes more than a million surface data points on a hardwood log to find the defects.

Screenshot of Hardwood Log Sawing Program showing control panel, log end and side views. Ed Thomas, USDA Forest Service

RAYSAW Computer Program Can Grade Hardwood Logs and Calculate its Value

Year: 2013

RAYSAW is a computer program developed by Forest Service scientists for hardwood log sawing that processes high-resolution laser-scanned hardwood logs with defects and user-described sawing patterns and creates a set of lumber that can be graded or processed using other available Forest Service computer programs. The laser scan surface allows log volumes and sawing residue to be precisely determined. Thus, a complete valuation of a log, including lumber, cant, bark, sawdust, and residue can be calculated.

Last modified: Wednesday, September 30, 2020