Publication in the Proceedings of the 1999 ASME NDE Engineering Topical Conference, San Antonio, TX

1. NDE OPPORTUNITIES FOR QUALITY ASSURANCE AND
PROCESS CONTROL IN THE FOREST PRODUCTS INDUSTRY
James R. Dickens, Ph.D.
Zhiyong Cai, Ph.D.
Temple-Inland Forest Products Corporation
Panel Products Division
Applied Research Center
P.O. Drawer N
Diboll, Texas 75941
ABSTRACT manufacturing facilities still rely on off-line destructive
testing methods for quality assurance.
In April 1998, Temple-Inland Forest Products Corporation
entered the medium-density fiberboard (MDF) market with This paper will highlight some of the existing NDE
the startup of a state-of-the-art mill in southern Arkansas. applications in our Company and to examine how NDE
Following the acquisition of two additional plants in October technologies may be applied to quality assurance and process
(Clarion, Pennsylvania, & Pembroke, Ontario), Temple control in the near future.
became the leading producer of MDF in North America –
virtually overnight. In recent years, Temple has strategically INTRODUCTION AND BACKGROUND
expanded its long-standing role in the particleboard and
gypsum wallboard markets. This rapid expansion includes yet Temple-Inland Forest Products Corporation operates 20 mills
another new product entry for the Company – cementitious which manufacture a variety of primary wood and building
fiberboard – which is expected to become a dominant product products, including lumber, plywood, particleboard, medium-
in the exterior siding market. density fiberboard (MDF), gypsum wallboard, cement
fiberboard, softboard sheathing, and hardboard siding & trim.
As the Company builds new mills and modernizes older mills, To support these operations, the Company owns and manages
computer control technologies are being integrated into the approximately 2.2 million acres of forestland, the majority of
production processes. PLC systems, which run from client- which is located in East Texas. The Company also owns and
server platforms, rely on accurate feedback of process data for operates a gypsum quarry in Fredricksburg, Texas. Annual
precise control. This is a relatively well-defined task with sales of the Building Products Group exceed $600 million.
respect to temperature, pressure, mass flow, position, velocity,
and the like. As yet, accurate on-line measurement of finished Within the Building Products Group, production facilities are
product characteristics is unavailable, and would lead to divided between two divisions: Solid Wood and Panel
potential breakthroughs in process control and optimization. Products. The Solid Wood Division operates five sawmills
and one plywood mill; the Panel Products Division operates
NDE has been in use for many years in the forest products the remaining 14 mills which include four particleboard mills,
industry, primarily as a means for process quality assurance. three MDF mills, four gypsum mills, one cement fiberboard
Examples of commercial applications include: ultrasonic blow mill, one fiber products mill, and one laminating operation.
detection systems, laser-based thickness measurement, With the exception of the laminating mill, all operations run
infrared moisture measurement, and x-ray density 24 hours/day, 7 days/week. Total annual production capacity
profilometry. While laboratory methods have shown promise of these 14 mills exceeds 3.5 billion square feet of product –
for quantitative measurement of mechanical properties, the enough to circle the globe 7 times!

2. Of the 14 mills in our division, fully half are less than five
years old; five mills are less than three years old; one is still X-ray Methods
under construction. The remaining 7 mills represent older X-ray techniques are used for density profilometry and may
technology, but have been upgraded during the last few years either be on-line, real-time systems or desktop versions used
to replace analog controls with state-of-the-art PLC and in the quality control lab. In both cases the goal is to
client-server control systems. The newer mills are being determine density profile through the thickness of the panel
designed from the ground up with state-of-the-art technology. (this is not an imaging process, per se). Control of the density
profile is critical for achieving good mechanical properties
Computerization opens up new possibilities with respect to and other measures of board performance such as
process control. Dozens of individual process settings can be machinability, surface quality, and warp.
continuously and simultaneously monitored and adjusted with
respect to established process targets. Even so, feedback Density profilometry is the only NDE technique we use which
control systems rely on accurate process measures in order to provides quantitative, real-time feedback on the product – but
be effective. For variables such as temperature, pressure, mass this technology is only being used in one MDF plant. On-line
flow, position, velocity, etc., accurate sensors are already (real-time) density profilometry is particularly useful for
available. The remaining challenge is to quantitatively adjusting press variables (e.g., temperature, time, closing
measure the output – i.e., the final product – of the process, in speed) which directly affect density profile. Depending on the
order to judge the effectiveness of process targets and control end use of the product, it may be more desirable to have
schemes. At present, we continue to rely on offline higher density on the surface and lower density in the core
destructive testing of a small sample of finished product as a (i.e., a U-shaped profile), or it might be more desirable to
means for representing the population of all material have a relatively flat density profile.
produced. This is a costly, time-consuming, and relatively
impractical task. Small sample sizes lead to high probabilities While average density is highly correlated to the internal bond
for both Type I (missed defect) and Type II (false call) errors, strength of wood products, it is by no means the only
resulting in a potentially high risk to producer and consumer determinant. Resin content and press variables also strongly
alike. influence internal strength, thus, measurement of density
alone is not a sufficient indicator of internal board strength.
NDE TECHNIQUES CURRENTLY IN USE For this reason, density profile is not the most robust method
for determining internal bond strength.
While the forest products industry has not widely embraced
the state-of-the-art with respect to nondestructive evaluation Optical & Visual Methods
(NDE) for process control, there exists already a number of Laser thickness gauges have been in use for 20 years in our
sensors that employ some of the basic nondestructive three older particleboard plants. These are used to provide
techniques, e.g., acoustic/ultrasonic, radiographic, & feedback on out-of-press thickness. When indicated by
optical/visual methods. This section will briefly describe how changes in board thickness, adjustments may be made to the
each method is employed in the various manufacturing press closing variables. There is a delicate balance to strike
processes. with respect to out-of-press thickness – if the board is too
thin, the sanded board will have low spots and must be
Ultrasonic Methods downgraded. If the board is thicker than necessary to prevent
Ultrasonic “blow” detectors have been in use for 30 years in low spots, excess energy and material is lost in the sanding
wood panel manufacturing facilities [1]. The blow detector is process and surface defects may result. Even under optimal
simply an array of stationary ultrasonic transducers conditions, sanding removes 5-10% of the material from the
configured for through-transmission. The blow detector is board.
located on the finishing end of the production line, usually
right after the hot press. Each board passes through the sensor Moisture content is one of the most critical process variables
array on its way down the production line. in wood manufacturing. The offline method of oven-drying a
small sample of wood is not practical from a process control
The function of the blow detector is to assess the integrity of standpoint, as fluctuations in moisture content can be rapid
the panel, i.e., to ensure that no internal delaminations exist. and significant. Moisture content depends on many factors,
Delaminations may be caused either from resin precure, including the initial moisture content of the wood, the manner
incomplete cure, or excess steam pressure inside the panel. in which wet and dry wood is mixed, and processing variables
Current systems in use today rely on pass/fail criterion for such as dryer temperature, mass flow rate, and resin variables.
rejecting a defective product. As a result, it is possible to pass Excessive moisture leads to blows during the press cycle;
board which has internal continuity, but lacks sufficient insufficient moisture leads to delaminations and generally
mechanical strength to meet performance specifications. requires longer curing times. Thus a careful balance must be
Because the blow detector uses a pass/fail decision criterion, maintained.
the potential remains high for Type I and Type II errors.

3. The most common in-process method for measuring moisture finished board properties. This helps the operator make
utilizes near-infrared (NIR) technology. By analyzing the adjustments to the blending & pressing variables in order to
spectroscopic characteristics of reflected NIR radiation, the compensate for changes in the raw material stream. As an
moisture content can be estimated. While there is some loss of indication of the potential significance of this technique, the
accuracy compared to oven-dry testing, the need for authors note that six papers related to NIR are scheduled to be
continuous feedback from the process is of primary presented at the annual meeting of the Forest Products Society
importance and outweighs any loss of accuracy. on June 27-30, 1999 – enough papers to justify a full-day
technical session devoted to emerging measurement
Visual inspection is still relied upon for final product quality technologies.
assurance. Typically, the sander operator is responsible for
grading the boards as they feed out of the sander. At Numerous controlled laboratory studies have shown some
production speeds, the operator has about 2-4 seconds to promise with respect to quantitative NDE of wood materials
make a final visual check of the board as it passes by the [5-8]. The most significant hurdle appears to be transferring
inspection station on its way to the unitizer (boards are the technology from a controlled lab environment to the plant
typically 4’x8’, but in our largest particleboard plant the site where heat, humidity, dust, and vibration are the norm.
operator must inspect a 9’x25’ panel!). The inspection task is Another key obstacle with respect to plant deployment is the
assisted somewhat by the use of mirrors and bright lights; fact that the freshly-pressed board is undergoing gradual
nevertheless, heavy demands are placed upon the operator for changes in temperature, moisture content, degree of resin
maintaining accuracy and consistency over long periods of cure, and – consequently – strength. Thus, time-temperature-
time. property relations complicate deployment into the plant
environment.
LITERATURE REVIEW
DISCUSSION AND CONCLUSIONS
While the techniques described in the previous paragraphs
give an indication of the utility of NDE for process control, There are a number of potential benefits that make it
the ability to accurately predict mechanical properties of the worthwhile to pursue an on-line measurement system. For
finished product in real time remains elusive. Two examples example:
of developments in real-time NDE are cited here, each with
caveats. • reduced process variability, leading to lower
process targets, and reduced material and energy
Back in the mid-80’s, one of Temple’s major competitors usage
developed an on-line ultrasonic internal bond measurement • reduced downfall and reject:
system which was subsequently deployed into four • less frequent destructive tests
particleboard plants between 1988 and 1990. An overview of • more timely operator response to changes in
the system is reported in [2], where the system utilized a the process output
through-transmission arrangement with rolling wheel • lower risk of Type I errors
transducers. While the paper indicates that deployment was • improved utilization of press time, leading to
initially successful, we understand from one of the system increased production capacity
developers that no further installations were made, and the
• reduced labor costs
four systems originally deployed have since fallen into disuse.
• reduced risk of Type II errors, leading to improved
The reasons why are not known to us at this time.
customer satisfaction and fewer product claims
About two years ago a commercial vendor advertised an
In addition to improving quality control, an on-line board
“add-on” to their standard ultrasonic blow detector, which
measurement system would provide a valuable tool to guide
claimed to predict internal bond strength by means of a neural
process improvement efforts. Feedback from the process
network [3]. At that time we concluded that the device was
would be available in minutes rather than hours, so that the
still at an early stage of development and deployment. No
net effect of process changes would be more quickly
other commercially-available quantitative systems are known
understood.
to be available at this time.
Operationally, quantitative NDE methods would prove most
NIR techniques are stepping up as a legitimate means for
beneficial to our primary wood products – particleboard and
improving product quality. One application that we are aware
MDF – which we produce at a rate of 1.1 billion square feet
of utilizes NIR technology to characterize the wood raw
per year, with relatively high quality costs. Gypsum products
material [4]. This principle appears to be similar to that used
represent about 2.1 billion feet of production, but traditionally
with NIR for moisture detection, but the sensors are calibrated
have much lower quality costs. Thus, NDE technology might
to detect specific chemical differences in the constituent wood
yield a higher return by focusing on improved efficiency of
which are then empirically correlated to variations in the

4. the gypsum manufacturing process. In products like cement
fiberboard and hardboard siding, internal bond and flexural
strength are important to ensure good long-term performance.
As with the other panel products, NDE might prove beneficial
either as a process control tool or a quality assurance tool for
exterior products.
Based on our experience with building materials, we feel that
the method with the greatest likelihood of success for QNDE
is air-coupled ultrasound. Air-coupled ultrasound has been
studied for many years, but only recently have advances in
transducer design and electronic instrumentation brought air-
coupled ultrasound to a high level of technical feasibility [9].
Our goal during the next few months is to prove the viability
of air-coupled ultrasound under laboratory conditions, then
move into one of the production operations with a prototype
system for continuous on-line testing.
REFERENCES
1. Birks, A.S., 1972, “Particleboard blow detector,” Forest
Products Journal, Vol. 22 No. 6, pp. 23-26.
2. Rodgers, J.M., et al., 1991, “Acousto-ultrasonic
measurement of internal bond strength in composite
wood products,” Materials Evaluation, pp. 566-571.
3. Electronic Wood Systems GmbH, Germany.
4. Johnsson, B., et al., 1998, “Process modeling system for
particleboard manufacturing, incorporating near-infrared
spectroscopy on dried-wood particles,” Proceedings of
the 32nd International Particleboard/Composite Materials
Symposium, Washington State University, p. 164.
5. Dickens, J.R., 1996, “Evaluation of finger-jointed
lumber strength using critically-refracted longitudinal
waves and constituent wood properties,” Ph.D.
dissertation, Texas A&M University, College Station,
TX.
6. Tucker, J., 1996, “Angled-beam ultrasonic inspection of
particleboard quality,” M.A. Record of Study, Texas
A&M University, College Station, TX.
7. Han, M-B., 1993, “Nondestructive evaluation of end-
jointed lumber using ultrasonic techniques,” Ph.D.
dissertation, Texas A&M University, College Station,
TX.
8. Bethi, R., 1994, “Grading of lumber using stress waves,”
M.S. thesis, Texas A&M University, College Station,
TX.
9. Grandia, W.A., and Fortunko, C.M., 1995, “NDE
applications of air-coupled ultrasonic transducers,”
Proceedings of the IEEE UFFC Symposium, Seattle,
WA.