Use of Dense Hardwood in Engineered Wood Product ...

Summary
Use of Dense Hardwood in Engineered Wood Product Manufacturing
Stephen Delahunty
Graduate Research Assistant
Faculty of Forestry and Environmental Management, University of New Brunswick
Fredericton, New Brunswick, Canada
Ying H. Chui
Professor
Faculty of Forestry and Environmental Management, University of New Brunswick
Fredericton, New Brunswick, Canada
There is an apparent world-wide trend to utilize hardwood in producing engineered wood products
(EWP). Dense hardwood lumber provides a better potential for producing high strength EWP than
softwood lumber. A procedure to evaluate the economics of using a new hardwood source material
in EWPs that considers lumber quality is presented. The mechanical property used here to
characterize lumber quality is modulus of elasticity. This procedure is demonstrated using a case
study in which red maple lumber was used as flange stock and laminates to produce predetermined
I-joist and glulam beams.
1. Introduction
The engineered wood product (EWP) industry in North America has experienced exceptional
growth in recent years. Currently, virtually all lumber based EWPs are made from high quality
softwood dimension lumber. This is despite the fact that hardwood lumber generally possesses
better mechanical properties than softwood lumber, which in theory should make hardwood lumber
a more suitable material than softwood lumber in EWP applications. Hardwood lumber has
traditionally been used in making appearance products and its use in structural applications has been
limited. One exception is the use of aspen/poplar in manufacturing oriented strandboard. The
fabrication of structural wood members using hardwood components has recently attracted the
attention of research organizations and industry in North America and Europe. In Europe extensive
research on utilizing birch (Solli 2004) and beech (Glos et al 2004) in manufacturing glulam has
been conducted. The United States Forest Products Laboratory undertook a number of studies to
evaluate the use of hardwood species as dimension lumber (McDonald et al 1993, Green and
McDonald 1993), and as glued laminated timber (glulam) (Manbeck et al 1993, Moody et al 1993,
Janowiak et al 1995). In the United States, hardwood glulam products are covered in the American
Institute of Timber Construction (AITC) standard (AITC 1996) however specifications for
hardwoods are lacking in the corresponding Canadian products and design standards. A Canadian
wood products research institute, Forintek Canada Corp, evaluated a structural laminated product
concept of combining low-grade white birch lumber as face laminates and balsam fir as core
laminates (Blanchet 2003).
The majority of lumber based EWPs produced in eastern Canada are made using lumber sawn from
dense black spruce trees that grow in northern Quebec. The slow growth rate of these trees, along
with the presence of small encased knots, makes the resulting lumber ideal for structural
applications. There is, however, a growing raw material supply shortage concern in eastern Canada
mainly due to the outstanding increase in production of EWPs in the recent past.
This paper explains a procedure that was developed to evaluate the economics of introducing a new
hardwood source material for EWPs taking into consideration lumber quality so that the total value

of each log can be maximized. Although this procedure is primarily based on modeling this paper
focuses on the characterization of lumber quality which is expressed here in terms of modulus of
elasticity (MOE). The overall procedure provides a link between logs sawn into lumber, and lumber
manufactured into EWPs. A series of lumber yield predictions are developed and incorporated into
a simulation model which uses Monte Carlo sampling techniques to model the variability in lumber
yield and lumber quality. The output from the simulation model, the inventory of mechanically
graded E-rated lumber available to produce EWPs, is transferred to an optimization model designed
to maximize the total revenue resulting from the sale of the finished EWPs. A case study in which
red maple was used in the production of wood I-joists and glulam is described to demonstrate the
lumber quality component of this procedure.
2. Procedure
Approximately 300 red maple sawlogs were obtained and geometric log characteristic data was
collected. This log data was fitted to probability distributions to simulate the variation of log
geometry. The log geometry values assigned by the model are used to determine lumber yields. The
logs were divided into two groups with one group processed into 38mm x 64mm lumber to be used
as flanges for I-joists and the other group processed into 38mm 89mm lumber to be used as
laminates for glulam beams. All lumber was then dried to a target moisture content of 15 percent
(oven-dried basis) and planed to the proper finished dimension. A sample of lumber (n = 32) was
selected using a stratified sampling method so that the top diameter of the logs would be
proportionately represented. The lumber MOE was measured according to ASTM D198 test
procedure (ASTM 2004). The same sample of lumber were then passed through a two pass constant
deflection Cook-Bolinder machine to measure the apparent MOE, termed Emean. Since lumber is
heterogeneous, MOE depends on span, orientation (edgewise or flatwise in bending), load speed of
test (static or dynamic) and method of loading (tension, bending, concentrated or uniform) (Green
and Kretschmann 1991). The static bending test for MOE outlined in ASTM D198 was used here as
a baseline measure of the true MOE.
MOE (MPa)
20000
18000
16000
14000
12000
10000
8000
6000
4000
2000
0
R 2 = 0.74
9000 11000 13000 15000 17000 19000 21000
Emean (MPa)
Figure 1 Correlation between MOE and Emean
Figure 1 is a linear regression analysis
using Emean as an indicator of MOE and
equation [1] describes the resulting line
of best fit. Finally, the remaining
lumber from the red maple sawlogs was
passed through the grading machine
and MOE was calculated for each piece
to be used in the simulation model. An
analysis revealed no significant
relationship between log diameter and
lumber MOE, therefore all MOE values
were grouped into a single probability
distribution to represent the variation of
lumber MOE.
( E ) 1944
MOE = 0. 94*
mean − (MPa) [1]

5,780 MPa 11,200 MPa 16,300 MPa
Figure 2 Histogram of MOE with fitted probability density function
The MOE values were fitted
to a probability distribution
to represent the variation of
lumber MOE. Figure 2 is a
histogram of the lumber
MOE values with overlying
beta probability distribution
function (α = 3.38, β = 3.50).
The distribution type that
contains the lowest mean
square error was selected to represent the observed data. The simulation model was run a total of six
times with the number of logs in each run set to 1000 and the log length set to 2.4m. The resulting
lumber from each run was divided into E-rated lumber groups. The average MOE of each group is
based on the MOE requirements needed to produce the I-joist flanges and glulam laminates. Table 1
summarizes a hypothetical inventory of I-joists (IJ) and glulam beams (GL), which were established
to demonstrate the optimization model. The lumber was grouped, by dimension, according to the Erated
lumber groups given in Table 1. The I-joist prices are wholesale prices for 6.1m long I-joists
containing conventional black spruce flanges. Glulam beam lay-ups were taken from AITC 119-96
standard specifications for structural glued laminated timber of hardwood species (AITC 1996)
using the MOE requirements for red maple. The wholesale prices for 6.1m long industrial grade
24f-EX Douglas fir-larch was used as representative of the red maple glulam beams since they
contain similar design properties.
Table 1 Summary of I-joist and glulam beam inventory
Product
Flange/laminate
dimension (mm)
No. flanges/laminates
(MOE) (MPa)
Depth (mm)
Price
($Cdn)
IJ-1 38 x 64 2 (13,800) 302 35
IJ-2 38 x 64 2 (10,300) 302 30
IJ-3 38 x 89 2 (11,200) 302 35
GL-1 38 x 89 2 (13,800), 2 (12,400), 4 (NoE) 302 313
GL-2 38 x 89 2 (13,800), 4 (12,400), 4 (NoE) 381 391
GL-3 38 x 89 4 (13,800), 2 (12,400), 6 (NoE) 457 470
3. Results and discussion
The output from each simulation model run was included in the optimization model with the
product definition and prices. Table 2 gives the optimal product mix that produces the maximum
total revenue.
Table 2 Optimal product mix yielding maximum total revenue
Run #
IJ-1
I-joist product tally
IJ-2 IJ-3
Glulam product tally
GL-1 GL-2 GL-3
Total revenue
1 91 344 186 42 0 48 $55,696
2 94 329 165 99 0 14 $56,486
3 86 324 174 73 0 30 $55,749
4 100 362 175 89 0 16 $55,846
5 100 343 163 101 0 12 $56,733
6 98 356 181 86 0 20 $56,746
Average 95 343 174 82 0 23 $56,209
Std error 2.3 6.0 3.6 8.9 0.0 5.6 $204
Since Emean from the grading machine is a good indicator of the baseline measure for the true MOE,
the grading machine can be calibrated so that every piece of lumber can be evaluated for MOE.

It can be seen from Table 2 that the variation of the total revenue is relatively stable, with a
coefficient of variation of 1.4 percent. Although the total revenue is stable, the optimal glulam
product mix varies considerably between runs. This suggests that the optimal glulam product mix is
sensitive to the available inventory of E-rated lumber. In each run no ‘GL-2’ beams were produced.
This is because there was a relatively low proportion of E-rated lumber with an average MOE of
12,400MPa available, and since this lay-up required a high proportion of this group, the model
selected the other two available products. It was found that approximately half of the lumber with
no MOE requirement (NoE) was left over as a surplus inventory. This could be identified and
corrected in practice by either finding an alternative use for the surplus lumber or by modifying the
inventory of products so that no surplus remained. The major limitation of this model is that it does
not account for production costs. If the cost to produce each product was included the model would
be more valuable in that it could be used to maximize the profitability of the operation rather than
simply the total revenue.
4. Conclusion
This method of evaluating the economics of low grade hardwood in EWPs is not only a useful tool
to predict the optimal product mix that maximizes the total revenue, but it is also useful to simply
simulate the operation. Various “what-if” scenarios can be evaluated such as how a change in log
supply will affect the outcome of the optimal product mix. The ability to answer such questions
efficiently and inexpensively without disruption to the operation is a valuable tool for making
informed decisions.
5. References
[1] McDonald, K.A., Green, D. W., Dwyer, J. and Whipple, J.W. 1993. Red maple stress-graded 2 by 4 dimension
lumber from factory-grade logs. Forest Products Journal, 43(11/12):13-18.
[2] Green, D. W. and K. A. McDonald. 1993. Mechanical properties of red maple structural lumber. Wood and
Fiber Science. 25(4):365-374.
[3] Manbeck, H.B, Janowiak, J. J., Blankenhorn, P. R., Labosky, P., Moody, R. C., and Hernandez, R. 1993.
Performance of red maple glulam timber beams. Research Paper FPL-RP-519, USDA Forest Service, Forest
Products Laboratory, Madison, WI.
[4] Moody, R. C., Hernandez, R., Davalos, J. F., and Sonti, S. S. 1993. Yellow Poplar glulam timber beam
performance. Research Paper FPL-RP-520, USDA Forest Service, Forest Products Laboratory, Madison, WI.
[5] Janowiak, J. J., Manbeck, H.B, Hernandez, R., Moody, R. C., Blankenhorn, P. R. and Labosky, P. 1995.
Efficient utilization of red maple lumber in glued-laminated timber beams. Research Paper FPL-RP-541, USDA
Forest Products Laboratory, Madison, WI.
[6] Blanchet, P. 2003. Special applications of low grade hardwood in EWP. Report to Canadian Forest Service,
Forinetek Canada Corp, St-Foy, PQ.
[7] AITC. 1996. Standard specifications for hardwood glued-laminated timber. AITC 119-96. American Institute of
Timber Construction, Vancouver, WA.
[8] Solli, K. H. 2004. Tensile strength of Nordic birch. Proceedings of Meeting Thirty-seven of CIB-W18, paper 37-
6-1, University of Karlsruhe, Karlsruhe, Germany.
[9] Glos, P., Denzler, J. K. and Linsenmann, P. W. 2004. Strength and stiffness behavior of beech laminations for
high strength glulam. Proceedings of Meeting Thirty-seven of CIB-W18, paper 37-6-3, University of Karlsruhe,
Karlsruhe, Germany.
[10] ASTM. 2004. Standard test methods of static tests of lumber in structural sizes. Designation D198-02. American
Society for Testing and Materials, West Conshohocken, PA.
[11] Green, D. W. and D. E. Kretschmann. 1991. “Lumber property relationships for engineering design standards”,
Wood and Fiber Science 23(3):436-456.