1

3 Structural Performance of Hybrid SPFs-LSL Cross-Laminated Timber

4 Panels
5

7 William G. Davids*1, Nicholas Willey1, Roberto Lopez-Anido1, Stephen Shaler2, Douglas Gardner2, Russell
8 Edgar3, and Mehdi Tajvidi2

9
10
1
11 Department of Civil and Environmental Engineering, University of Maine, 5711 Boardman Hall, Orono,
12 Maine, USA, 04469-5711
2
13 School of Forest Resources, University of Maine, 5755 Nutting Hall, Orono, Maine, USA, 04469-5755
3
14 Advanced Structures and Composites Center, University of Maine, 35 Flagstaff Road, Orono, Maine,
15 USA, 04469-5793
*
16 Corresponding author: william.davids@maine.edu, +1 207 581-2116

17

18

19

© 2017. This manuscript version is made available under the Elsevier user license
http://www.elsevier.com/open-access/userlicense/1.0/
20 Abstract
21 The bending and shear performance of hybrid cross-laminated timber (CLT) panels made from Spruce-

22 Pine-Fir (South) (SPFs) and laminated strand lumber (LSL) are examined. Four configurations of three-

23 layer CLT were fabricated: all-SPFs control specimens, all-LSL specimens, hybrid specimens with SPFs

24 faces and an LSL core, and hybrid specimens with LSL faces and an SPFs core. Bending tests were

25 conducted to assess flexural strength and stiffness. Additionally, three-point bending tests were

26 performed to assess shear performance. The incorporation of LSL in the core of CLT panels increased

27 mean panel bending stress at failure by 23% through mitigation of rolling shear failure.

28 Keywords: cross-laminated timber, laminated strand lumber, wood shear strength, wood flexural

29 strength, structural testing

2
30 1 Introduction

31 Originally introduced in Austria and Germany in the mid-1990s, cross-laminated timber (CLT) has

32 become an increasingly popular alternative for multi-story timber construction in Europe [1]. CLT has

33 recently garnered interest in North America with the establishment of several CLT and nail-laminated

34 timber plants in Canada and the United States. CLT panels are suitable for use in walls, floors and roofs,

35 and are typically fabricated from an odd number of flat-wise layers of solid-sawn lumber placed in

36 alternating 90 degree directions. In the majority of cases, individual layers of boards are adhesively

37 bonded although nail- and screw-laminated CLT is also produced. Alternative forms of CLT have been

38 considered including placing laminations at +/- 45 degrees as well as hollow, box-based systems [2].

39 Compared to typical concrete construction, CLT structures are lightweight, sequester more carbon,

40 possess better thermal insulation properties, and are more rapidly erected [3].

41 Research on the structural performance of CLT can be separated into the broad categories of seismic

42 behavior [4-7], fire resistance [8,9], and determination of the mechanical properties of CLT. A number of

43 studies have focused on the determination of CLT mechanical properties in flexure and shear, which are

44 primary design properties for panels subjected to out-of-plane loading [10-15]. Sikora et al. [10] present

45 a current and thorough review of the existing literature on this topic. Others have focused on CLT

46 mechanical response due to in-plane loading [16-18].

47 As with plywood, an issue which can limit the capacity of CLT subjected to out-of-plane loading is failure

48 in perpendicular-to-grain shear, commonly called rolling shear. Rolling shear also contributes to

49 deflections of CLT panels. Because of its significance, several investigations have considered rolling shear

50 properties and failure mechanisms. Zhou et al. [19] examined the effect of rolling shear deformations in

51 3-layer, black spruce CLT, measuring rolling shear modulus and conducting three-point bending tests of

52 CLT specimens. Zhou et al. [19] also proposed a deflection adjustment factor to account for rolling shear

3
53 deformations, and also concluded that bending specimen width did not significantly affect apparent

54 elastic modulus and apparent shear modulus. Li et al. [20] implemented a torsional test for evaluating

55 rolling shear strength in CLT, observing that thinner cross-layers tended to have higher rolling shear

56 strengths. Hochreiner et al. [14] studied CLT plates subjected to concentrated loads and examined the

57 evolution of rolling shear failure modes by tracking fracture development and load-deformation history

58 using digital image correlation. Li and Lam [21] experimentally assessed rolling shear damage

59 accumulation in CLT attributable to load cycling, and calibrated a damage accumulation model that can

60 be used for future studies on duration-of-load behavior of CLT under rolling shear. Noting the

61 significance of rolling shear on CLT structural performance, Aicher et al. [22] assessed the rolling shear

62 modulus and strength of European beech, which typically has much better rolling shear properties than

63 softwoods normally used in CLT construction, concluding that the use of beech in CLT cross-layers could

64 be beneficial for CLT strength. Wang et al. [23] assessed the use of laminated strand lumber (LSL) in both

65 cross-layers and face layers of hybrid CLT panels, demonstrating increased flexural capacities relative to

66 conventional all-softwood CLT.

67 The literature indicates that rolling shear failure can be a limiting factor for the strength of CLT subjected

68 to out-of-plane loading. The focus of the research reported in this paper was the structural assessment

69 of hybrid CLT panels made from LSL and softwood lumber with the objective of increasing strength by

70 mitigating rolling shear failures in the core layer. LSL is an engineered composite lumber that is made

71 from approximately 300 mm long strands of fast-growing species (often aspen or poplar) that are

72 bonded and densified during manufacture and oriented with the long axis of the structural member. LSL

73 typically possesses good dimensional stability and very predictable strength and stiffness values

74 compared to solid-sawn lumber. Additionally, the authors are aware of no published experimental

75 research specifically examining the use of Northeastern U.S. Spruce-Pine-Fir (South) (SPFs) lumber in

76 CLT. SPFs is an economically significant group of lumber species harvested in the United States that

4
77 includes Eastern Spruces, Balsam Fir, Red Pine, Jack Pine, Englemann Spruce, Lodgepole Pine, Sitka

78 Spruce and Norway Spruce. All species are subject to the same grading rules and have the same design

79 values. CLT panel production using SPFs harvested and milled in the Northeastern US may become an

80 important new market for lumber producers in the United States as CLT markets grow. The research

81 reported in this paper includes characterization of the SPFs and LSL lumber used for CLT manufacturing,

82 assessment of the bond between SPFs and LSL using a polyurethane adhesive, and testing to assess both

83 major-axis flexural and shear strength and stiffness.

84 2 Materials and Methods

85 Materials were 38 mm x 184 mm x 3 m kiln-dried SPFs No. 2, 38 mm x 184 mm x 3 m grade 1.35E LSL

86 boards (without wax coating on the board edges), and Henkel PURBOND HB E452 single-component

87 polyurethane adhesive. The No. 2 grade of SPFs is a standard grading category corresponding to specific

88 stiffness and strength design values in the US codes, and is a commonly produced grade of SPFs lumber.

89 The SPFs lumber was procured in bulk quantities from Pleasant River Lumber in Dover-Foxcroft, Maine,

90 USA, and as discussed later, a small percentage of the SPFs was No. 1, a higher grade with higher design

91 values. The LSL was provided by Louisiana-Pacific Corporation’s plant in Houlton, Maine, USA. The

92 designation “1.35E LSL” refers to a specific grade of LSL that has a nominal elastic modulus of 9310 MPa.

93 The 1.35E LSL was selected as opposed to a higher grade – 1.55E and 1.75E grades with moduli of 10,700

94 MPA and 12,070 MPA are also available – because its flexural strength and stiffness were expected to be

95 similar to the SPFs. Two, three-layer CLT panels were laid up for each of four configurations as described

96 in Table 1. The CLT panels were made from continuous boards as opposed to the finger-jointed lumber

97 typically used in commercially fabricated CLT panels.

98

5
 99 2.1 Lumber Characterization and Preparation

100 Each SPFs and LSL board was first E-rated using a Metriguard 340 E-Computer, its moisture content (MC)

101 taken with a Delhorst J2000 pin moisture meter, and its dimensions measured and density calculated. In

102 the E-rating process, the board is placed flatwise on two supports, one of which contains a small load

103 cell. The board is struck at mid-span with a hammer, and the dynamic load cell readings are used to

104 compute a dynamic modulus of elasticity (MOE). A total of over 900 SPFs and over 700 LSL boards were

105 measured to permit the fabrication of additional panels beyond those discussed here. Table 2

106 summarizes the results of the lumber characterization study. The MOE values were adjusted to 12% MC

107 using the procedure defined in ASTM D1990 [24] to allow direct comparison with design values.

108 The average SPFs MOE was significantly higher than expected. The National Design Specification [25]

109 reports the mean MOE for SPFs as 7.58 GPa for No. 2 and 8.27 GPa for No. 1, and the average MOE of

110 the all SPFs used in this study exceeded 8.27 GPa by 34%. While only 4.2% of the SPFs lumber was

111 stamped No. 1, a visual inspection indicated that the vast majority of the SPFs lumber was red spruce

112 (Picea rubens). For comparison, the Wood Handbook [26] gives an average MOE for clear red spruce at

113 12% MC of 11.45 GPa. In contrast, the LSL MOE was only 1.5% less than the tabulated value of 9.31 GPa

114 [27]. Further, as expected the LSL MOE was much less variable than the SPFs MOE. To ensure that no

115 excessively compliant material was used in CLT panel fabrication, the SPFs boards with MOE values in

116 the lower 5% of the distribution, which corresponded to an MOE of less than 6.89 GPa, were removed

117 from the lot. This shifted the mean MOE to from 11.05 GPA to 11.35 GPa and reduced the coefficient of

118 variation (CoV) in MOE from 19.6% to 15.2%.

119 Following MOE testing, both the SPFs and LSL were conditioned in a dehumidification dry kiln to reduce

120 the MC differential between the two materials and promote better adhesive bonding. The SPFs lumber

121 was conditioned for five days after which it had reached an average MC of 10.8%. The LSL boards were

6
122 conditioned for 27 days, reaching a MC of 9.4%. The resulting MC differential of 1.4% was well within

123 the recommended moisture content differential of no more than 5% specified in PRG 320 [28].

124 2.2 Assessment of Bond Strength

125 The manufacturer-recommended spread rate for the PURBOND adhesive was 100 – 180 g/m2. This

126 relatively wide range, combined with the uncertainty associated with bonding LSL to SPFs, dictated that

127 an adhesive spread rate study be conducted. To accomplish this, adhesive compression shear block

128 testing was performed per ASTM D905 [29] for adhesive spread rates of 98, 122, 146 and 171 g/m2 for

129 SPFs to SPFs and SPFs to LSL. For each adhesive spread rate and lay-up, a 127 mm x 305 mm two-layer

130 lamination was made from which 10 shear block specimens were cut. Specimens were fabricated from

131 conditioned boards that had been planed to a thickness of 19 mm. Laminates were pressed at 0.01 MPa

132 for two hours per the product standard, and cured per the requirements of ASTM D905. Following each

133 shear block test, strength and percent wood failure were recorded, and each specimen was oven-dried

134 and weighed to determine MC. Results of the shear block tests are given in Table 3. Based on these

135 results, an adhesive spread rate of 146 g/m2 was chosen for CLT panel manufacturing. This adhesive

136 spread rate gave the highest percent wood failure for the SPFs-SPFs specimens, and the highest average

137 shear stress for the SPFs-LSL specimens. We note that the spread rates reported here will likely not be

138 applicable to other brands and types of adhesives.

139 2.3 Panel Fabrication and Test Specimen Preparation

140 Two 2.45 m-long X 1.32 m-wide panels of each of the four CLT configurations were fabricated. Both SPFs

141 and LSL boards with minimal warp, twist, bow or cupping were used to ensure reasonable dimensional

142 tolerances. Within two hours of adhesive application, each board was planed to a final thickness of 35

143 mm, with approximately 1.6 mm removed from each face. Average MC was determined using a pin

144 moisture meter at the time of panel lay-up. Prior to adhesive application, the lumber surface was

7
145 moistened with a light water spray. A pre-measured amount of PURBOND adhesive was applied using

146 putty knives.

147 Following the application of adhesive and panel lay-up, pipe clamps were used to squeeze the panels in

148 the transverse direction and ensure gaps of no more than 3.2 mm between boards in the face layer. The

149 CLT panels were placed in a 1.22 m x 2.44 m hydraulic press for two hours at a pressure of 1.03 MPa.

150 With the exception of one all-SPF specimen, which had a gap of 4.8 mm between two adjacent boards in

151 one face layer, the panels met the tolerance requirements of PRG 320 [28]. Three flexure and three

152 shear specimens were cut from each panel. Six specimens were tested in flexure for all four layups, and

153 six specimens were tested in shear for layups L1, L2 and L4.

154 2.4 Flexure Test Protocol

155 Quasi-static four-point flexural tests were conducted in accordance with ASTM D198 [30] as shown in

156 Figure 1. Load was applied with a single 145 kN hydraulic actuator, and two wooden load heads with a

157 406 mm radius were attached to a spreader beam. The specimen span was 2.32 m measured between

158 the centerline of the supports, and the load heads contacted the beam at the third points of the simple

159 span. Each support had a steel top plate that rested on a roller allowing free rotation. To ensure that the

160 relatively long supports did not affect the response and that the support reactions were uniformly

161 distributed to the specimen, 102 mm long neoprene bearing pads were sandwiched between the

162 support top plate and the bottom of the beam. Seven string potentiometers were attached at mid-

163 depth on one side of the specimen: one string pot at each support, one at each load head, one midway

164 between each support load head, and one at mid-span.

165 Prior to testing, each specimen was weighed. A pre-load of 1.33 kN was applied, and a load rate of 5

166 mm/min ensured specimen failure within 6-20 minutes as specified by ASTM D198 [30]. After loading,

167 failure mode and location were noted, and a 25.4 mm thick cross-sectional slice was taken from an

8
168 undamaged region of the beam. Dimensions and weight of the cross-sectional slice were recorded, and

169 it was then oven-dried for at least 15 days at 105o C to determine average moisture content at the time

170 of testing.

171 2.5 Shear Test Protocol

172 Quasi-static three-point bend tests were conducted in accordance with ASTM D198 [30] to assess shear

173 strength. Load was applied at the middle of the 0.619 m span with a single, 406 mm radius wooden load

174 head. The supports were identical to those employed in the flexural tests. String potentiometers were

175 attached to one side of the specimens at each support and at mid-span (see Figure 2).

176 As with the flexure tests, each specimen was weighed and its dimensions measured prior to the start of

177 the test, and post-test, oven-dry moisture content was determined. A constant displacement rate of

178 1.52 mm/min was used, which produced failure after about 10 minutes.

179 2.6 Assessment of Specimen Stresses at Failure from Test Results

180 The maximum bending stress f bmax was determined only for the flexural specimens, and was calculated

181 using Equation 1. The maximum moment Mmax and effective section modulus Seff were computed using

182 Equations 2 and 3, respectively. Pmax is the maximum total actuator load, which was evenly split

183 between both load heads. Equation 2 applies for four-point bending with a span length L and the

184 effective section modulus assumes a transformed section, with the effective depth heff taken as the

185 distance between the extreme fibers of the longitudinal layer(s). PRG 320 [28] recommends that the

186 transformed section modulus Itr be computed assuming a modular ratio n of 30 for the center layer.

187 However, n = 17.6 was used here based on tabulated stiffness properties for red spruce given by Bodig

188 and Jayne [31], since the majority of the lumber used appeared to be red spruce.

M max
f bmax  Eq. 1
Seff

9
 Pmax L
M max  Eq. 2
6
2I
S eff  tr Eq. 3
heff
189

190 The maximum shear stress f vmax was computed from both the flexural and shear test results using Eqs. 4

191 and 5 as recommended in the U.S. CLT Handbook [32]. In Eq. 5, the summation upper limit of 3/2

192 indicates that the shear stress is evaluated at the center of the cross-section. The elastic modulus of the

193 outer layers was taken as the mean value determined from board testing and the elastic modulus of the

194 center layer was taken as E/17.6 for the SPFs. For the LSL, the elastic modulus of the center layer was

195 taken as 1.03 MPa, the value recommended by the LSL manufacturer. The effective elastic bending

196 rigidity EIeff was computed per Equation 6. The thickness of each layer is hi and zi is the distance from

197 the center of a given layer to the center of the cross-section.

Pmax  Q 
f vmax    Eq. 4
2  Ib eff
3

Q
  
 2
i 1
E i hi z i
Eq. 5
 Ib  eff EI eff
3  h3 
EI eff  i 1  Ei bi i  Ei Ai z i2  Eq. 6
 12 
198

199 2.7 Assessment of Specimen Stiffnesses from Flexure Test Results

200 The load-displacement response for each specimen was used to determine the effective flexural rigidity

201 EI eff , which accounts for only bending deformations; the apparent flexural rigidity EI app , which

202 incorporates the effect of both bending and shear deformations; and GAeff , the effective shear rigidity.

203 All stiffness values were computed based on specimen response between 20% and 50% of maximum

10
204 measured load Pmax. Therefore, loads generically denoted as P in the following equations are computed

205 as 0.5Pmax  0.2 Pmax . The term P which appears in Eqs. 7-9 was computed from a linear regression of


206 the appropriate measured load-displacement response between 0.2 Pmax and 0.5Pmax .

207 EI eff was computed first for each specimen using Eq. 7, which applies given that shear deformations

between the load heads are zero. Here the shear-free deflection within the load span,  b , was
ls
208

209 computed as the difference between the mid-span displacement and average of the displacements at

210 the load points.

 P  L3
EI eff   ls  Eq. 7
  432
 b 
211

212 EI app was determined for each specimen using Eq. 8, where  m is the total mid-span displacement

213 minus the average support displacement, and includes both bending and shear deformations.

 P  L3
EI app    Eq. 8
  m  56
214

215 Finally, GAeff was computed using Eq. 9, where k was taken as the shear correction factor for a

216 rectangular section of 5/6. Eq. 9 was applied independently at the location of both load heads and mid-

217 span, and the resulting three values of GAeff were averaged to give the reported value for each

218 specimen.

P L
GAeff    Eq. 9
  s  6k
219

11
220 In Eq. 9, the shear deflection,  s , was computed as the total measured deflection minus the bending

deflection at the load head,  b , and mid-span,  b , computed with Euler beam theory and the
0.33L 0.5 L
221

222 experimentally estimated value of EI eff determined using Equation 7. Equations 10 and 11 apply for

223 computing bending deflections at the load heads and mid-span, respectively.

5 PL3
0b.33L  Eq. 10
324 EI eff
23PL3
0b.5 L  Eq. 11
1296 EI eff
224

225 3 Results and Discussion

226 3.1 Flexure Tests: Capacities and Failure Modes

227 Typical flexural load-deflection responses for a single specimen of each layup are given in Figure 3. The

228 reported mid-span deflection is the value measured by the stringpot at mid-span minus the average of

229 the deflections recorded at the support centerlines. Failure load, failure mode, stiffness characteristics

230 and computed stresses are listed for each specimen in Table 4, and averages are presented for each

231 layup in Table 4. The failure loads, stiffness values, and stresses in Table 4 have been adjusted to 12%

232 MC using the methods recommended by ASTM D1990 [24]. Further, reported stresses were calculated

233 based on actual cross-sectional dimensions assuming equal thickness layers and using methods

234 discussed later in this paper.

235 The L2 specimens generally exhibited the lowest degree of variability in both failure load and stiffness,

236 which can be attributed to their being made entirely of LSL, an engineered lumber that typically has

237 more consistent mechanical properties than solid wood. While the failure load CoV was the highest for

238 layup L4, this could be attributed to specimen L4-1 failing at the location of a large knot in the SPFs

12
239 bottom face layer at a relatively low load of 48.9 kN. Further, all L4 specimens failed in flexural tension

240 of the SPFs bottom layer, and therefore variability in peak load was to a large extent driven by defects in

241 the SPFs. Discarding specimen L4-1 gives a mean failure load of 75.2 kN with CoV of 8.9% for layup L4;

242 this CoV lies between the peak load CoV for lay-up L1 (all SPFs) and L3 (LSL faces and SPFs core).

243 Failure modes were consistent within a given lay-up. All L1 specimens exhibited perpendicular-to-grain

244 shear failures, five of six L2 specimens exhibited flexural tension failures, five of six L3 specimens

245 exhibited flexure tension failures, and all six L4 specimens exhibited flexural tension failures. A typical

246 tension failure is shown in Figure 4, and a typical shear failure is shown in Figure 5. The fact that the

247 average bending stress at failure for the L4 specimens was 23% higher than for the L1 specimens

248 combined with the shift in failure mode from shear to flexural tension supports the premise that

249 incorporating LSL in the core can increase CLT panel shear strength.

250 The L2 and L3 specimens had very similar average failure loads and exhibited consistent flexural tension

251 failures. This indicates that the 1.35E LSL used in the faces of the L2 and L3 panels had a lower average

252 flexural strength than the SPFs used in the faces of the L1 and L4 layups. This may seem counter-

253 intuitive based on the larger allowable flexural stress of 1.35E LSL compared to No. 2 or better SPFs.

254 However, as noted earlier, the lower 5% of the SPFs boards based on MOE were discarded, and

255 relatively straight SPFs boards were selected from the remaining stock for panel manufacturing. This

256 board selection likely increased the overall quality of the SPFs used in panel fabrication.

257 3.2 Flexure Tests: Stresses at Failure

258 The nearly identical average bending stress at failure for layups L2 and L3 can be attributed to the fact

259 that for 10 of these 12 specimens, flexural tension failures occurred in the bottom layer of LSL. The

260 highest average flexural stress was for layup L4, where the high shear strength of the LSL core drove

261 flexural tension failures in the bottom SPFs face layer. All of the L1 specimens exhibited shear failures,

13
 max max
262 which explains why their average f b of 40.7 MPa is 19% less than the average f b of 50.1 MPa

max
263 achieved by the L4 specimens. The average f b of 50.1 MPa for the L4 layup is 9.4 times greater than

264 the allowable bending stress of 5.34 MPa for No. 2 SPFs given by the National Design Specification [25].

265 While a direct comparison between allowable and observed actual maximum stress should be done with

266 caution, this large discrepancy indicates that the No. 2 or better SPFs used for CLT panel fabrication was

267 of relatively high quality. This is consistent with the high MOE values of the SPFs determined from board

268 testing as discussed earlier.

269 The average shear stress at failure for the L1 lay-up was 1.26 MPa, and should reflect the actual shear

270 strength of the SPFs core material under this loading condition given that all L1 specimens failed in

max
271 shear. This value of f v falls within the shear strength range of 1 - 2 MPa reported by Sikora et al. [10]

272 for Sitka spruce CLT specimens.

max
273 Specimen L3-6, the only L3 specimen that failed in shear, had a computed f v of 1.17 MPa, which is

274 only 7.1% less than average shear stress of 1.26 MPa for the L1 specimens. Computed shear stresses for

275 all other L3 specimens were lower, ranging from 0.94 to 1.13 MPa. The average shear stress in the LSL

276 core of the L4 specimens was 1.62 MPa. If the results for specimen L4-1 are discarded because of its very

277 low failure load driven by a large knot on the bottom face, the average computed shear stress at failure

278 for the remaining five L4 specimens increases to 1.72 MPa.

279 3.3 Flexure Tests: Stiffnesses

280 The L1 lay-up had the highest average EI eff and the L2 lay-up the smallest, which is consistent with the

281 higher average MOE values for the SPFs boards than the LSL boards. The L3 lay-up, with LSL in the faces,

282 had a slightly (4.1%) greater EI eff than the all-LSL L2 lay-up, and EI eff for the L4 lay-up with SPFs in the

283 faces is only 4.6% lower than EI eff for the L1 lay-up. However, when both bending and shear

14
284 deformations are considered via EI app , the L4 lay-up is stiffer than L1. This is likely attributable to the LSL

285 having a greater shear modulus than the SPFs. This is consistent with the larger value of GAeff

286 determined for the L4 lay-up, which reflects the LSL core.

287 The experimentally-determined EI eff values were compared with those determined analytically using

288 the average measured board MOE value of 11.35 GPa for the SPFs, the average measured MOE of 9.16

289 GPa for the LSL, and transformed section analysis. This approach gives EI values of 320, 259, 258 and

290 321 kN-m2 for layups L1, L2, L3 and L4, respectively. Comparing with the mean values in Table 4 shows

291 that the analytical approximations compare reasonably well with experimentally determined means.

292 Using the experimental values for reference, the errors in approximations are -13%, -3%, -7%, and -9%

293 for layups L1 – L4, respectively.

294 3.4 Shear Tests

295 Based on the flexural test results, short-span shear tests were conducted on layups L1, L2 and L4. Layup

296 L3 was not considered further, since the flexure tests indicate no clear advantage to using low grade LSL

297 face material with an SPFs core. Results for all specimens are provided in Table 5.

max
298 The L1 specimens all failed in shear (see Figure 6) at an average shear stress, f v , of 2.03 MPa, 61%

299 higher than average shear stress of 1.26 MPa at failure for the L1 flexure specimens. However, bending

300 stresses at failure were 54% higher for the L1 flexure specimens than for the L1 shear specimens. It is

301 possible that the observed L1 flexure specimen shear failures initiated near the load heads and were

302 therefore influenced by simultaneous significant bending stresses. Ig this was the case, the average

303 shear stress at failure of 2.03 MPa may be more representative of the perpendicular-to-grain shear

max
304 strength of the SPFs core material. This value of f v is only slightly higher than the upper bound of

305 shear strength of 2 MPa reported by Sikora et al. [10] for Sitka spruce CLT specimens.

15
 max
306 The L2 specimens, made entirely of LSL, had an average f v of 2.61 MPa. However, four of the L2

307 specimens failed in tension, and the shear strength of the LSL therefore cannot be confidently inferred

308 from the L2 data. All the L4 shear specimens exhibited shear failures (see Figure 7) with an average

309 f vmax of 2.96 MPa. This is 83% greater than the average shear stress at failure of 1.62 MPa for the L4

310 long-span flexure specimens. However, as noted earlier, all the L4 long-span specimens failed in flexural

311 tension, indicating that their average shear stress at failure of 1.62 MPa is a lower bound on LSL shear

max
312 strength. Therefore, the shear stress at failure f v of 2.96 MPa is likely the most accurate estimate of

313 the LSL perpendicular-to-grain shear strength, which implies that the LSL has a perpendicular-to-grain

314 shear strength 46% greater than that of the SPFs. Wang et al. [23] reported a mean perpendicular-to-

315 grain shear strength for LSL of only 1.43 MPa, about half the value of 2.96 MPa found in this study.

316 However, Wang et al. [23] also noted that the short-span LSL specimens they tested failed in tension as

317 opposed to shear, which could explain this large difference.

318 4 Summary and Conclusions

319 This study examined the bending and shear performance of three-layer CLT consisting of four-different

320 layups: all SPFs (lay-up L1), all LSL (lay-up L2), LSL-SPFs-LSL (lay-up L3) and SPFs-LSL-SPFs (lay-up L4).

321 Based on long-span bending tests, the L4 panels were the strongest, followed by the L1, L3 and L2

322 panels. The hybrid L4 panels had a mean bending stress at failure 23% greater than the all-SPFs L1

323 panels, and the inclusion of low-grade LSL in the core shifted the failure mode from perpendicular-to-

324 grain shear to flexural tension. Short-span shear test results indicate that the low-grade LSL has a shear

325 strength about 46% greater than that of the SPFs. This supports the observation that the inclusion of LSL

326 in the core prevents shear failures in the three-layer CLT panels studied here.

327 These results indicate that hybrid panels with an LSL core may have a structural advantage over all-SPFs

328 panels, allowing for increased span lengths or load-carrying capacity for a given panel span-to-depth

16
329 ratio. In contrast, the L2 and L3 lay-ups showed only a 3% difference in average bending strength, which

330 reflects the fact that 83% of these panels failed in flexural tension of the LSL face layer. Further, the

331 average bending stress at failure of the all-LSL L2 specimens was 35% less than that of the L4 specimens,

332 and 16% less than that of the L1 specimens.

333 However, as noted earlier in this paper, a low grade (1.35E) of LSL was used in panel fabrication. Further,

334 based on measured MOE, the SPFs material appears to have been higher quality than typical No. 2 or

335 better material. Therefore, the results reported here may not apply for three-layer CLT panels made

336 from a higher grade of LSL and/or a more typical sample of SPFs material. Further, comparisons between

337 mean strength values do not directly apply to design strengths, which take into account LSL’s inherently

338 low material variability. Finally, the specimens tested here were made from continuous boards as

339 opposed to the finger-jointed lumber typically used in commercially fabricated CLT panels. Finger joints

340 tend to reduce board tensile capacity, and as a result could decrease the differences in flexural strength

341 observed for different layups. Despite these caveats, the inclusion of low-grade LSL as a core material

342 can clearly increase the strength of CLT panels that would otherwise fail in shear.

343 5 Acknowledgements

344 This project was supported by Agriculture and Food Research Initiative Competitive Grant no. 2013-

345 34638-21491 from the USDA National Institute of Food and Agriculture. We are also grateful for the

346 materials provided by Pleasant River Lumber, Louisiana-Pacific, and Henkel Adhesives.

17
347 6 Literature Cited
348
349 [1] Mohammad M, Gagnon S, Douglas BK, Podesto L (2012). Introduction to Cross-Laminated Timber.
350 Wood Design Focus, 22(2): 3-12.
351 [2] Chen Y, Lam F (2013). Bending performance of box-based cross-laminated timber systems. Journal of
352 Structural Engineering, 139(12): 04013006.
353 [3] Robertson A. (2007). “A comparative life cycle assessment of mid-rise office building construction
354 alternatives: Laminated timber or reinforced concrete.” Master’s thesis, Univ. of Toronto, Toronto,
355 ON, Canada.
356 [4] Fragiacomo M, Dujic B, Sustersic I (2011). Elastic and ductile design of multi-storey crosslam massive
357 wooden buildings under seismic actions. Engineering Structures, 33(11): 3043–3053.
358 [5] Gavric I, Fragiacomo M, Ceccotti A (2015a). Cyclic behavior of CLT wall systems: experimental tests
359 and analytical prediction models. Journal of Structural Engineering, 141(11): 04015034.
360 [6] Gavric I, Fragiacomo M, Ceccotti A (2015b). Cyclic behavior of typical metal connectors for cross-
361 laminated (CLT) structures. Materials and Structures, 48: 1841-1857.
362 [7] Pei S, van de Lindt JW, Popovski M (2013). Approximate R-factor for cross-laminated timber walls in
363 multistory buildings. Journal of Architectural Engineering, 19(4): 245-255.
364 [8] Frangi A, Mario F, Hugi E, Jöbst R (2009). Experimental analysis of cross-laminated timber panels in
365 fire. Fire Safety Journal, 44: 1078-1087.
366 [9] Fragiacomo M, Menis A, Clemente I, Bochicchio G, Ceccotti A (2013). Fire resistance of cross-
367 laminated timber panels loaded out of plane. Journal of Structural Engineering, 139(12): 0401308.
368 [10] Sikora K, McPolin DO, Harte AM (2016). Effects of thickness of cross-laminated timber (CLT) panels
369 made from Irish Sitka spruce on mechanical performance in bending and shear. Construction and
370 Building Materials, 116: 141-150.
371 [11] Hindman DP, Bouldin JC (2015). Mechanical properties of southern pine cross-laminated timber.
372 Journal of Materials in Civil Engineering, 27(9): 04014251.
373 [12] Ido H, Nagao H, Masaki H, Kato H, Ogiso J, Miyatake A (2016). Effect of the width and lay-up of sugi
374 cross-laminated timber (CLT) on its dynamic and static elastic moduli, and tensile strength. Journal
375 of Wood Science, 62: 101-108.
376 [13] Park H-M, Fushitani M, Sato K, Kubo T and Byeon H-S (2006). Bending creep performances of three-
377 ply cross-laminated woods made with five species. Journal of Wood Science, DOI 10.1007/s10086-
378 005-0750-7.
379 [14] Hochreiner G, Füssl J, Eberhardsteiner J (2014a). Cross-laminated timber plates subjected to
380 concentrated loading: experimental identification of failure mechanisms. Strain, 50: 68-81.
381 [15] Hochreiner , ssl J, Serrano E and Eberhardsteiner J (2014b). Influence of Wooden Board Strength
382 Class on the Performance of Cross-Laminated Timber Plates Investigated by Means of Full-Field
383 Deformation Measurements. Strain, 50(2): 161-173.

18
384 [16] Vessby J, Enquist B, Petersson H, Alsmarker T (2009). Experimental study of cross-laminated timber
385 wall panels. Eur. J. Wood Prod. 67: 211–218.

386 [17] Bogensperger T, Moosbrugger T, Silly G (2010). Verification of CLT-plates under loads in plane.
387 Proceedings of the 11th World Conference on Timber Engineering, Riva del Garda, Italy, 2010.

388 [18] Christovasilis IP, Brunetti M, Follesa M, Nocetti M, Vassallo D (2016). Evaluation of the mechanical
389 properties of cross laminated timber with elementary beam theories. Constr. Build. Mater. 122:
390 202–213
391 [19] Zhou Q, Gong M, Chiu YH, Mohammad M (2014). Measurement of rolling shear modulus and
392 strength of cross laminated timber fabricated with black spruce. Construction and Building
393 Materials, 64: 379-386.
394 [20] Li M, Lam F, Li Y (2014). Evaluating rolling shear strength properties of cross-laminated timber by
395 torsional shear tests. Proceedings of the 2014 World Conference on Timber Engineering, Quebec
396 City, Canada, August 10-14.
397 [21] Li Y and Lam F (2016). Low cycle fatigue tests and damage accumulation models of the rolling shear
398 strength of cross-laminated timber. Journal of Wood Science, 62: 251-262.
399 [22] Aicher S, Christian Z, Hirsch M (2016). Rolling shear modulus and strength of beech wood
400 laminations. Holzforschung, DOI 10.1515/hf-2015-0229.
401 [23] Wang Z, Gong M, Chui Y-H (2016). Mechanical properties of laminated strand lumber and hybrid
402 cross-laminated timber. Construction and Building Materials, 101: 622-627.
403 [24] ASTM (2014). D1990-14 standard practice for establishing allowable properties for visually-graded
404 dimension lumber from in-grade tests of full-size specimens. American Society of Testing and
405 Materials, West Conshohocken, PA: ASTM International.
406 [25] American Wood Council (2012). National Design Specification: Design Values for Wood Construction
407 (2012 ed.). Leesburg, VA: American Wood Council.
408 [26] Forest Products Laboratory (2010). Wood Handbook - Wood as an Engineering Material. Madison,
409 WI: U.S. Department of Agriculture.
410 [27] ICC Evaluation Service (2015). LP SOLIDSTART Laminated Strand Lumber (LSL) and Laminated Veneer
411 Lumber (LVL). Technical Report No. ESR-2403, ICC Evaluation Service.
412 [28] American National Standard Institute & APA - The Engineered Wood Association (2012). ANSI/APA
413 PRG 320-2012. In Standard for Performance-rated Cross-Laminated Timber. Tacoma, WA: APA - The
414 Engineered Wood Association.
415 [29] ASTM (2008). D905-08 standard test method for strength properties of adhesive bonds in shear by
416 compression loading. American Society of Testing and Materials (2013 ed.). West Conshohocken,
417 PA: ASTM International.
418 [30] ASTM (2014). D198-14 standard test methods of static testing of lumber in structural sizes. In
419 American society of testing and materials. West Conshohocken, PA: ASTM International.
420 [31] Bodig J and Jayne BA (1982). Mechanics of Wood and Wood Composites. New York, NY: Van
421 Nostrand Reinhold Company.

19
422 [32] FPInnovations & Binational Softwood Lumber Council (2013). CLT Handbook: Cross-Laminated
423 Timber (E. Karacabeyli & B. Douglas, Eds.). Pointe-Claire, QC: FPInnovations.
424

20
425
426
427
428
429
430
431
432 Table 1: CLT Configurations

Face Core
Configuration
Material Material
L1 SPFs SPFs
L2 (2.8%)
LSL LSL
L3 LSL SPFs
L4 SPFs LSL
433
434

21
435
436
437
438
439
440
441
442
443
444 Table 2: Summary of Results from Board E-Rating
445 (MC = moisture content)
Avg. MOE Specific Density MC
(GPa) Gravity (kg/m3) (%)
SPFs 11.05 (19.6%) 0.42 (2.8%) 479 14.5
LSL 9.16 (5.1%) 0.68 (7.8%) 718 5.0
446 (CoV in parentheses)

447
448

22
449
450
451
452
453
454
455
456
457 Table 3: Summary of ASTM D905 Shear Block Test Results
458 (MC = moisture content)
Average Range of
Spread Rate Avg. Shear
Lay-Up % Wood % Wood MC (%)
(g/m2) Stress (MPa)
Failure Failure
SPFs - SPFs 98 11.1 (5.5%) 98 80-100 11.9 (1.3%)
SPFs - SPFs 122 11.6 (5.9%) 90 40-100 11.2 (3.7%)
SPFs - SPFs 146 9.4 (8.4%) 100 100-100 11.7 (1.5%)
SPFs - SPFs 171 10.2 (12.5%) 98 80-100 11.9 (2.7%)
SPFs - LSL 98 8.5 (19.5%) 75 50-100 6.4 (1.3%)
SPFs - LSL 122 9.0 (20.4%) 88 20-100 6.7 (2.9%)
SPFs - LSL 146 10.0 (11.5%) 91 80-100 6.4 (2.4%)
SPFs - LSL 171 8.8 (11.7%) 98 95-100 6.5 (4.4%)
459 (CoV in parentheses)

460
461

23
462
463 Table 4: Long-Span Flexure Test Experimental Results

464 (MC = moisture content; EIeff = effective bending rigidity; EIapp = apparent bending rigidity; GAeff =
max max
465 effective shear rigidity; f b = maximum bending stress; f v = maximum shear stress)

Failure
MC EIeff EIapp GAeff f bmax f vmax Failure
Specimen Load
(%) (kN-m2) (kN-m2) (kN) (MPa) (MPa) Mode
(kN)
L1-1 62.8 11.1 386 346 6914 44.8 1.40 Shear
L1-2 54.3 9.2 356 297 3661 39.3 1.20 Shear
L1-3 61.5 10.9 340 302 5419 43.9 1.37 Shear
L1-4 50.2 9.4 313 263 3376 35.9 1.12 Shear
L1-5 61.4 10.1 420 322 2817 44.1 1.37 Shear
L1-6 52.2 9.1 392 298 2528 36.1 1.11 Shear
56.7 10.0 368 305 4119 40.7 1.26
Mean L1 –
(10.4%) (8.9%) (10.6%) (9.1%) (41.3%) (10.1%) (10.7%)
L2-1 48.2 7.6 259 236 5393 33.9 1.11 Tension
L2-2 47.8 7.6 285 240 3125 33.7 1.10 Tension
L2-3 48.5 8.9 267 239 4652 34.5 1.11 Tension
L2-4 46.0 6.9 262 230 3764 32.7 1.05 Tension
L2-5 48.5 8.1 265 238 4637 34.8 1.13 Tension
L2-6 43.1 7.5 262 227 3445 30.6 0.98 Shear
47.0 7.6 266 235 4170 33.4 1.08
Mean L2 –
(4.5%) (3.8%) (3.5%) (2.4%) (20.7%) (4.5%) (5.0%)
L3-1 45.9 7.6 264 218 2560 32.7 1.03 Tension
L3-2 47.2 7.6 287 228 2282 33.3 1.07 Tension
L3-3 45.2 8.9 275 231 2972 32.3 1.01 Tension
L3-4 42.3 6.9 275 228 2729 30.2 0.94 Tension
L3-5 50.2 8.1 295 241 2723 35.6 1.13 Tension
L3-6 52.4 7.5 268 236 3986 37.4 1.17 Shear
47.2 7.8 277 231 2875 33.6 1.06
Mean L3 –
(7.7%) (8.6%) (4.2%) (3.4%) (20.5%) (7.5%) (7.9%)
L4-1 48.9 9.3 363 309 5393 34.6 1.12 Tension
L4-2 80.3 9.9 367 336 3125 56.8 1.85 Tension
L4-3 69.5 9.4 317 295 4652 49.3 1.59 Tension
L4-4 84.2 9.2 352 316 3764 59.9 1.91 Tension
L4-5 70.2 8.5 386 336 4637 49.3 1.61 Tension
L4-6 71.6 9.2 322 310 3445 50.7 1.62 Tension
70.8 9.3 351 317 7964 50.1 1.62
Mean L4 –
(17.3%) (4.8%) (7.7%) (5.2%) (49.1%) (17.5%) (17.2%)
466 (CoV in parentheses)

467

24
468
469
470 Table 5: Short-Span Shear Test Experimental Results
max
471 (MC = moisture content; f v = maximum shear stress)

Failure
MC f vmax Failure
Specimen Load
(%) (MPa) Mode
(kN)
L1-1 79.8 8.3 1.76 Shear
L1-2 88.1 8.2 1.92 Shear
L1-3 100.4 8.0 2.19 Shear
L1-4 95.0 8.3 2.11 Shear
L1-5 104.8 8.3 2.34 Shear
L1-6 84.1 7.9 1.85 Shear
92.0 8.2 2.03
Mean L1 –
(10.6%) (2.1%) (10.9%)
L2-1 115.2 6.3 2.59 Tension
L2-2 117.9 6.5 2.68 Shear
L2-3 120.4 6.4 2.72 Tension
L2-4 117.9 6.7 2.66 Tension
L2-5 106.5 6.2 2.41 Tension
L2-6 115.4 6.3 2.66 Shear
115.5 6.4 2.61
Mean L2 –
(4.2%) (2.9%) (4.3%)
L4-1 119.0 7.2 2.73 Shear
L4-2 130.5 6.7 2.96 Shear
L4-3 109.2 7.6 2.51 Shear
L4-4 144.0 8.1 3.25 Shear
L4-5 147.2 7.7 3.31 Shear
L4-6 142.4 7.7 3.04 Shear
132 7.5 2.96
Mean L4 –
(11.6%) (6.2%) (10.3%)
472 (CoV in parentheses)

473
474

25
Figure
Click here to download Figure: CBM_05_2017_figures.docx

Figure 1: Long-Span Four-Point Flexure Test

Figure 2: Short-Span Three-Point Shear Test
Figure 3: Typical Load-Displacement Response
Figure 4: Typical Flexural Tension Failure Observed in Long-Span Tests
Figure 5: Typical Shear Failure Observed in Long-Span Tests
Figure 6: Typical Shear Failure Observed in Short-Span Tests of L1 Specimens
Figure 7: Typical Shear Failure Observed in Short-Span Tests of L4 Specimens