Investigation of the Temperature-Dependent

Investigation of the Temperature-Dependent Mechanical Behavior of a Polypropylene-Pine Composite Andrew J. Schildmeyer1; Michael P. Wolcott2; and Donald A. Bender, P.E., M.ASCE3 Abstract: Wood-plastic composites have been recognized as versatile and practical materials for use in many light-structural uses. Recently, more structurally demanding applications have surfaced, which require an improved understanding of mechanical performance and design methodologies. Research addressing the influence of service temperature on mechanical performance with the goal of assigning structural design values is lacking. This study examines the effect of temperature on the mechanical performance of a polypropylene-pine composite formulation. Static tests were performed at temperatures between 21.1 and 80.0° C to determine the material constitutive relations and ultimate properties in tension and compression. A statistical approach was proposed to assess design thermal loads according to geographical location. Both Young’s modulus and ultimate stress were found to decrease with temperature while maximum strain increased linearly with temperature. Temperature adjustment factors were developed over the range studied and were found to decrease properties by as much as 50% at the highest service temperatures. A simple thermal load methodology based on an ASHRAE standard was proposed for determining prevailing thermal conditions in design. DOI: 10.1061/共ASCE兲0899-1561共2009兲21:9共460兲 CE Database subject headings: Composite materials; Fiber reinforced plastics; Temperature effects; Wood; Polymers.

Introduction Wood-plastic composites 共WPC兲 are a class of materials made primarily from synthetic thermoplastics and natural fibers. Currently, WPC materials are primarily used in applications with low structural demands, such as automobile substrates, residential decking, deck rails, and windows 共Clemons 2002兲. Water resistance, durability, and low maintenance are intrinsic features emphasized and have driven WPCs to applications where less emphasis is given to structural performance. Inherent beneficial qualities combined with the prospect of improved structural function in strength, material stiffness, and long-term performance give WPCs significant advantages over traditional materials in selected applications where structural performance is the primary focus. To address the market expansion into more demanding structural applications, a testing and evaluation process to ascertain allowable design values must be established. However, because of the variability among WPC formulations, it has been difficult to fully characterize the behavior of the entire material class. Specific formulations of WPCs vary in thermoplastic and wood type, as well as various polymeric and inorganic additives. The most 1

Design Engineer, PCS Structural Solutions, 950 Pacific Ave., Ste 1100, Tacoma, WA 98402. E-mail: [email protected] 2 Louisiana-Pacific Professor, Wood Materials and Engineering Laboratory, Washington State Univ., Pullman, WA 99164-1806 共corresponding author兲. E-mail: [email protected] 3 Weyerhaeuser Professor and Director, Wood Materials and Engineering Laboratory, Washington State Univ., Pullman, WA 99164-1806. E-mail: [email protected] Note. This manuscript was submitted on March 6, 2007; approved on February 18, 2009; published online on August 14, 2009. Discussion period open until February 1, 2010; separate discussions must be submitted for individual papers. This paper is part of the Journal of Materials in Civil Engineering, Vol. 21, No. 9, September 1, 2009. ©ASCE, ISSN 0899-1561/2009/9-460–466/$25.00.

commonly used thermoplastics are polyethylene, polyvinyl chloride, and polypropylene 共PP兲, whereas frequently used wood species include pine, maple, and oak 共Wolcott 2001; Clemons 2002兲. Previous research has addressed the mechanical performance of different formulations due to the components included 共i.e., polymer type, wood species, wood form, coupling agents, etc.兲, as well as a comparison of performance in different loading modes 共i.e., tension, compression, flexure, and dowel bearing兲共Haiar 2000; Slaughter 2004; Kobbe 2005兲. However, little data have been published to characterize the influence of temperature on the performance of WPCs. Use of dynamic mechanical thermal analysis is common to discern phase transitions in the thermoplastic composite 共Amash and Zugenmaier 1997; Bengtsson et al. 2005; Harper et al. 2009兲. This testing does provide data on temperature sensitivity for mechanical properties but the focus of the analysis is seldom in the temperature use range of the end product. In contrast, when developing adjustment factors for design properties, it is best to evaluate the static properties of the material over the range of temperatures that the material is expected to experience in use. Testing to develop this temperature-dependent performance, our formulation will provide a testing procedure by which other formulations can be tested and compared, perhaps allowing for a more general WPC characterization in the future. This research is focused on the influence of service temperature on the constitutive relations and ultimate performance of a PP-based WPC formulation. Elevated temperature performance is especially important for WPCs because the thermoplastic component exhibits strong temperature dependence for both static and time-dependent behaviors. The use of thermally stable fillers 共e.g., wood兲 within the plastic phase decreases the effect that temperature exerts upon a thermoplastic. Central to this work is establishing the degree to which elevated in-service temperatures influence strength and Young’s modulus 共E兲 performance. In support of developing a methodology to assess the temperature dependence of mechanical properties for WPCs, the specific objectives of this work are to:

460 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / SEPTEMBER 2009

Downloaded 09 Jan 2010 to 134.121.180.48. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

1. 2. 3.

Quantify the changes in ultimate stress, maximum strain and elastic modulus over a valid service temperature range. Use analytical models to describe constitutive relations for this material over a range of service temperatures. Develop temperature-dependent adjustment factors for design of WPC materials by evaluating the changes in mechanical performance with temperature.

Materials and Methods The WPC was manufactured using a formulation composed of 58.8% pine 共Pinus spp.兲 flour 共American Wood Fibers 6020兲, 33.8% PP 共Solvay HB9200兲, 4.0% talc 共Luzenac Nicron 403兲, 2.3% maleated PP 共MAPP, Honeywell 950P兲, and 1.0% lubricant 共Honeywell OP100兲 by mass 共Slaughter 2004兲. Commercial 60mesh pine flour was dried in a steam tube dryer to a moisture content of less than 2%. The formulation components were dry blended using a 1.2-m drum mixer in a series of 25-kg batches. An 86-mm conical counterrotating twin-screw extruder, operating between 5 to 12 rpm, was used to produce the sections from which specimens were obtained. Details on materials and extruder operating parameters are given by Schildmeyer 共2006兲. The formulated WPC material was extruded into a three-box cross section using a patented stranding die 关T. C. Laver, “Extruded synthetic wood composition and method for making same,” U.S. Patent No. 5,516,472 共1996兲兴. The section 共Fig. 1兲 had overall nominal dimensions of 45.7 by 165.1 mm with nominal wall and flange thicknesses of 10.2 mm. Test specimens were machined from 1.22-m lengths of this extruded profile. The dimensions of individual tension and compression specimens were measured using digital calipers. Mechanical Testing Mechanical testing was conducted to determine material properties at various temperatures using a 222-kN servohydraulic test frame with an environmental chamber mounted within the frame. Data were collected during testing by computer at a sampling rate of 2 Hz. Displacement over a 25.4-mm gauge length was measured using an extensometer. Applied loads were determined by 22.2- and 244.7-kN in-line load cells for tension and compression tests, respectively. A constant strain rate of 0.01 mm/mm was applied by a controlled crosshead displacement rate of 2.03 mm/ min for both tension and compression. These conditions followed the ASTM D 683 共1999兲and D 695 共1996兲 standards as specified in ASTM D 7031 共2004兲. The environmental chamber controlled test conditions to 65% relative humidity and 21.1, 30, 40, 50, 65.6, and 80° C within a tolerance of ⫾2 ° C during all tests. It was considered that the

Fig. 1. Triple box extrusion profile including nominal member dimensions, along with specimen location in cross section

mechanical behavior could potentially be affected by molecular rearrangement at higher temperature tests, thereby relaxing potential processing stresses in some higher-temperature conditions but not in others. This was addressed by conditioning all specimens at 65.6° C for 48 h prior to testing at any of the prescribed service conditions. This temperature was judged as an appropriate upper bound to the realistic service conditions. At each temperature level, 28 specimens were tested to facilitate estimation of 5% nonparametric tolerance limits at 75% confidence level if needed. The E of this material was determined using a secant modulus technique, applied between 5 and 10% of ultimate load 共Kobbe 2005兲. This procedure was adopted to maintain consistency in analyzing the nonlinear stress-strain behavior of this material. Tension Mechanical properties in tension were established by following procedures outlined in ASTM D 683 共1999兲 with the exceptions of the test temperature and conditioning procedures. Type III dogbone specimens were sampled from the top and bottom flanges of the three-box boards. During preparation, the flanges were cut and planed to ensure uniform thickness and eliminate surface defects. These planed flanges were then cut to the required dimensions and shaped to their final configuration using a guide and router. Compression Mechanical properties in compression were established by following procedures outlined in ASTM D 695 共1996兲, except for test temperatures, conditioning procedures, and specimen geometry. Previous experiments testing small-scale compressive specimens in direct accordance with ASTM D 695 standard have yielded unrepresentative values for full-scale specimen performance. For this reason, a single-box compression specimen 203-mm long was cut from the outer boxes of the three-box section. This specimen was produced by detaching the outer boxes at the two flanges and machining the cut edges until smooth. All specimen dimensions were measured using digital calipers and recorded. These dimensions were used in relevant calculations for section area. The nominal specimen dimensions were 45.7 by 61.0 mm with four 10.2-mm-thick walls.

Results and Discussion Static Mechanical Properties The temperature dependence of tension and compression properties 关ultimate stress 共␴ult兲, maximum strain 共␧max兲, and E兴 are presented in Figs. 2–4, respectively. In general, increases in service temperature resulted in decreased values for ␴ult and E while ␧max showed an increase, indicating a more ductile response. Representative mean curves were computed by averaging the load values at common strain levels from the 28 specimens in each loading condition. For both tension and compression, ␴ult decreased linearly with increasing temperature. At 21.1° C, the ultimate tensile strength of this material was found to be 18.14 MPa and decreased to 12.03 MPa at 80° C. A similar trend is found in compression, however the ␴ult at ambient temperature was approximately 4 times greater at 48.91 MPa. Again, as temperature increased to 80° C, ␴ult decreased linearly to 24.90 MPa. When examining ␧max values, the tension and compression

JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / SEPTEMBER 2009 / 461


Fig. 2. Relationship between maximum specimen stress and increasing temperature

Fig. 4. Relationship between secant elastic modulus with respect to temperature

gain differed in magnitude, with compression displaying a much more ductile response. For tension, the ␧max linearly doubled from a value of 0.00901 mm/mm at 21.1° C to 0.01925-mm/mm at 80° C. The compression strains were nearly 3.5 times larger than tension strains at ambient temperatures. The temperature dependence for ␧max in compression was more modest than the tensile trend, however, and remained nearly constant at 0.035 mm/mm regardless of temperature tested. A decreasing trend for E is expected due to decreasing strength and increasing ductility with respect to temperature. At ambient temperature E values for both tension and compression were nearly identical with values of 3,489 MPa and 3,447 MPa for tension and compression, respectively. In both modes of loading, the values of E decreased. In tension, E decreased according to a second-order function to a value of 1,593 MPa at 80° C. In compression, a linear decrease was found at 80° C with a value of 1,751 MPa. Summary statistics for ␴ult, ␧max, and E at each temperature level can be found in Table 1. For ␧max, the coefficients of variation 共COV兲 values were between 10 and 16% for tension and 10 and 22% for compression. There was slightly less scatter for the E

data where COVs ranged between 9 and 18.5% in tension and 7.5 to 13% in compression. The least scatter among the groups of data is found in ultimate strength. COV values for this were all below 10% with only one exception at 21.1° C in tension where the value is 11.2%. Constitutive Relations The nonlinear nature of this material requires a more complex relation than materials where linear proportionality exists. Separate works by Conway 共1967兲, Lockyear 共1999兲, Murphy 共2003兲, and Kobbe 共2005兲 have investigated expressions using hyperbolic functions to describe constitutive relations for nonlinear materials. Two possible constitutive relations could be appropriate for modeling this material, one using the arc-hyperbolic sine and the other using the hyperbolic tangent functions. Previous research on this specific formulation by Kobbe 共2005兲 has indicated that the arc-hyperbolic sine function with two curve fitting parameters 共a and b兲 most accurately represents the initial relation between stress 共␴兲 and strain 共␧兲 behavior

Table 1. Mean Values for Testing Parameters with respect to Loading Mode and Temperature Tension Temperature 共°C兲 21.1 30.0 40.0 50.0 65.6 80

Fig. 3. Relationship between maximum specimen strain and increasing temperature

Note:

␴ult MPa

␧max

Compression E MPa

␴ult MPa

␧max

18.14 0.0090 3,487.3 48.91 0.0315 共0.112兲共0.151兲共0.183兲共0.0525兲共0.213兲 17.49 0.0115 3,047.1 43.54 0.0343 共0.0941兲共0.157兲共0.177兲共0.0619兲共0.113兲 15.30 0.0128 2,367.1 40.08 0.0367 共0.0620兲共0.167兲共0.160兲共0.0699兲共0.174兲 14.36 0.0148 1,990.9 35.85 0.0366 共0.0763兲共0.111兲共0.162兲共0.0413兲共0.104兲 12.77 0.0173 1,794.0 29.47 0.0358 共0.0851兲共0.132兲共0.0956兲共0.0381兲共0.145兲 12.03 0.0193 1,593.9 24.90 0.0357 共0.0804兲共0.108兲共0.102兲共0.0263兲共0.119兲 Coefficients of variation values are given in parentheses.



E MPa 3,434.9 共0.127兲 3,074.4 共0.106兲 2,787.5 共0.0776兲 2,474.4 共0.0808兲 2,059.2 共0.0964兲 1,755.7 共0.0904兲

Table 2. Values for the Parameters a and b for the Constitutive Relation Modeling at Each Temperature for Each Mode of Loading Tension Temperature 共°C兲 21.1 30.0 40.0 50.0 65.6 80.0

Compression

a 共MPa兲

b

a 共MPa兲

b

9.35 8.20 6.79 6.20 4.90 4.62

387.91 402.63 400.54 383.82 468.80 441.21

19.12 16.02 14.59 12.06 9.38 7.66

213.93 236.87 242.91 292.53 351.25 385.85

␴ = a · a sinh共b · ␧兲

共1兲

Values for the constants a and b were determined for mean stress-strain curves at all temperatures and are presented in Table 2. Values for these constants were determined by minimizing the residual sum of the squares between the predicted and experimental data. Clear trends for these empirical parameters can be seen in Figs. 5 and 6 showing a decreasing trend for a and an increasing trend for b with increasing temperatures. Because of the wide range of temperatures tested, the likely in-service high temperature condition of any end-user application should fall within this range. Therefore, appropriate constitutive equations can be interpolated from this data to arrive at reasonable predictions. The quality of fit can be judged in Figs. 7 and 8, where the constitutive relations are plotted against experimental data. Development of Temperature Design Factor Coefficient Temperature effects for civil engineering materials other than timber provide little guidance for expected service temperatures. For steel, concrete, and masonry design, no reductions are proposed for the temperature range studied in this work 共Salmon and Johnson 1996; MacGregor 1997兲. Decreases for steel and concrete are only assessed at much higher temperatures, i.e., fire resistance and welding considerations. Since timber construction is the only material widely used within the civil engineering community with temperature-dependent strength and WPC products

Fig. 5. Curve fitting constitutive relation parameter, a, with respect to temperature

Fig. 6. Curve fitting constitutive relation parameter, b, with respect to temperature

are likely to be used as replacements for timber components, an approach similar to the timber design code is logical. For this reason, the same ranges for temperature factor limits are considered for WPCs as in the timber code 关AF&PA 2005兴: T ⬍ 37.8° C, 37.8° C ⬍ T ⬍ 51.7° C, and 51.7° C ⬍ T ⬍ 65.6° C. To arrive at an appropriate allowable design value for a given loading property 共Fx兲, the mean value of the property must be first adjusted for variability to arrive at a characteristic value 共B兲, that is then further modified by a series of adjustment factors 共Ci兲, which account for service conditions that differ from testing conditions. The following equation has previously been proposed for WPC materials and is similar in form to the timber code method 共AF&PA 2005兲: Fallowable = B · Ca · Ct · Cm · Cv · Cd

共2兲

where B represents the characteristic allowable property modified for variability, C indicates various property adjustment factors, and subscripts a, t, m, v, and d represent adjustments for safety, temperature, moisture, volume, and load duration, respectively

Fig. 7. Mean tensile testing curves 共solid lines兲 for each temperature level along with the fit curves using the arc-hyperbolic sine model 共dashed, gray lines兲



Fig. 8. Mean compressive testing curves 共solid lines兲 for each temperature level along with the fit curves using the arc-hyperbolic sine model 共dashed, gray lines兲

Since the characteristic allowable property is taken at ambient temperature, the values for Ct were determined from the experimental data by normalizing each property of interest 共X兲 by the mean value for that property at ambient temperature 共T = 21.1° C兲 Ct =

X共T兲 X共21.1兲

共3兲

From the strength and E trends found in this investigation, an equation to calculate specific Ct factors for specific thermal loads has been found 共Figs. 9 and 10兲. This equation takes a quadratic polynomial form to encompass second-order effects that are found in material stiffness degradation Ct = 1 − ␤1共⌬T兲 − ␤2共⌬T兲2

共4兲

where Ct = temperature adjustment for a specific thermal load; ␤1,2 = empirical coefficients 共Table 3兲; and ⌬T = difference in thermal load temperature from ambient 共21.1° C兲. Eq. 共4兲 can be applied to calculate reduction factors for both strength and E. Furthermore, investigating the reduction in these

Fig. 10. Temperature adjustment factor with change from ambient temperature 共material stiffness adjustment兲

properties indicated unique decreases in properties between tension and compression. Therefore, different empirical factors should be applied to calculate the thermal reduction parameter depending on the mode of loading that a member will experience. Table 3 contains the empirical ␤-coefficients that apply to Eq. 共4兲 above. A more simplified approach can also be employed to determine the Ct factor for these materials. This approach considers temperature ranges over which a single reduction factor is calculated. The values of Ct have been calculated for the same temperature intervals as the timber code 共AF&PA 2005兲 and are presented in Table 4. It is important to mention that this approach is always conservative by converting the prevailing temperatures to the low end of the interval. Considerations for Thermal Loads Reductions in WPC strength and material stiffness with increased temperature are more severe than for timber because stronger temperature dependence exists. Furthermore, because the loss in strength and E for timber at high temperatures is often offset by corresponding gains from lower moisture contents 共Breyer et al. Table 3. Summary of ␤-Coefficients for the Equation Form to Calculate Ct ␴ult Tension ␤1 ␤2 R2

6.329⫻ 10 0 0.961

E Compression

−3

8.755⫻ 10 0 0.986

−3

Tension

Compression −2

1.918⫻ 10 −1.719⫻ 10−4 0.989

8.807⫻ 10−3 0 0.983

Table 4. Summary of Recommended Temperature Adjustment Factors for ␴ult and E Temperature °C

Fig. 9. Temperature adjustment factor with change from ambient temperature 共stress adjustments兲

T ⬍ 37.8 37.8⬍ T ⬍ 51.7 51.7⬍ T ⬍ 65.6 T ⬎ 65.6

␴ult

E

Tension

Compression

Tension

Compression

0.80 0.80 0.70 0.65

0.80 0.70 0.60 0.50

0.70 0.60 0.50 0.40

0.80 0.70 0.60 0.50



1999; AF&PA 2005兲. Application of temperature factors in timber design focuses upon sustained long-term elevated temperatures upon a member which causes degradation in the hemicellulose components and is nonrecoverable. A different approach is appropriate for WPCs because strength and material stiffness degradation occur quickly and is recoverable. Mechanical performance decreases with temperature in WPCs because of a softening of the thermoplastic matrix, rather than chemical degradation as in the case of timber. Strength and E of WPCs is recoverable then as the temperature decreases and the polymer matrix hardens. Transient periods of high temperature may or may not coincide with the highest load demands on a given structure. It is important then to address what temperature should be considered as design level temperature 共Fig. 10兲. Structural loads are often determined on a 50-year recurrence interval 共e.g., ground snow loads and design wind speeds兲. A similar approach could be used to find a thermal load for WPCbased geographical location. A good source for data are the ASHRAE Fundamentals Handbook 共2005兲. This source includes extreme drybulb temperatures at any desired return period based on the assumption that annual maxima and minima are distributed according to the Gumbel Type I extreme value probability distribution. For example, the 50-year extreme maximum temperatures for Phoenix, Arizona and Seattle, Washington are 49.2 and 39.4° C, respectively. In addition, the ASHRAE fundamentals handbook gives methodologies for computing surface temperatures due to incident solar radiation. Climate data are presented for hundreds of cities in the United States and Canada, as well as for selected cities worldwide. Not all locations in the United States are included in ASHRAE tables. In these cases, it is recommended to contact a local source for information on extreme temperatures as is done for ground snow loads and design wind speeds. Appropriate engineering judgment must be employed when choosing coincident design temperatures and structural loads. At extreme high temperatures when solar incident heating would occur, oftentimes the application may not be subjected to full design loads. The issues of transient solar incident heating and assessing nonuniform thermal gradients within a cross section are additional areas that require engineering judgment. A designer must consider the probability of experiencing both maximum solar incident heating and full design load. Furthermore, in applications such as decking, to achieve a full design load on the deck would effectively shade the decking material and reduce solar heat loading. Considerations for Implementation Final implementation of temperature factors should include some consideration for the mode of yielding for the designed member. When members are subjected to only compressive loads, factors based on the compressive testing here would recommend values of 0.80, 0.70, 0.60, and 0.50 for the respective temperature ranges reported earlier. This would apply to situations like braced deck foundation columns where only compressive axial forces are of interest. Similarly, calculations for buckling stress and other performance issues relating to stiffness should be checked using a reduced E to ensure adequate stiffness during high temperature conditions. If the mode of loading is a flexural or pure tensile application, the ␴ult temperature factors should more closely resemble the tensile factors of 0.80, 0.80, 0.70, and 0.60. At a “design level” stress, a flexural member would be designed such that maximum

tensile and compressive stresses are less than or equal to approximately 40% of their respective ultimate strengths. At this loading level, compressive and tensile behavior should be roughly equal in magnitude 共Kobbe 2005兲. However, the disparity between compressive and tensile capacities is such that a load level equal to the maximum design level stress in tension 共40% of ultimate strength兲 would only be 15% of the compressive capacity on the opposite extreme fibers in the member. Thus, failures will initiate in the tensile face of flexural members 共Kobbe 2005兲. The factors developed for the tensile mode would then be the appropriate mode for flexure. Applying the more conservative 共smaller in magnitude兲 compressive factors would create an overly conservative case because of the disparity in ultimate strengths between tension and compression.

Summary and Conclusions Static tests were performed at temperatures between 21.1 and 80.0° C to determine the material constitutive relations and ultimate properties in tension and compression. Different magnitudes of ␴ult were measured for tension and compression but the trend for both loading modes with respect to increasing temperature both decrease linearly. Tension loadings for ␧max increased appreciably with temperature, whereas a nearly constant maximum strain was determined for compression regardless of temperature. Both tension and compression E was approximately 3.5 GPa at ambient temperatures and decreased by nearly half at 80° C. A linear inverse relationship of mechanical performance with temperature was found for compression and a quadratic relationship for tension. An arc-hyperbolic sine function was found to adequately describe the nonlinear constitutive relation for this material. A method for determining thermal loads based on geographical location was proposed and sample calculations were presented. This method is based on historical climactic data in the ASHRAE handbook and uses a 50-year recurrence interval similar to other structural loads. Engineering judgment and consideration of competing factors 共i.e., reduced live loads at high temperatures and issues regarding solar incident heating兲 should be taken into account by the designer for appropriate use of these materials. Based on the test results and trends over this temperature range, adjustment factors were proposed at temperature levels similar to timber design for both ultimate stress and material stiffness in tension and compression. Considering ultimate stresses from ambient temperature to 37.5° C, a reduction of 0.80 is appropriate for both tension and compression members. Correction factors for the next temperature range, 37.8– 51.7° C, were determined to be 0.80 and 0.70 for tension and compression, respectively. The next range found factors of 0.70 for tension and 0.60 for compression and applies between 51.7 and 65.6° C. Above 65.6° C, a factor of 0.60 for tension and 0.50 for compression will appropriately reduce the allowable stress allowed on a member to account for the reduction in strength at the elevated temperature condition. Reductions in E which are similar to those in ␴ult are also proposed for the same temperature levels to address decreases in E. From ambient to 37.8° C, factors of 0.70 and 0.80 were determined for tension and compression, respectively. Above that, for temperatures between 37.8 and 51.7° C, reduction values of 0.50 and 0.60 apply for tension and compression. Lastly, at tempera-



tures greater than 65.6° C, factors of 0.40 and 0.50 are recommended to adequately reduce E.

References Amash, A., and Zugenmaier, P. 共1997兲. “Thermal and dynamic mechanical investigations on fiber-reinforced polypropylene composites.” J. Appl. Polym. Sci., 63共9兲, 1143–1154. American Forest & Paper Association 共AF&PA兲. 共2005兲. National design specification for wood construction, AF&PA, Washington, D.C. American Society of Heating, Refrigerating and Air-Conditioning Engineers 共ASHRAE兲. 共2005兲. “Section 28: Climatic design information.” Handbook—Fundamentals, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc., Atlanta. ASTM. 共1996兲. “Standard test methods for compressive properties of rigid plastics.” ASTM D 695–96, West Conshohocken, Pa. ASTM. 共1999兲. “Standard test methods for tensile properties of plastics.” ASTM D 683–99, West Conshohocken, Pa. ASTM. 共2004兲. “Standard guide for evaluating mechanical and physical properties of wood-plastic composite products.” ASTM D 7031, West Conshohocken, Pa. Bengtsson, M., Gatenholm, P., and Oksman, K. 共2005兲. “The effect of crosslinking on the properties of polyethylene/wood flour composites.” Compos. Sci. Technol., 65共10兲, 1468–1479. Breyer, D. E., Fridley, K. J., and Cobeen, K. E. 共1999兲. Design of wood structures ASD, 4th Ed., McGraw-Hill, New York, 4–45. Clemons, C. 共2002兲. “Wood-plastic composites in the United States, the interfacing of two industries.” For. Prod. J., 52共6兲, 10–18.

Conway, J. B. 共1967兲. Numerical methods for creep and rupture analyses, Gordon and Breach, New York. Haiar, K. J. 共2000兲. “Performance and design of prototype wood-plastic composite sections.” MS thesis, Washington State Univ., Pullman, Wash. Harper, D. P., Laborie, M. P. G., and Wolcott, M. P. 共2009兲. “The impact of polypropylene-graft-maleic anhydride on the crystallization and dynamic mechanical properties of isotactic polypropylene.” J. Appl. Polym. Sci., 111共2兲, 753–758. Kobbe, R. G. 共2005兲. “Creep behavior of wood-polypropylene composites.” MS thesis, Washington State Univ., Pullman, Wash. Lockyear, S. A. 共1999兲. “Mechanical analysis of transversely loaded wood/plastic sections.” MS thesis, Washington State Univ., Pullman, Wash. MacGregor, J. G. 共1997兲. Reinforced concrete: Mechanics and design, 3rd Ed., Prentice-Hall, Upper Saddle River, N.J., 74–75. Murphy, J. F. 共2003兲. Characterization of nonlinear materials, USDA Forest Products Laboratory Technical Note, Madison, Wis. Salmon, C. G., and Johnson, J. E. 共1996兲. Steel structures: Design and behavior, 4th Ed., Prentice-Hall, Upper Saddle River, N.J., 55–56. Schildmeyer, A. J. 共2006兲. “Temperature and time dependent behaviors of a wood-polypropylene composite.” MS thesis, Washington State Univ., Pullman, Wash. Slaughter, A. E. 共2004兲. “Design and fatigue of a structural wood-plastic composite.” MS thesis, Washington State Univ., Pullman, Wash. Wolcott, M. P. 共2001兲. “Wood-Plastic composites.” Encyclopedia of materials: Science and technology, Buschow et al., eds., Elsevier Science, New York, 9759–9763.

