Minimizing the creep of recycled polypropylene/rubberwood flour composites with mixture design experiments

Abstract

Composites of rubberwood flour (RWF) and recycled polypropylene (rPP) were produced into panel samples by using a twin-screw extruder. The effects on creep behavior of mixture fractions of rPP, RWF, maleic anhydride-grafted polypropylene (MAPP), and ultraviolet (UV) stabilizer were studied in a D-optimal mixture design. Creep was significantly affected by the composition. Increasing the fraction of RWF decreased creep, while MAPP and UV stabilizer increased it. The models fitted were used to optimize a desirability score that balanced multiple creep characteristics. The model-based optimal formulation 50.5 wt% rPP, 44.9 wt% RWF, 3.5 wt% MAPP, 0.1 wt% UV stabilizer, and 1.0 wt% lubricant was experimentally validated to have low creep closely matching the model predictions.

Keywords

Wood-plastic composites recycled polypropylene rubberwood flour mixture experimental design creep behavior

Introduction

Wood-plastic composites (WPCs) have been extensively developed and used in non-structural applications.¹ For example, WPCs are increasingly used to replace softwood lumber in deck building, to improve durability.^2,3 The advantages of WPCs include high specific strength and stiffness, resistance to water absorption, and positive impact on environmental issues. These have stimulated the development of WPC materials for also structural applications.^1,4,5 However, these composites are poorly suited for some applications due to long-term creep under loading. This study is aimed to evaluate and improve the creep characteristics of specific WPCs.

Waste materials locally available in southern Thailand were used as raw materials because of both environmental benefits and low cost. Rubberwood (Hevea brasiliensis) waste is mainly produced by sawmills and furniture industry, both prevalent in southern Thailand. Of their total wood intakes, these industries generally generate about 34% wood wastes and about 54% rejects of plantation wastes, while only 12% of the rubberwood ends up in the products.⁶ Most of the wood waste can be used in medium-density fiberboard and particle board.⁷ However, the use of wood waste as reinforcement in plastic composites is of great interest, with environmental and economic benefits. The advantages of wood particles include biodegradability, low health hazard during handling, and non-abrasive nature,^8,9 when substituted for synthetic fillers such as glass fiber, carbon fiber, and other inorganic fillers. In addition, plastic waste is one of the major constituents of global municipal solid waste.¹⁰ In 2008, at least 33.6 million tons post-consumer plastics were generated in the USA, of which 28.9 million tons went to landfills, 2.6 million tons to combustion and energy recovery, and only 2.2 million tons to recycling^11,12—only a tiny fraction of plastic waste is recycled. Blending post-consumer plastics with wood flour to create value-added products could increase the value of plastic waste and impact its reuse practices.¹¹ Plastic waste is a promising raw material for WPCs because of low cost¹⁰ and properties similar to virgin materials. For example, composites made from recycled high-density polyethylene (rHDPE) have similar or, in some cases, better mechanical properties than composites made from virgin HDPE.^13,14 The mechanical properties of composites are not better with virgin polystyrene than with recycled polystyrene.¹⁵ Ashori¹⁶ studied the potential of municipal solid waste materials for making wood plastic composites. Waste wood and paper can replace inorganic fillers in thermoplastic composites, and these composites can be reclaimed and recycled repeatedly. Ashori and Sheshmani¹⁷ made hybrid composite materials with a combination of recycled newspaper fiber (RNF) and poplar wood flour as reinforcement, and with recycled polypropylene (rPP) as the polymer matrix. They found that the composites with a high fraction of RNF had high water absorption. Madhoushi et al.¹⁸ studied the effects of sanding dust loading and nanoclay content on the physical and mechanical properties of polypropylene. Addition of sanding dust significantly decreased tensile and flexural properties of the composites, and flexural, tensile, and withdrawal strength of fasteners were improved by the addition of 2 wt% nanoclay in the matrix. Nourbakhsh and Ashori¹⁹ evaluated the effects of the fiber content and compatibilizing agent concentration on the mechanical properties and water absorption of composites from poplar fiber and rHDPE. The compatibilizer polyethylene-grafted maleic anhydride improved the flexural properties that now increased with wood content. In another study, no statistically significant differences were found in mechanical properties of composites, on comparing recycled plastics (HDPE and polypropylene) with virgin plastics.²⁰ Polypropylene waste and wood waste are promising alternative raw materials for making low cost WPCs.²¹ To reduce solid waste disposal in landfills and have low cost WPC products¹³ with good mechanical properties and low creep deformation, suitable WPC formulations need to be developed.

Design of experiments contributes to efficiently finding the best formulations. Typical designs include Taguchi method, factorial design, and mixture design.²² The fractions of components in a mixture cannot be changed independently because they must add up to 100%, and mixture designs make use of this fact.²² A D-optimal mixture experimental design allows to fit models that can be used to optimize the formulation of a composite material.²³ It also allows placing restrictions on the formulations, such as lower or upper limits on the fractions of some components.^23,24 Mixture designs have recently been applied in food and pharmaceutical industries to find optimal formulations because they appear efficient in providing useful models with a comparatively small number of experiments. However, prior studies on WPCs seem not to have used D-optimal mixture designs. A four-factor central composite design was applied to develop a response surface model and to study the foamability of rigid PVC/wood-flour composites.²⁵ A 2⁴ factorial design was used to determine the effects of two hindered amine light stabilizers (HALS), a colorant, an ultraviolet absorber, and their interactions, on the photostabilization of wood flour/HDPE composites.²⁶ A Box-Behnken design with response surface method was adopted to determine which variables influenced board performance significantly.²⁷ In the current study, a D-optimal mixture design was used to model the creep of WPCs. The ultimate goal of this work was to optimize the composite formulation using rPP and rubberwood flour (RWF) for minimal creep.

Materials and methods

Materials

rPP pellets, with a melt flow index of 11 g/10 min at 230℃, were purchased from Withaya Intertrade Co., Ltd (Samutprakarn, Thailand). RWF, used as a natural reinforcement, was collected from a local furniture factory (Songkhla, Thailand). Its chemical composition (by dry weight) was cellulose 39%; hemicellulose 29%; lignin 28%; and ash 4%.⁶ The interfacial bonding between wood flour filler and polymer matrix was also modified, using maleic anhydride-grafted polypropylene (MAPP) with 8–10% of maleic anhydride, supplied by Sigma-Aldrich (Missouri, USA). HALS additive under the trade name MEUV008, chosen as the ultraviolet (UV) stabilizer, was supplied by TH Color Co., Ltd (Samutprakarn, Thailand). Paraffin wax, chosen as the lubricant (Lub), was purchased from Nippon Seiro Co., Ltd (Yamaguchi, Japan).

Experimental design to optimize formulation

The D-optimal design of mixture experiments was created with Design-Expert software (version 8.0.6, Stat-Ease, Inc.) to statistically evaluate and model the effects of component fractions on creep properties and to optimize the formulation. The optimal experimental design of WPC formulations specified the component fractions of rPP (x₁), RWF (x₂), MAPP (x₃), UV (x₄), and Lub (x₅). The upper and lower limits of experimental range for the compositions are shown in Table 1. Despite the fraction of Lub being held constant, it is included as a variable because it contributes to the 100% in the mixture. The total number of runs was 20, as shown in Table 2, including 15 different formulations and five duplications to evaluate reproducibility or variances.

Table 1.

Constraints for the mixture design of experiments.

Component	Proportion restriction (wt%)
rPP (x₁)	50 ≤ x₁ ≤ 70
RWF (x₂)	25 ≤ x₂ ≤ 45
MAPP (x₃)	3 ≤ x₃ ≤ 5
UV stabilizer (x₄)	0 ≤ x₄ ≤ 1
Lub (x₅)	=1

Table 2.

Experimental compositions in mixture experimental design and measured responses.

Run no.	Mixture proportion (wt%)					Creep strain (%)
Run no.	x ₁	x ₂	x ₃	x ₄	x ₅	C_e	C_ve6000	C_t6000
1	63.9	29.9	4.5	0.7	1.0	0.98	0.33	1.31
2	70.0	25.0	3.0	1.0	1.0	1.03	0.42	1.45
3	50.0	43.0	5.0	1.0	1.0	0.77	0.27	1.04
4	54.9	38.9	4.5	0.7	1.0	0.75	0.25	1.00
5	59.5	34.5	5.0	0.0	1.0	0.80	0.37	1.17
6	55.4	39.9	3.5	0.2	1.0	0.74	0.29	1.03
7	59.5	34.5	4.0	1.0	1.0	0.78	0.31	1.09
8^a	59.5	34.5	5.0	0.0	1.0	0.89	0.38	1.27
9	50.0	44.3	4.3	0.5	1.0	0.67	0.27	0.94
10	68.0	25.0	5.0	1.0	1.0	1.07	0.43	1.50
11	50.0	45.0	3.0	1.0	1.0	0.70	0.25	0.95
12^a	50.0	43.0	5.0	1.0	1.0	0.71	0.36	1.07
13	60.3	35.3	3.0	0.5	1.0	0.88	0.34	1.22
14	64.9	30.4	3.5	0.2	1.0	0.90	0.38	1.28
15^a	70.0	25.0	3.0	1.0	1.0	1.03	0.40	1.43
16	51.0	45.0	3.0	0.0	1.0	0.70	0.32	1.02
17^a	51.0	45.0	3.0	0.0	1.0	0.69	0.26	0.95
18^a	50.0	45.0	3.0	1.0	1.0	0.73	0.25	0.98
19	70.0	25.0	4.0	0.0	1.0	0.94	0.42	1.36
20	69.0	25.0	5.0	0.0	1.0	0.96	0.33	1.29

Duplicate experiment.

Composites processing

Before compounding, the RWF was sieved through an 80 mesh standard sieve (particles smaller than 180 µm pass) and dried in an oven at 110℃ for 8 h to minimize moisture content. WPCs were then produced in a two-stage process. In the first stage, WPC pellets were produced: rPP and RWF were dry-blended, and then melt-blended into wood-plastic composite pellets using a twin-screw extruder machine (Model SHJ-36 from En Mach Co., Ltd, Nonthaburi, Thailand). The extrusion barrel with 10 temperature zones was controlled at 130–170℃ to avoid degradation of the components, while the screw rotating speed was maintained at 70 rpm. The extruded strand passed through a water bath and was subsequently pelletized. In the second stage, WPC panels were produced: the WPC pellets were again dried at 110℃ for 8 h. WPC pellets, MAPP, UV stabilizer, and lubricant compositions indicated in Table 2 were then dry-mixed and added into the feeder of the twin-screw extruder. The temperature profile for extruding was 130–190℃, with 50 rpm screw feed. Melt pressure at the die varied between 0.10 and 0.20 MPa, depending on wood flour content. Vacuum venting at nine temperature zones was also used to purge volatile compounds. The samples were extruded through a rectangular 9 mm × 22 mm die and cooled in atmospheric air. The specimens were machined for flexural creep testing, following the standard of American Society for Testing and Materials (ASTM).

Characterization

Three-point bending creep tests of rPP/RWF composites were carried out on an Instron Universal Testing Machine (Model 5582 from Instron Corporation, MA, USA) in Figure 1, according to ASTM D2990 standard. All the tests were performed on 13 mm × 4.8 mm × 100 mm (width × thickness × length) rectangular samples, and a test span of 80 mm. Before the creep tests, the specimens were equilibrated for 15 min, and the tests were conducted at a temperature of 25℃ (ambient conditions). The total time of the testing was 100 min (6000 s) under a constant stress of 19 MPa. Five replications of each formulation were tested.

Morphological analysis

The formation of cracks and interfacial morphology between the wood flour and the polymeric matrix were analyzed with a scanning electron microscope (SEM). The fracture surface of the specimen before creep testing was fractured in liquid nitrogen. Likewise, the fracture surface of the specimen after creep testing was fractured by the creep characterization. SEM imaging was performed using a FEI Quanta 400 microscope (Oregon, USA) at an accelerating voltage of 20 kV. The samples (fracture surfaces) before and after creep tests were sputter-coated with gold to prevent electrical charging during the observation. Specimens were imaged at magnifications of 150× and 2500×.

Results and discussion

The D-optimal mixture design of experiments, with five fractions as (mutually dependent) variables (that sum to one), had 20 runs in a randomized order. The three determined responses were the values of the instantaneous creep strain (C_e), of the viscoelastic creep strain after 6000 s (C_ve6000), and of the total creep strain after 6000 s (C_t6000), and the results are summarized in Table 2.

Statistical analysis of the response surface model

The data for C_e, C_ve6000, and C_t6000 were fit with linear models by multiple linear regression, with no statistical need for quadratic, special cubic, and cubic models. For example, a summary of modeling the C_t6000 response is shown in Table 3. The sequentially fit linear model is significant (p-value less than α = 0.05), but the higher order terms are not. The adjusted coefficient of determination (adj-R²) and predicted coefficient of determination (pred-R²) shown in Table 4 have fairly good values at 0.8780 and 0.8413, respectively. The values in Table 4 came from an analysis of variance (ANOVA) on the significant effects relative to the creep responses. The ANOVA shows statistical significance of these linear models, indicated by p-values less than α (α = 0.05). This result implies that each modeled output, C_e, C_ve6000, and C_t6000, was significantly affected by at least one of the four controlled variables. The R² value for C_ve6000 is relatively poor, partly because its determination was “noisy” with a high CV.

Table 3.

Fit summary of C_t6000 response.

Source	Sequential p-value	Lack of fit p-value	Adj-R²	Pred-R²
Linear	<0.0001*	0.1125	0.8780	0.8413	Suggested
Quadratic	0.2092	0.1415	0.9045	0.7242
Special cubic	0.2454	0.1161	0.9279	−13.77
Cubic	0.1161	–	0.9497	–	Aliased

P < 0.05 indicates that model terms are significant.

Table 4.

P-values from analysis of variance and model adequacy indicators for each modeled response.

Source	C_e (%)	C_ve6000 (%)	C_t6000 (%)
Model	Linear	Linear	Linear
Model	<0.0001*	0.0006*	<0.0001*
Linear mixture	<0.0001*	0.0006*	<0.0001*
Lack of fit	0.3331	0.3590	0.1125
R ²	0.9143	0.6555	0.8973
Adj-R²	0.8982	0.5909	0.8780
Pred-R²	0.8742	0.4296	0.8413
CV (%)	4.99	12.03	5.52

P < 0.05 indicates that model terms are significant.

The R² values of the C_e, C_ve6000, and C_t6000 are 0.9143, 0.6555, and 0.8973, indicating that 8.57%, 34.45%, and 10.27%, respectively, of the total variability in observations is not explained by the models; R² values close to 1 indicate good fits.²⁸ R² values will always increase when a variable is added to the model,²⁹ and the computed adj-R² should be close to R² value of the model selected. This indeed is the case for the fitted models, indicating it is unlikely that the models have insignificant terms included.³⁰ The pred-R² value of C_e was 0.8742, meaning that the fitted model is estimated to explain about 87% of variability in new cases, and this is in reasonable agreement with the adj-R² of 0.8982. For C_ve6000 all of R², adj-R², and pred-R² have relatively low or poor values, because C_ve6000 was calculated as C_t6000−C_e and this increased its relative inaccuracy. The coefficients of variation (CV), of C_e, C_ve6000, and C_t6000 were estimated at 4.99%, 12.03%, and 5.52%, respectively, based on the residual variation. Low CV values indicate good precision of the determinations.

Model adequacy checking

Model adequacy checking is always necessary with a fitted model.³⁰ Figure 2(a) displays normal probability plots of the residuals for elastic creep strain C_e, and the visually good fit with a straight lines suggests the residuals are about normally distributed. The interpretation is that the residuals are Gaussian measurement noise, while the explanatory variables (fractions of rPP, RWF, MAPP, and UV stabilizer) explain the deterministic part of the relationship. Likewise, the presence of outliers is not strong indication, as such a failed experiment would give a large residual disabling the good straight line fit in a probability plot.²² A plot of the residuals vs. the predicted values for the model of C_e is shown in Figure 2(b). There is no obvious pattern remaining, and therefore no suggestion for adding some nonlinear terms to the fit.²² Figure 2(c) shows the C_e model predictions vs. observations. The model outputs fit the actual observations quite well, with C_e model deviating from actual by less than about 5%, in alignment with the estimated CV. The model adequacy was similarly checked for C_ve6000 and C_t6000, with essentially similar conclusions.

Effect of composition on the elastic creep strain, and optimal formulation

The linear regression model fitted to experimental C_e value was

C_{e} = 0.99 x_{1} + 0.66 x_{2} + 0.89 x_{3} + 1.78 x_{4}

(1)

in which all coefficients are positive. RWF (x₂) has the smallest coefficient so it should be maximized to minimize creep. The UV stabilizer (x₄) has the largest coefficient, so its addition should be as small as possible. The experimentally covered formulations are shown in Figure 3(a) and (b), with color coding for the modeled C_e. In the triangular contour plot of Figure 3(a), the three pure components (rPP, RWF, and MAPP) are represented by the corners, while the additive levels were fixed (UV stabilizer at 0.5 wt% and Lub at 1 wt%). The contours in the colored areas, that include the experimental observations, present the C_e regression fits varying from 0.7% to 1%. The creep C_e clearly decreases with increasing RWF content. High wood flour content increases the modulus of elasticity (MOE) of composites,³¹ so that higher stress is required for the same deformation.^7,32 The choice of MAPP content between 3 and 5 wt% barely affected the C_e. Generally, the addition of coupling agent in the WPCs decreases the creep strain due to the improved filler dispersion and improved interfacial adhesion between wood flour and polymer matrix,^33–35 whereas too much MAPP relative to wood flour will cause self-entanglement, resulting in slippage with the PP molecules.^7,36 This slippage in the WPC structures leads to easier creep deformation of the WPC specimens. The triangular contour plot in Figure 3(b) also shows that addition of 1 wt% UV stabilizer slightly increased the elastic creep strain from 0.82% to 0.88% because UV stabilizer has some chemical reactions with the other components.³⁷ Likewise, it is known to reduce the flexural properties but to increase creep strain, due to non-homogeneous spatial distribution of wood flour, polymer, and UV stabilizer.³⁸ Therefore, the load-bearing capacity of WPC specimens decreased, resulting in increase of the creep deformation. Using 1 wt% of UV stabilizer may be unnecessary, and its fraction should be minimized to minimize creep.

The numerically optimized compositions for each creep characteristic, based on fitted models, are shown in Table 5. In all three cases, the formulations have about 50.0 wt% of rPP and 45.0 wt% of RWF, with minor variation in MAPP and UV stabilizer fractions. Since the requirements of optimizing the different creep characteristics are not in much of a conflict, they can be approximately optimized simultaneously.

Table 5.

The optimal formulations that minimize each creep characteristic, with predicted responses.

Property (%)	Mixture proportion (wt%)					Predicted response	Desirability
Property (%)	rPP	RWF	MAPP	UV	Lub	Predicted response	Desirability
C_e	50.0	45.0	3.9	0.1	1.0	0.67	0.996
C_ve6000	50.0	45.0	3.0	1.0	1.0	0.26	0.906
C_t6000	50.1	45.0	3.5	0.3	1.0	0.95	0.962

Effect of composition on the viscoelastic creep strain, and optimal formulation

The linear regression model for the viscoelastic creep strain (C_ve6000) was

C_{ve 6000} = 0.40 x_{1} + 0.27 x_{2} + 0.45 x_{3} + 0.26 x_{4}

(2)

with positive coefficients. In Figure 4 C_ve6000 (in the range of 0.28 to 0.40%) increases for high fractions of rPP, because the mobility of polymer chains increased in the WPCs and contributed to viscoelasticity. The concentration effect of MAPP on C_ve6000 was insignificant, similar to elastic creep strain. The optimal composition minimizing viscoelastic creep strain coincided with formulation 11, see Table 5.

Effect of composition on the total creep strain, and optimal formulation

The linear regression fit for the total creep strain (C_t6000) was

C_{t 6000} = 1.39 x_{1} + 0.93 x_{2} + 1.33 x_{3} + 2.04 x_{4}

(3)

with positive coefficients. RWF (x₂) has the lowest coefficient, while UV stabilizer (x₄) had the largest coefficient. Figure 5(a) shows that total creep strain varies in range of 1.0–1.3% and decreases with RWF loading. RWF had a lower coefficient than rPP because wood flour is stiffer than the neat plastic.³⁹ Therefore, the deformation of the composites is reduced with increased wood flour content. Furthermore, the addition of increasing wood flour replaced the plastic matrix. It reduced the volume and mobility of the macromolecules chains in the composite structures, resulting in decrease of the creep deformation. The creep of WPCs is mainly caused from the plastic matrix, and the chains mobility leads to eventual failure of the WPC specimens. Figure 5(b) presents the effects of MAPP and UV stabilizer contents on the total creep strain. The total creep strain slightly increases with MAPP and UV stabilizer concentrations, with reasons similar to what was discussed in relation to C_e. The optimal formulation based on the numerical model is also shown in Table 5. In addition, the microstructure analysis of the composites before and after creep experiment was also observed from the SEM micrographs in Figure 6 (Figure 6(a) and (b) before creep testing and 6(c) and (d) after creep testing). Irregular short fibers are seen in the composites. The composites before creep testing exhibited no crack formation and good interfacial contact between RWF and PP matrix. According to this SEM study, the coupling agent used in the composites improves the compatibility, resulting in the good interfacial adhesion and enhancement of creep performance.⁷ In contrast, the composites after creep testing had large cracks and poor interfacial contact. This is due to great extension of the composites bearing load for long time, and the wood flour was pushed out from the plastic matrix, resulting in failure of the composites.

Optimal formulation for all creep characteristics

Multiobjective optimization using all of the regression models was performed with the Design-Expert software, using its default settings to construct a desirability score that balances all of the fitted models. The plot in Figure 7 shows the formulation that was considered optimal, along with contours of the desirability score. The optimal formulation found was 50.5 wt% rPP, 44.9 wt% RWF, 3.5 wt% MAPP, 0.1 wt% UV stabilizer, and 1.0 wt% Lub, corresponding to a high desirability of 0.945. All the previous optima, in Table 5, were at practically the same formulation. The model predictions were validated experimentally, and the results are given in Table 6 for the jointly optimal formulation. The maximum deviations between model predictions and experimental averages are of the same order as the earlier estimated CV accuracies of determinations.

Table 6.

Predicted and observed responses with the formulation optimized jointly for all the creep characteristics.

Mixture component proportion (wt%)					Creep strain (%)
rPP	RWF	MAPP	UV	Lub	C_e	C_ve6000	C_t6000
Predicted	50.5	44.9	3.5	0.1	1.0	0.67	0.27	0.95
Observed	50.5	44.9	3.5	0.1	1.0	0.71	0.29	1.00
(0.01)	(0.02)	(0.03)

Note: The values in parentheses are standard deviations from five replicates.

Conclusions

Design and analysis of D-optimal mixture experiments were used to efficiently obtain the optimal formulation of rPP/RWF composites that minimizes creep. All the component fractions experimentally varied, namely of rPP, RWF, MAPP, and UV stabilizer, which significantly affected all the creep characteristics (C_e, C_ve6000, and C_t6000). In general, a high fraction of RWF reduced all of these, and the optima found had 45 wt% RWF which was the maximum in the experimental design. At this wood flour loading, the modulus of elasticity was maximized, so that a comparatively high stress is required for a given creep deformation. Increasing the fraction of MAPP from 3 to 5 wt% only slightly affected the creep strain, lacking statistical significance. The addition of 1 wt% UV stabilizer slightly increased creep. The approximately optimal formulation minimizing jointly all creep characteristics was 50.5 wt% rPP, 44.9 wt% RWF, 3.5 wt% MAPP, 0.1 wt% UV stabilizer, and 1.0 wt% Lub. The joint optimization maximized a desirability score that balanced the multiple objectives, and the jointly optimal formulation was experimentally validated to produce low creep nearly as predicted.

Footnotes

Funding

The authors gratefully acknowledge the financial support from the graduate school of Prince of Songkla University, the Government budget Fund (Research Grant Code: 2555A11502062), and Rubberwood Technology and Management Research Group (ENG-54-27-11-0137-S) of Faculty of Engineering, Prince of Songkla University, Thailand.

Conflict of Interest

None declared.

Acknowledgement

We thank the Research and Development Office (RDO) and Assoc. Prof. Seppo Karrila for editing this article.

References

Tamrakar

Lopez-Anido

Kiziltas

. Time and temperature dependent response of a wood-polypropylene composite. Compos A: Appl Sci Manuf 2011; 42: 834–842.

Carroll

Stone

Sirignano

. Structural properties of recycled plastic/sawdust lumber decking planks. Resour Conserv Recycl 2011; 31: 241–251.

Ganguly

Eastin

. Trends in the US decking market: a national survey of deck and home builders. Forest Chroni 2009; 85: 82–90.

Klyosov

. Wood-plastic composites. 1, Hoboken, NJ: John Wiley & Son, Inc, 2007.

Ratanawilai

Lekanukit

Urapantamas

. Effect of rubberwood and palm oil content on the properties of wood-polyvinyl chloride composites. J Thermoplast Compos Mater 2012. DOI: 10.1177/0892705712454863.

Petchpradab

Yoshida

Charinpanitkul

. Hydrothermal pretreatment of rubber wood for the saccharification process. Ind Eng Chem Res 2009; 48: 4587–4591.

Homkhiew

Ratanawilai

Thongruang

. Composites from recycled polypropylene and rubberwood flour: effects of composition on mechanical properties. J Thermoplast Compos Mater 2013. DOI: 10.1177/0892705712475019.

Peng

Fan

Hartley

. Properties of natural fiber composites made by pultrusion process. J Compos Mater 2011; 46: 237–246.

Francucci

Rodrıguez

Vazquez

. Experimental study of the compaction response of jute fabrics in liquid composite molding processes. J Compos Mater 2011; 46: 155–167.

10.

Cui

Lee

Noruziaan

. Fabrication and interfacial modification of wood/recycled plastic composite materials. Compos A: Appl Sci Manuf 2008; 39: 655–661.

11.

Ratanawilai

Thanawattanasirikul

Homkhiew

. Mechanical and thermal properties of oil palm wood sawdust reinforced post-consumer polyethylene composites. ScienceAsia 2012; 38: 289–294.

12.

Themelis NJ, Castaldi MJ, Bhatti J, et al. Energy and economic value of nonrecycled plastics (NRP) and municipal solid wastes (MSW) that are currently landfilled in the fifty States. EEC Study of non-recycled plastics, Earth Engineering Center, Columbia University, 2011, p.8.

13.

Adhikary

Pang

Staiger

. Dimensional stability and mechanical behaviour of wood-plastic composites based recycled and virgin high-density polyethylene (HDPE). Compos B: Eng 2008; 39: 807–815.

14.

Selke

Wichman

. Wood fiber/polyolefin composites. Compos A: Appl Sci Manuf 2004; 35: 321–326.

15.

Lisperguer

Bustos

Saravia

. Thermal and mechanical properties of wood flour-polystyrene blends from postconsumer plastic waste. J Appl Polym Sci 2011; 119: 443–451.

16.

Ashori

. Municipal solid waste as a source of lignocellulosic fiber and plastic for composite industries. Polym Plas Technol Eng 2008; 47: 741–744.

17.

Ashori

Sheshmani

. Hybrid composites made from recycled materials: moisture absorption and thickness swelling behavior. Biores Technol 2010; 101: 4717–4720.

18.

Madhoushi

Chavooshi

Ashori

. Properties of wood plastic composite panels made from waste sanding dusts and nanoclay. J Compos Mater 2013; 48(14): 1661–1669.

19.

Nourbakhsh

Ashori

. Preparation and properties of wood plastic composites made of recycled high-density polyethylene. J Compos Mater 2009; 43: 877–883.

20.

Najafi

Hamidinia

Tajvidi

. Mechanical properties of composites from sawdust and recycled plastics. J Appl Polym Sci 2006; 100: 3641–3645.

21.

Nourbakhsh

Ashori

Tabari

. Mechanical and thermo-chemical properties of wood-flour/polypropylene blends. Polym Bull 2010; 65: 691–700.

22.

Montgomery

. Design and analysis of experiments, 7th ed. New York, NY: John Wiley & Sons, Inc, 2009.

23.

Khosrowshahi

Salem

. Influence of polyvinyl alcohol and carboxymethyl cellulose on the reliability of extruded ceramic body: application of mixture design method in fabricating reliable ceramic raschig rings. Inter J Appl Cera Technol 2011; 8: 1334–1343.

24.

John

RCS

. Experiments with mixtures, ill-conditioning, and ridge regression. J Qual Technol 1984; 16: 81–96.

25.

Matuana

Mengeloglu

. Manufacture of rigid PVC/wood-flour composite foams using moisture contained in wood as foaming agent. J Vinyl Addit Technol 2002; 8: 264–270.

26.

Stark

Matuana

. Ultraviolet weathering of photostabilized wood-flour-filled high-density polyethylene composites. J Appl Polym Sci 2003; 90: 2609–2617.

27.

Jun

Xiang-ming

Jian-min

. Optimization of processing variables in wood–rubber composite panel manufacturing technology. Biores Technol 2008; 99: 2384–2391.

28.

Amini

Younesi

Bahramifar

. Application of response surface methodology for optimization of lead biosorption in an aqueous solution by Aspergillus niger. J Hazard Mater 2008; 154: 694–702.

29.

Eren

Kaymak-Ertekin

. Optimization of osmotic dehydration of potato using response surface methodology. J Food Eng 2007; 79: 344–352.

30.

Myers

Montgomery

Anderson-Cook

. Response surface methodology: process and product optimization using designed experiments, 3rd ed. Hoboken, NJ: John Wiley & Sons, Inc, 2009.

31.

Marcovich

Villar

. Thermal and mechanical characterization of linear low-density polyethylene/wood flour composites. J Appl Polym Sci 2003; 90: 2775–2784.

32.

Rahman

Huque

Islam

. Improvement of physico-mechanical properties of jute fiber reinforced polypropylene composites by post-treatment. Compos A: Appl Sci Manuf 2008; 39: 1739–1747.

33.

Mingyin

Ping

Yongsheng

. Creep behaviour of wood flour/poly(vinyl chloride) composites. J Wuhan Univer Technol-Mater 2009; 24: 440–447.

34.

Bengtsson

Gatenholm

Oksman

. The effect of crosslinking on the properties of polyethylene/wood flour composites. Compos Sci Technol 2005; 65: 1468–1479.

35.

Nunez

Sturm

Kenny

. Mechanical characterization of polypropylene-wood flour composites. J Appl Polym Sci 2003; 88: 1420–1428.

36.

Mohanty

Verma

Nayak

. Influence of fiber treatment on the performance of sisal–polypropylene composites. J Appl Polym Sci 2004; 94: 1336–1345.

37.

Homkhiew

Ratanawilai

Thongruang

. The optimal formulation of recycled polypropylene/rubberwood flour composites from experiments with mixture design. Compos B: Eng 2014; 56: 350–357.

38.

Wechsler

Hiziroglu

. Some of the properties of wood–plastic composites. Build Environ 2007; 42: 2637–2644.

39.

Garcia

Hidalgo

Garmendia

. Wood–plastics composites with better fire retardancy and durability performance. Compos A: Appl Sci Manuf 2009; 40: 1772–1776.