OPTIMISATION STUDIES ON THE MECHANICAL PROPERTIES OF NEEM BARK FLOUR-FILLED HIGH DENSITY POLYETHYLENE COMPOSITES USING CENTRAL COMPOSITE DESIGN

Government Rabboni Mike. This research is carried out to study the effect of filler content and the particle size on the mechanical properties of neem bark flour-filled high density polyethylene composites. The fiber was characterized. Response surface methodology was used to optimize the preparation of the composite. A second order polynomial model was developed to predict the mechanical properties based on central composite design. The results show that the neem bark used for this study contains 9.04% lignin, 65.43% cellulose, 23.55% hemicellulose. It was found that the composite design was best fit for a quadratic regression model. The selected optimum condition for this composite is 21.63 mesh particle size and 40wt% filler content at 77.7% desirability. The mechanical properties at optimum are 14.51MPa, 5.28%, 0.45GPa, 92.31KJ/m, and 242.14Pa for tensile strength, percentage elongation, tensile modulus, impact strength and hardness respectively. The optimum conditions were validated with little error less than 0.2%.

with n-hexane and methanol. Then, 0.7g of the treated fiber was added in a 95% solution of nitric acid and ethanol mixture. The mixture was filtered and the residue washed first with hot water, then with ethanol to completely remove the residual acid. The residue was oven dried at 100 0 C to a constant weight. The cellulose corresponds to the insoluble fraction of the sample. The hemicellulose content of the plant fiber was determined by neutral detergent fiber method, [4]. Neutral detergent fiber was prepared by refluxing 0.7g of the fiber sample with sodium lauryl sulphate solution (neutralized to pH 7.0) for 1 hour. The solution was filtered and the residue washed with hot distilled water and ethanol. The residue was subsequently oven dried to constant weight at 100 0 C for 8 hours. The weight obtained is the neutral detergent fiber weight. Hemicellulose content was calculated as the difference in weight of neutral detergent fiber and the acid detergent fiber prepared from acid hydrolysis of the same mass of the sample. The lignin content was estimated according to Ververis et al. [5]. In determination of the lignin content, 0.7g of the fibers were heated with 5ml of 72% w/w H 2 SO 4 solution for 4.5 hours in order to hydrolyze the cellulose and hemicellulose. After the heating, the solution was filtered and the residue thoroughly washed with hot distilled water and absolute ethanol to completely remove any acid present. The solid residue was dried at 105 0 C for 24hrs and weighed. This residue is known as the acid detergent fiber with weight, w 1 . The residue was then transferred to a pre-weighed dry porcelain crucible and heated at 600 0 C for 5hrs. After cooling, its weight w 2 was determined. Acid insoluble lignin was then calculated by the difference (W 1 -W 2 ). Each test was carried out three times.

Preparation of bio composite:-
Neem bark was ground and sieved to a particle size of 10mesh-40mesh . Each particle size of the neem bark flour and the HDPE were mixed using an internal mixer at 50 rpm screw speed. This process was done at a room temperature until a homogeneous mixture was obtained. The mixture was poured into an injection molding machine at temperature of 180 0 C and the composite was injection molded. The injection molded biocomposites are shown in Fig 1. After cooling, the solidified samples were prepared for the test. Mechanical Test:-All mechanical tests were carried out at the temperature of 23 0 C ± 2 and relative humidity of 50 ± 5%. The tensile strength test was carried out in a universal testing machine (Hounsfield Tensometer, model number 8889, with an accuracy of BSS 1610) according to ASTM D638 at a crosshead speed of 5mm/min. The impact test was performed using an impact pendulum tester on notched rectangular specimens according to ASTM D256. The specimen was loaded on the testing machine and the pendulum from the impact tester was allowed to strike the specimen. The Izod notched impact energy absorbed was determined. Brinell hardness test was conducted in accordance with ASTM E103. Each test was carried out three times. The average result was then calculated and presented.
Experimental design and statistical analysis:-Response surface methodology (RSM) was used to optimize the conditions for the fiber preparation to obtain optimum mechanical properties. The design of experiments was done using Minitab17.0. The screening experiment shows that only particle size and filler content were significant factors. Thus, the two independent variables were employed by central composite design (CCD). The experiments were randomized to protect against unknown bias distorting the experimental result. The design matrix for the experiment is presented in Table 1. The response functions measured were tensile strength, impact strength, percentage elongation, Brinell hardness and tensile modulus. A second order polynomial equation was fitted for each factor as follows: Y= β 0 + β 1 x 1 + β 2 x 1 + β 11 x 1 2 + β 22 x 2 2 + β 12 x 1 x 2 where y is the estimated response; β 0 , β 1, β 2, β 11, β 22 and β 12 are constant parameters; x 1 and x 2 are the values of the independent variables of particle size and filler contents respectively. Table 1 shows the design matrix with the uncoded values of the process variables and their corresponding responses. The variance of each factor was partitioned into linear, quadratic and interactive terms.

Results and discussion:-Model Selection and Verification of the Mechanical Properties:-
The experimental data were analyzed using Minitab 17.0 software. Analysis of variance (ANOVA) and regression analysis were used to analyze all the responses in order to evaluate the coefficient terms and for model fitting. The results are tabulated in Table 2-6. The ANOVA demonstrated that the quadratic regression model for all the mechanical properties were highly significant. The test shows very low P-value; P < 0.001 for models of all responses. This probability value means that there is less than 0.1% chance that a model value of this magnitude could occur due to noise. However, the lack-of-fit for all mechanical properties were significant. This implies that the model requires further analysis. Many insignificant model terms were also observed. In order to remove the insignificant terms and improve the model, backward elimination regression with alpha to exit = 0.10, was used to modify the original quadratic model. The goodness-of-fit of the modified models were further inspected using the R 2 values. The R 2 values are shown in Table 7. The values show that 94% for tensile strength, 95% for percentage elongation, 93% for tensile modules, 94% for impact strength and 97% for Brinell hardness of the total variability of the response data around its mean was explained by the model. The difference between the predicted and adjusted R-square was also considered. A rule of thumb is that the adjusted and predicted R-square should be within 0.2 of each other [6]. Table 7 Shows reasonable agreement between adjusted R-square and predicted R-square (within 0.2 each of the others). Adequate precision was used to measure the signal to noise ratio. A ratio greater than 4 indicates an adequate signal and shows that the model can be used to navigate the design space [7]. The adequate precisions in Table 7 indicate that the model can be used to navigate the design space.       (6) Where; x 1 is the particle size X 2 is the filler content Residual analysis:-Residuals play important role in judging model adequacy [6]. Residuals are the difference between the actual and predicted values. Table 8-12 contains actual values, predicted values and residuals for the mechanical properties of the developed composites. To check whether the residuals followed a normal distribution, a normal probability curve of the residuals was constructed. If the residual plots approximately along a straight line, then the normality assumption is satisfied. Figures 2,3,4,5 and 6 show a normal plot of residuals for the responses. These figures show that there is no apparent problem with normality as the residuals plot approximately follows along a straight line.        The actual response value versus the predicted response value graph was used to determine if the model is a satisfactory fit to the data. The condition is that the data point should be approximately split evenly by the 45 degree line [7]. Figure 7-11 shows the plot of predicted versus actual values for the tensile responses. The plots show that the data points were, approximately, evenly split by the 45 degree line. This show that the models are satisfactory fit to the data. All the values were well predicted by the data. From the above analysis, it can be concluded that this model is suitable for predicting the mechanical properties of forest flame pod flour-filled high density polyethylene composites within the limits of the experiment.

Analysis of Response Surface:-
The 3D response surface plot for the combined effect of particle size and filler content on tensile strength, percentage elongation, tensile modulus, impact strength and Brinell hardness value are shown in figures 12 through 16 respectively. The figures show that there is a quadratic effect of particle size and filler content on the responses except for tensile modulus. Figure 12 indicates that tensile strength decreases slightly with an increase in filler content. This result is similar to the one obtained by Obidiegwu et al [8] in "Walnut Shell Powder on the Properties of Polypropylene Filled Composite" and Salmah et al. [9] in "Coconut Shell Reinforced Unsaturated Polyester Composites." According to Salmah et al [9], the decrease in tensile strength as the filler loading increases. This is due to the poor adhesion of the filler-matrix interface and the agglomeration of filler particles. From Fig 13, it can be inferred that the elongation at break of the composite decreased with an increase in the filler content. According to Ahmed [10], with growing filler content, the stiffness of the composite becomes gradually enhanced, with parallel diminution in the elongation at break. With the improvement in rigidity, the ductility of composites declines, consequently the composites break at lower elongation. The incorporation of filler that has a poor union to the matrix appears to cause disruption in the alignment of the polymer chains. The tensile modulus increases as both the filler content and size increased, as can be seen from Fig 14. This may be ascribed to the fact that the toughness of composites is improved by the addition of filler [11]. The impact strength of the composite increased with the increase in filler content as shown in Fig 15, while lower particle size favours impact strength. Such an increase in impact strength of a thermoplastic composite with an increase in filler content has been reported by Ahmed [10]; Bigg [12] in his study on "Mechanical Properties of Particulate filled Polymers and Igwe [13]. As the concentration of the filler content increased, Izod notched impact energy increased due to the more energy required for crack propagate [11]. The composite displayed the highest hardness values at higher filler contents as shown in Fig 16. This may be ascribed to the fact that the addition of fibers into plastic composite improved the matrix surface resistance to the indentation [14].  It can be viewed in Table 12, that the neem bark has very good cellulose, hemicellulose and lignin content. This result shows that bark cellulose content is close to some fibers in the literature.
Process Optimisation:-Numerical optimization was used to explore the design space to determine factor settings that met the design goal.
The goal is to maximize the responses. Desirability was used as the criteria for selecting factor settings used for the optimization. In this study, the factor settings that gave the combined optimum responses are 21.63 mesh size and 40% filler content at 77.7% desirability.

Conclusion:-
The work studied the effect of filler content and particle size on the mechanical properties of forest neem bark Flourfilled high density polyethylene composites. A quadratic regression model was developed for modeling the responses. Quadratic model was selected because of its high significance level. The model for all the responses shows high values of R 2 and the adequate precision for all the responses are above 4; thus the model can be used to predict the mechanical properties of neem bark flour-filled high density polyethylene composites. From the 3D response surface, it was found that increase in particle size improves all the mechanical properties within the experimental limit. Increase in filler content resulted in decrease of tensile strength and percentage elongation while tensile modulus, impact strength and hardness values improved. The optimal conditions with regards to the mechanical properties were found to be at 21.63mesh particle size and 40% filler content with a desirability of 77.7%. The corresponding responses at the optimal condition was found to be 14.51MPa, 5.28%, 0.45GPa, 92.31KJ/m 2 and 242.14Pa for tensile strength, percentage elongation, tensile modulus, impact strength and hardness value respectively. The optimum conditions were validated with minimum error less than 2 %.