In this work, the stir casting method was applied to obtain composites based
on the alloy AN EW 6061 used as a metal base, and Al2O3 particles as a
reinforcement. Composites play a significant role as engineering materials.
Therefore, it is necessary to study, in detail, the production methods and
the factors that affect their mechanical properties. For this purpose, we
have carried out a planned experiment wi ASM International th the aim to use
regression analysis to predict the influence of particle size and mass
fraction on hardness of the obtained composites. The full factorial
experimental design with two factors was used, which was analyzed at three
levels. Hardness was observed as a system response, while particle size and
mass fraction were set as influencing factors. Influencing factors were
observed at three levels: 50, 80 and 110 ?m for the particle size and 2, 5
and 8 mass%. Measured hardness values of the composites ranged from 72 HV10
to 80 HV10. Based on the probability values (p<0.05), it was determined
which factors are important for the system response. Statistical analysis
has shown that linear terms of the influence factors (size and mass fraction
of reinforcement particles) and the square term of the mass fraction have
statistical significance on the hardness change. The square term of the
particle size and the interaction term of the influencing parameters do not
have a statistically significant contribution in predicting the hardness
value. Thus, a second-order polynomial model was obtained by the regression
analysis. Influence of input factors on the system response and the adequacy
of the obtained mathematical model were determined by using the Analysis of
Variance (ANOVA). Based on the statistical data analysis, it was established
that, the particle mass fraction has a greater influence on hardness of the
obtained composite in relation to the particle size. By comparing the
experimental and predicted values, a high degree of agreement was achieved
so that the chosen model of the factorial experiment was adequate
(R2=0.989). It can be also concluded that the developed regression model can
be applied to predict hardness of the aluminum composite reinforced by Al2O3
particles in the chosen variation interval of particle size and mass
fraction.