THE ACCURACY OF SCHEFFE’S THIRD DEGREE OVER SECOND-DEGREE, OPTIMIZATION REGRESSION POLYNOMIALS

Abstract

Pozzolanas are materials containing reactive silica and/or aluminum. When the material is mixed with lime in powdered form and in the presence of water, it will set and harden like cement. Rice Husk Ash (RHA) is a form of natural pozzolana. It is found locally in abundance in Nigeria and in many other parts of the world. Udeala (2003), produced a mix containing RHA with 45 per cent slaked lime. This work uses Scheffe's (1958), simplex theory to optimize the compressive strength of concrete made from RHA pozzolan based on (4,2) and (4,3) simplex lattices. The strengths predicted by the models are in good agreement with their corresponding experimentally observed values. The accuracy of strength predicted by the third degree polynomial is about 21 per cent higher than that of the second-degree polynomial. This implies that, except for need of high accurate results, the use of higher order polynomial, that is cumbersome and time wasting, is not necessary. With the models, any desired strength of hardened concrete, given any mix proportions, is easily evaluated. The computer programs are written in BASIC language.