Taku et. al., Vol. 13, No. II, December 2017, pp 1-14. 1 OPTIMIZATION MODEL FOR COMPRESSIVE STRENGTH OF SANDCRETE BLOCKS USING CASSAVA PEEL ASH (CPA) BLENDED CEMENT MORTAR AS BINDER 1 Amartey Y. D., 2 Taku J. K. * , 1 Sada B. H. 1 Department of Civil Engineering, Ahmadu Bello University, Zaria, Nigeria 2 Department of Civil Engineering, University of Agriculture, Makurdi, Nigeria * Corresponding author’s email: kumataku@yahoo.com Received 24 August, 2016; Revised 20 August, 2017 ABSTRACT This research work applies Scheffe’s second degree simplex theory to formulate a regression model for the optimization of the compressive strength of sandcrete blocks using cassava peel ash (CPA) blended Portland cement (OPC) as binder material for different mix ratios as multivariate functions with the proportions of the sandcrete block ingredients serving as variables. The experimental values of the compressive strength were obtained by performing destructive strength tests on the blocks after curing for 28 days, with a binder-aggregate ratio of 1:8 and water binder ratio ranging from 0.45 to 0.60, the OPC being replaced with CPA at 0 – 30% for the respective water-binder ratios. The optimization model from the Scheffe’s mixture method for a (4, 2) factor space was found to be y= f(x) = 1.95x1 (2x1-1) + 1.84x2 (2x2-1) +1.81x3 (2x3-1) +1.79x4 (2x4-1) + 6.08x1x2 + 5.72 x1x3 + 1.89 x1x4 + 7.28 x2x3 + 1.80 x2x4 + 7.16 x3x4. The model was tested using the student t- test at 95% accuracy and found to be accurate. Thus, the model can be used to predict any desired compressive strength value for CPA- OPC blended sandcrete blocks given any water-cement ratio between 0.45 and 6.0 and vice versa. Keywords: Sandcrete Blocks, Cassava Peel Ash, Optimization Model, Scheffe’s Simplex design and Student t-test. INTRODUCTION Predictive modeling is the name given to a collection of mathematical techniques employed to derive mathematical relationships that are generated using experimental data between a dependent variable and a number of independent variables with the sole aim of measuring and inserting the values of the predators into the model to predict or determine the value of the target variable within the shortest possible time [1]. Thus the use of predictive modeling saves time, energy and resources [1]. Mathematical modeling has found various applications in concrete technology, the commonest being the application of predictive models like those derived from Henry Scheffe’s mixture models to predict concrete properties like strength [2–6]. Scheffe’s mixture model is a single step multiple comparison ORIGINAL RESEARCH ARTICLE OPEN ACCESS