The ash yield from the combustion of a mixture of Africa star apple and tropical almond seeds shells (biocomposite biomass) with ammonium dihydrogen phosphate as an additive in a furnace was optimized using I-Optimal Design under the Combined Methodology of the Design Expert Software. The data obtained were analysed statistically using Analysis of Variance (ANOVA), Artificial Neural Network (ANN) for the prediction of ash yield and Principal Component Analysis (PCA) to determine the coefficient of determination (R²) between variables. Proximate analysis was used to evaluate Moisture Content (MC), Fixed Carbon Content (FCC), and Volatile Matter (VM) values while the Higher Heating Value (HHV) of the mixtures that gave the highest and lowest ash yields was evaluated numerically. The optimum conditions of process variables for the compositions of tropical almond, African star apple, and ammonium dihydrogen phosphate, as well as the temperature, were 30%, 60%, 10% and 704 oC, respectively leading to a minimum ash yield of 24.8%. The mathematical models for the ash using the I-optimal design indicate a good fit to the Quadratic model with a R² of 0.9999. The ANN model agreed significantly with the experimental results with an R² of 0.9939.  The VM, FCC, MC, AC and HHV of the highest ash yield were 11.00%, 2.34%, 3.20%, 33.80% and 4487.747 , respectively. The study established the suitability of optimisation tool to develop solid fuel mixtures for possible use in grate furnaces and its efficiencies.