Optimization of Thermal Decomposition Conditions of Bone to Achieve the Highest Percentage of Crystalline Phase in Bone Char using Gene Expression Programming and Artificial Neural Network

Bone char (BC) is one of the most common adsorbent with extensive applications in the removal of pollutions. The adsorption capability of BC is proportional to the crystalline index, i.e, the atomic ratio of Ca/P. This study is an attempt to model the crystalline index of BC that by thermal decomposition of natural bone using artificial neural network (ANN) and genetic expression programming (GEP). In this regard, 100 various experimental data used to construct the ANN and GEP models, separately. Through the data collection step, heating rate, the type of precursor, calcination temperature, and residence time selected as the inputs for the preset output as Ca/P ratio. The results reveal that the minimum amount of Ca/P ratio are at the heating rate 10 °C/min, HNO3 1.6 M as activation agent, calcination temperature 1000 °C, and residence time 2 h. R squared indices is used to compare the performance of extracted models. Finally, the best ANN uses to investigate the effect of each practical variable by sensitivity analysis and revealed that the residence time is the most effective parameter on the crystalline index while acid activation is of secondary importance.