Application of artificial neural networks for the prediction of aluminium agglomeration processes

<p>Aluminium is universal and vital constituent in composite propellants and typically used to improve performance. Aluminum agglomeration takes place on the burning surface of aluminized propellants, which leads to reduced combustion efficiency and 2P flow losses. To understand the processes and behaviour of aluminum agglomeration, particles size distribution of composite propellants were studied using a quench particle collection technique, at 2 to 8 MPa and varying quench distances from 5mm to 71mm. To predict the agglomerate diameter of metallized/ultra-fine aluminium of composite propellants, a new artificial neural network (ANN) model was derived. Five Layered Feed Forward Back Propagation Neural Network was developed with sigmoid as a transfer function by varying feed variables in input layer such as Quench distance (QD) and pressure (P). The ANN design was trained victimization stopping criterion as one thousand iterations. Five ANN models dealing with the combustion of AP/Al/HTPB and one ANN model of AP/UFAl/HTPB composite propellants were presented. The validated ANN models will be able to predict unexplored regimes wherein experimental data are not available. From the present work it was ascertained that, for agglomeration produced by quench collection technique, the ANN is one of a substitute method to predict the agglomerate diameter and results can be evaluated rather than experimented, with reduced time and cost. The resulting agglomerates sizes from ANN model, matches with the experimental results. The percentage error is less than 3.0% of the label propellants used in this work.</p>