Experimental and theoretical investigation of water removal from DMAZ liquid fuel by an adsorption process

2-dimethylaminoethylazide (DMAZ) is a new liquid fuel that has made significant progress in bio/mono propellant rocket engines in recent years. Purification of DMAZ fuel by reducing its water content using various adsorbents including zeolites, calcium chloride and nano-particles is experimentally and theoretically investigated. The highest water adsorption of 92.6% from the DMAZ solution is obtained by the CaCl2 adsorbent within 10 min. Four different artificial neural networks (ANN) are examined to correlate an extent of removed water from the DMAZ solution to its affecting parameters. The performed regression analysis indicated that water initial concentration (WIC), adsorbent types, solution temperature, contact time and adsorbent dosage are the most important affecting variables for water sorption from the DMAZ solution. The accomplished statistical analysis demonstrated a multi-layer perceptron neural network (MLPNN) with seven hidden neurons and is the most accurate approach for modeling the considered task. The obtained results showed that the proposed MLPNN model could be successfully employed for accurate prediction of an amount of water removal from the DMAZ fuel solution by the adsorption process.