Ibtisam Mogul

The current work aims to develop computational models for the thermal characteristics of turbulent CH4 flames for varying burner dimensions. This study develops a platform for data driven analysis of temperature prediction of turbulent non-premixed flames, in which the influence of flow and geometric parameters, including burner head diameter (D), half cone angles (α), and co-flow air velocity (Ucf), have been considered. The algorithms used were ridge regressor (RR), linear regressor (LR), and three variations of support vector regression (SVR): SVR with a linear kernel (SVR-LR), SVR with a radial basis function function (SVR-RBF), and SVR with a polynomial kernel (SVR-Poly). The performance of each computational model was evaluated and contrasted based on several metrics: mean absolute error (𝑀𝐴𝐸), regression coefficient (𝑅 2 ), mean absolute percent error (𝑀𝐴𝑃𝐸), and mean Poisson deviance (𝑀𝑃𝐷). From the modelling of the output data, it was observed that the SVR-RBF predictions were more accurate compared to those from the other algorithms, as it achieved the highest training 𝑅 2 value of 0.955. The testing predictions of RR, SVR-LR, SVR-RBF, and SVR-Poly algorithms were also robust, with 𝑅 2 values ranging between 0.91-0.94. It is, therefore, established that these computational models are effectively suited for predicting sensitive turbulent CH4 flame characteristics based on varying input factors.

The current study focusses on developing a back propagation neural network model for depth of cut during the abrasive water jet machining of a Ti-6AL-4V aluminum alloy. The study analyzed depth of cut for five different water jet abrasive parameters namely, water pressure, transverse speed, abrasive mass flow rate, abrasive orifice size, and nozzle to orifice diameter. Experiments were conducted as per the L27 Taguchi-design of experiments (DoE). The back propagation neural network model comprising of one input layer, one hidden layer and an output layer with an architecture of 1–5–6 was chosen for conducting the analysis. The algorithm predicted the Taguchi based output values for the experimental depth of cut with an accuracy of up to 95 %. The neural network algorithm further automated itself, generating 50 new data sets for K-cross validation, out of which 70 %, 20 %, and 10 % of the data were used for training, testing, and validation, respectively. Confirmatory experiments were conducted for depth of cut and assessed against the data set used for validation (10 %). The results showed that as the depth of cut was small, i.e., ranging from 3 mm to 5 mm, the algorithm was unable to predict the optimized parameters, however, the prediction improved as the depth of cut increased. Overall, the consistency between the neural network predicted and the experimental depth of cut throughout the algorithm confirmed the validity of the procedure and the appropriateness of the algorithm.