In the present work, the effect of attaching a combination of hollow baffles and winglets on the bottom surface of rectangular channel was investigated numerically with Reynolds number range from 〖3.8×10〗^3 to 〖2.4×10〗^4 using the commercial CFD software ANSYS FLUENT. First, the thermal and hydraulic characteristics of hollow trapezoidal baffles were investigated at the same pitch with different geometric parameters: baffle height (H), baffle width (S), baffle length (L), corner angle (α), inclination angle (β), and Reynolds number (Re). The artificial neural network (ANN) approach was used to determine the optimum design point. Backpropagation technique was used to determine the optimal ANN size with the Bayesian regularization algorithm. The performance of the hollow trapezoidal baffles was compared with the experimental data obtained by Sahin et al. [21] to validate the used numerical model. Two modified shapes of hollow baffle (elliptic and wavy top surface) were investigated in both aligned and staggered arrangements to maximize the benefits of using the hollow baffles. Based on the thermal hydraulic efficiency values, the optimum condition is obtained at α= 16°, β = 0°, and baffle dimensions of H= 20 mm, L= 25 mm, S= 26 mm. Additionally, the baffles of wavy top surface in staggered arrangement outperformed the straight and elliptic top surface ones. The spaces between baffles in the baffles array were optimized in both longitudinal (L/P_L=0.2 - 0.8) and transverse (S/P_t=0.2 - 0.8) directions. The best performance was obtained at a longitudinal length to pitch ratio L/P_L=0.2 and transverse length to pitch ratio S/P_t=0.4 which is consistent with what the ANN validation results. The thermal and hydraulic characteristics of a combination of baffles and winglets were then investigated using a three shapes of delta winglets, selected from the literature, to perform this study. The comparison of the three shapes of delta winglets at different orientations showed that, (shape 3) with orientation angle (θ = 30) is the best. Inserting a single row of winglets between baffles’ rows provided an increasing in thermal enhancement factor by about 10% larger than the no-winglet case. Two rows of the optimum winglet shape are then optimized in aligned and staggered arrangements at different spacing ratio (P_W/P_L = 0.3 – 0.7). The maximum value reached of thermal enhancement factor, at P_W/P_L = 0.5, is about 1.81 in aligned arrangements with an 13% increasing than the no-winglet case. Results of streamlines, contours, local heat transfer coefficient, overall-averaged Nusselt number, friction factor, and thermal enhancement factor were presented and discussed. Additionally, correlations for thermal enhancement factor, Nusselt number, and friction factor were presented for the design parameters of tested baffles in both experimental and numerical approach.