Obstructing the passage of blast waves is an effective method of mitigating blast pressures downstream of the obstacle. To this end, the interaction between a blast wave and a simplified structural shape, such as a cylinder, has been widely investigated to understand the complex flow pattern that ensues around the obstacle. The patterns include the interference zones of the incident wave, the diffracted wave, and other secondary waves in the downstream region. Such zones are responsible for causing significant modifications to the blast wave parameters. This research aims to identify and study the factors that serve to mitigate the resulting blast loads downstream of a cylindrical obstacle – both on the ground, and on a rigid wall target that the obstacle is aiming to protect. Inputs from this numerical study are also used to develop a fast-running predictive method based on an artificial neural network (ANN) model. It was found that the size of the cylinder, the strength of the blast wave, the position of the cylindrical obstruction, and the target length, all have remarkable effects on the development of the complex flow-field downstream, and on the impulse mitigation on a reflective target. A number of key mitigation mechanisms are identified, namely shadowing and interference, and their origins and significance are discussed. An ANN model trained using scaled input parameters could successfully predict impulse values on such a reflective target. Using this model to predict the response of previously unseen configurations (for the ANN) gave excellent correlation, thereby demonstrating the high fidelity of this fast-running tool, and its ability to predict the effectiveness of various wave-cylinder interactions in mitigating blast loading.