We optimize the placement of sensors for detecting a nuclear, biological, or chemical (NBC) attack in a dense urban environment. This approach draws from two main areas: (1) computational fluid dynamic (CFD) simulations and (2) sensor placement algorithms. The main objective was to minimize detection time of a NBC sensor network for attacks on a generic urban environment. To this end we conducted simulations in such environments using thirty-three (33) unique attack locations, thirty-three (33) candidate sensor locations, prevailing wind conditions, and the properties of the chemical agent, chlorine gas. A total of ninety-nine (99) simulated attack scenarios were created (three sets of thirty-three unique attack simulations) and used for optimization. Simulated chemical agent concentration data were collected at each candidate sensor location as a function of time. The integration of this concentration data with respect to time was used to calculate the contaminant "consumption" of the network and the sensor placement algorithm, along with contaminant-level detection, minimized consumption to the network while also minimizing the number of sensors placed. Our results show how a small number of properly placed sensors (eight (8), in our case) provides the best achievable coverage (additional sensors do not help).