The Particle Collision Algorithm (PCA) is a Metropolis-based algorithm loosely inspired by the physics of nuclear particle collision reactions, particularly scattering and absorption. This metaheuristic is applied to a Nuclear Power Plant Auxiliary Feedwater System surveillance tests policy optimization, and its performance is compared to previous results. The optimization problem consists in maximizing the system's average availability for a given period of time, considering realistic features such as (i) aging effects on standby components during the tests; (ii) revealing failures in the tests imply on corrective maintenance, increasing outage times; (iii) components have distinct test parameters (outage time, aging factors, etc.); (iv) tests are not necessarily periodic. The PCA outperforms the canonical genetic algorithm and two of its variants, ratifying the previous performance in another optimization problem.