There have been many proposed uses of unmanned underwater vehicles (UUVs) including military applications, port safety and security, environmental monitoring, and scientific research. Current trends are moving towards using multiple UUVs to execute these missions quickly and more efficiently. Unfortunately, UUV failures due to hardware malfunctions, software bugs, and unforeseen environmental conditions can disrupt mission execution, which will need to be addressed promptly. Therefore, an important issue is to re-organize the remaining UUVs and re-structure missions when these failures occur. In this work, we propose and evaluate a method for UUV failure recovery, developed at the Applied Research Lab. This method is based on a genetic algorithm that can help mission organizers determine the best action plan when one or more UUVs fail to continue mission execution with minimal disturbance. This paper evaluates our proposed method under different UUV failure patterns and quantifies the algorithm's effectiveness. We also discuss how our approach can be used to handle the scenarios when mission re-planning is necessary due to not only UUV failures, but also the dynamic changes in mission goals and constraints.