Manned spaceflight in outer/deeper space will require crew operations that are independent of ground support. This requires the crew to re-plan day-to-day activities, particularly in the case of unforeseen circumstances. To support these planning duties, we are developing a mixed-initiative planning tool that optimizes schedules in collaboration with astronauts. This paper highlights the tool's planning algorithm. The planning algorithm has two closely-coupled components: first, an optimization algorithm (optimizer) based on local search heuristics and, secondly, a computational model of the work that is to be performed. In this framework, the optimizer acts as a surrogate model of the more detailed computational models, such that new solutions can be efficiently explored. The computational work model is capable of simulating a plan through time, and can account for dynamic interactions between activities and work environment that are not modeled in the optimizer. Moreover, the computational model returns to the optimizer metrics that reflect required teamwork to coordinate activities between astronauts. The paper includes a description of the optimizer and computational simulation models as well as a case study with activities, agents and resources that are representative of a typical manned mission.