The solution of complex problems, including schedulling ones, must to be faced following the stages:
1. Right formulation of the overall schedulling problem.
2. Right decomposition of the formulated task
3. Right selection of the optimization methods to be used for solving the subtask derived from the decompositon and for the solutions composition between subtasks.
In the stage 1st could be included complex engineering procedures resulted from the specialty evolution, without any simplification.
For facing the 2-nd stage I have proposed approches that could be found, by example, in the work attached.
If the 2 first stages are right done, then the 3-th stage will be not difficult to complete. For doing it is available an important arsenal of exact and heuristic methods included modern heuristics.
Metaheuristics will never reach an optimum, if the optimisation problem is complex and/or very large. Running a few metaheuristics will do no good for anyone.
This is a good question. Exploring and studying the limitations of meta-heuristics will develop their performance day-to-day. I recommend studying the below publications: