Disclaimer I don't claim to understand the technical aspects ("orthogonal subspace update") nor the uranyl chemistry.
However, in my experience excitations are quite sensitive to the given geometry, especially when electronic states are beginning to interact with each other. Have you tried re-optimising the geometry of your complex first (e.g. with solvent environment), and then run the excitation afterwards? Sometimes finding and reproducing sensible geometries takes up a lot of time.
Also, depending on your molecule or the properties of your specific excitation, e.g. excitations with strong CT character, you might also need to find a more adequate functional. Try screening the relevant literature.
Good luck!
