Sometimes calcium aluminate castables fail to set in a reasonable time.    Often one finds hard spheres or sections thereof in the matrix of soft wet material.    I suspect that whereas normally castables begin crystallisation at billions of nucleation points, in this case only a few points succeed in nucleating, and the setting progresses only on the surfaces of the hydrate crystals.   These growth points never seem to coalesce to include the whole castable.   Something runs out or maybe something accumulates on the growth front.   

I would love to find out more about this phenomenon, and hear from others who have encountered it.    Perhaps a good topic for a thesis.    I have tried to seed castables with finely milled hydrate crystals without success.