Galerkin method is used for discretisning the differential operator in space for the boundary-value (PDEs) and initial-boundary-value problems (time dependent PDEs). While it may be used for pure initial value problems, I don't think it is worth the effort.
You will be better off by using the conventional time integration techniques such as Newmark-beta, generalised-alpha or Runge-Kutta schemes by transforming the equations into the space.
Evolutionary problems involving parabolic and hyperbolic equations in any number of space dimensions are formulated in such a way that Galerkin methods, using a local basis, can be used to find approximate solutions. These methods are particularly suited to problems involving general regions and derivative boundary conditions.