I would definitely not use separation of variables in obtaining an analytic solution to this problem. Here is the approach I would recommend:
1. Start with the analytic solution for the case of C = 0, and on the infinite x interval from minus infinity to plus infinity. This is a closed-form single term expression (not a series of sines and cosines).
2. Add the periodicity to this solution by putting reflective boundary conditions at x = L/2 and x =-L/2 by taking a superposition of terms in the form of 1. , based on delta functions being initially located not only at x = 0, but also at x = + nL and x = - nL, with n = 1,2,3,...
3. Add the translation (advection) by substituting x - Ct for x in all the terms of 2.
This will satisfy the differential equation, the initial condition, and the boundary conditions