Convergence monitoring!
 

I solving a jet impingement problem, (compressible Euler ) Computational domain is a trapezoid (i last plane has an inclination of 30 deg) with grids 31x25x25. CFL=1.0 iterations=200 cycles=5 (so 1000 iterations) I monitored the L2-norm , it initially drops but then at around 500-600 iterations it peaks and then drops off. For 2000 iterations there were peaks at iter no. 500-600 , around 1200 It almost seemed periodic. Any suggestions what must be causing this? Thanks