Keff Vs. Time graph coming unexpected after burnup

Hello, I have depleted a hexagonal fuel assembly for about 200 days, taking 10 steps, while depletion code runs smoothly, the output is very random and I have checked my geometry and materials and they are alright and the initial keff also matches with the benchmark file.

Can anyone tell me what could be causing the irratic behavior of keff during depletion?

image

The graph looks dramatic, but the keff is really only varying by less than 0.1%. The problem may just be statistical variance as the result of insufficient particle count.