Hard to comment on this without experience of the solver or a deeper look at what's happening. Does your anomaly also show up in pressure (I imagine it would) What happens if you ramp up from a really small time step?