I noticed from this advice that if VELOCITY PROFILES = 4(instead of 5), I can get no errors at BC(2)(3) even if the relaxation factor = 0.02.
Then I could get a reasonable Q(1) and H(1) with liquid boundary:[4 4 4] (not [5 5 5]).
I can't make use of the rating curve yet,and can't understand the differences between the manual description of [444] and [555] nor formula used in [444].
But Free HQ BC is working now, so I can take the next step.
Thanks.