TIP3 waters (residue WAT) have a H-H bond, and SHAKE should be used to keep them rigid. Thus there would be no internal energy for this model of water.
In 'amber41/dat/parm91.dat' there is no angle bending parameter
defined for HW-HW-OW. Is it appropiate to take the missing values from
'parm94.dat':
HW-HW-OW 0. 127.74 (found in crystallographic water with 3 bonds)
[Note: the Link and Edit programs that this question relates to are obsolete, and the distinction between parameters for crystallographic and program-added waters does not apply to Leap.]
Yes...when you use "crystallographic" waters (entered as WAT in the Link step) an angle is defined as above, which needs a parameter. (This angle is not generated when waters are generated by the BOX option of Edit). Note that the force constant above is zero, so this angle is not "really" there (i.e. always has zero energy because of the HW-HW bond which, with the use of SHAKE on H, keeps water rigid) but is necessary in some applications to include it in the parm.dat file. I believe that its absence from parm91.dat was an oversight.