The ten datapoints of USS NEW JERSEY´s 1943 trial results for 56,200 to 56,800ts displacement result in the following MMF model curve:
010000 SHP = 12.6kts (extrapolation)
016470 SHP = 15.3kts (primary data, trial)
016800 SHP = 15.5kts (primary data, trial)
020000 SHP = 16.4kts (interpolation)
030000 SHP = 18.7kts (interpolation)
038000 SHP = 20.0kts (primary data, trial)
038340 SHP = 20.0kts (primary data, trial)
040000 SHP = 20.5kts (interpolation)
050000 SHP = 21.9kts (interpolation)
060000 SHP = 23.1kts (interpolation)
070000 SHP = 24.1kts (interpolation)
078000 SHP = 24.9kts (primary data, trial)
080000 SHP = 24.9kts (interpolation)
081900 SHP = 25.1kts (primary data, trial)
090000 SHP = 25.7kts (interpolation)
100000 SHP = 26.3kts (interpolation)
120000 SHP = 27.5kts (interpolation)
126400 SHP = 27.9kts (primary data, trial)
140000 SHP = 28.4kts (interpolation)
160000 SHP = 29.2kts (interpolation)
163400 SHP = 29.3kts (primary data, trial)
170960 SHP = 29.7kts (primary data, trial)
180000 SHP = 29.9kts (interpolation)
200000 SHP = 30.6kts (interpolation)
212000 SHP = 30.9kts (rated design power)
220000 SHP = 31.1kts (interpolation)
221030 SHP = 31.0kts (primary data, trial)
240000 SHP = 31.6kts (extrapolation, max. power 1 -without cavitation)
254400 SHP = 31.9kts (extrapolation, max power 2 -assuming no cavitation takes place)
----------------------------
Higher speeds would be possible at lower load. Main issue with these data is that performance is somehow overstated at speeds in excess of 30kts but the error margin is + 0.15kts. One digit data, only.
----------------------------
Given the data we know from the three ships mentioned:
BISMARCK for displacements at 75% load
RICHELIEU for displacements at 25 to 50% load
and NEW JERSEY for displacements at 80 to 85% load
and assuming that the relevant rated powers (100% output) are:
134000 SHP for BISMARCK
150000 SHP for RICHELIEU, and
212000 SHP for NEW JERSEY
it´s possible at least to draw comparative performances for these respective loads (that is unaccounted for load differences, using trial reference as the only source), comparing the speed with the powerload (1% output to 120% power):
note that differences in displacement are unaccounted for. The IOWA class appears to be less efficient at very high speeds or else, RICHELIEU and BISMARCK are more efficient. The reasons for this I am not intending to explore, but I guess that lower propulsive efficiency caused by the close neighborhood to cavitation limits may in part be interesting to investigate.