Thank you Fifi, my first remarks:
DIESEL
Forward:
- Slow: speed 2% faster than espected, rpm 10% faster than espected
- One third, Standard, Full, Flank: speed as espected or 1% lower, rpm 3 to 6% faster than espected
Backward:
- Back Slow, Back Standard, Back Full: speed 8 to 10% faster than espected, rpm 12 to 14% faster than espected
- Back emergency: speed 11% slower than espected, rpm 5% faster than espected
ELECTRIC
Forward:
- Slow: speed 18% slower than espected, rpm 30% faster than espected
- One third, Standard: speed 3 to 4% slower than espected, rpm 7 to 8% faster than espected
- Full, Flank: speed 1% faster than espected, rpm 2 to 4% faster than espected
Backward:- Back Slow: speed 16% faster than espected, rpm 48% faster than espected
- Back Standard: speed 6% faster than espected, rpm 17% faster than espected
- Back Full: speed 14% faster than espected, rpm 13% faster than espected
- Back emergency: speed 11% slower than espected, rpm 8% faster than espected
The main problem is not the discrepancy between espected and observed outputs, but the fact that rpm/speed deviations are dishomogeneous, i.e. apparently SH5 U-boat speed is not growing linearly with engine rpm, whereas data from our sources suggest a linear relation among them.
In other words, unless TDW manages to find the culprit in game code, we can either match speeds or rpm to historical values. It might seem a secondary problem, but if (as it was in RL) engine noise is related with rpm, we risk being hydro-detected at speeds that should be classified as "silent". On this subject, it is useful noting again that the most efficient silent speed was at 90 rpm.
I hope that plotting your data on a chart and comparing the resuting graph with the espected lines, will cast some light on the issue. I will post the chart as soon as possible
