http://etheses.bham.ac.uk/793/1/Coppell10PhD.pdf
(sticking the link here for reference)
Having now read it, something made me uneasy; I can't follow all the
maths, but the seems to be a big hole in the scope of this sort of
modelling.
Then I worked out what it is: the 'feedback loop' is completely absent.
For instance, Anna uses the observed variable velocity of the water
through the stroke, but the thing is that that variable velocity is a
function, partly, of what's going on with the blade the first place.
So we have a model of how the water's behaving through a typical stroke,
which is good, but what's absent seems to be how much effort it takes to
make the blade describe that shape, and whether that effort is being put
to best use.
(Which wasn't part of the scope of the thesis, obviously, but seems to
me the most interesting question.)
AJP
The effort it takes to move the blade in a certain way requires
additional formulas. The entire contents of chapter 5 deal with this.
The CFD model tells how the water bends, the way the water bends tells
what the input in the CFD model should be.
"Through incorporating the propulsive properties of the oar blades
into a larger mathematical model of rowing, it should be possible to
determine the performance of the rowing boat. Therefore, propulsive
force data obtained from the oar blades in Chapters 3 and 4 were
integrated into a mathematical model of rowing."