## drag computation

### drag computation

I just ran across this short video about computational fluid dynamics work on shell hull design;

http://www.vespoli.com/vespoli/news/2009/02/26/racing-shell-hull-desi...

and I have a question.  This may be too technical to be of any general interest, but I know there are real experts among you who might weigh in.  These fellows talk about developing a sort of quasi-dynamic model of drag throughout the stroke cycle by first computing a set of static (steady-state) drag simulations (with a RANS code) for a range of shell speeds and pitch angles, and then stitching these solutions together, choosing one for each instant in time, based on a model of the stroke dynamics, to get the whole evolution of drag through the cycle.

I wonder how good this is.  It seems to me that both the pitch and speed of a real shell vary so quickly- much faster than the flow transit time along the hull-  that even the instantaneous mean flow, never mind the reynolds stress, is going to be quite different from a steady-state simulation at that speed and pitch.  For instance, there will be vertical components of flow as the boat pitches that wouldn't be present in steady-state.

This is related to something I've often wondered about, namely why drag seems to increase significantly with quite small amounts of surface chop.  From the Reynolds-averaged flow perspective, is it the perturbation by the waves of the mean flow that dominates, or is the development of the turbulence- the Reynolds stress- greatly affected, or both?

We all know, looking at our wakes, that when we row through ripples or small chop the boat leaves a flattened path behind it-  the turbulence we leave behind scrambles and flattens out the waves.  So I imagine that small surface waves must also interact strongly with the turbulent boundary layer as it is generated along the hull surface.  But I don't know what that interaction does to the turbulence and thereby to the drag on the hull.

Sorry if this is too technical (or too dumb a question), I'm still pretty new to the group and don't yet know the limits of what's tolerable.

thanks,
John G

### drag computation

Dear John,

It's actually a very good question. I think there is a lot of "confusion" or possibly ignorance when it comes to drag calculations on boats. I am not an expert on boats either; my area is fluid mechanics and computational fluid dynamics.

I am not sure if I agree with everything stated in the video that you mention. For instance, they estimate "Aero drag" at 3% of the total drag, which I think is, for typical rowing conditions, an underestimation. Also, the "viscous drag" is mentioned as "the opportunity", where probably all fast rowing boats know how to deal with this: minimize the surface cross sectional area, thus a U shape under your seat. So it is more of a standard than an opportunity.

If the viscous drag (I usually say friction drag) is really so ***, then the RANS and potential code they refer to is really not suitable - or at least not what does the calculation in a code. It's all boundary layer dynamics, for which a RANS code uses an empirical "log-layer" profile. But this is, strictly speaking, only valid for a 1D fully developed steady boundary layer. The RANS model far away from the boundary layer is also inappropriate, as the Reynolds number is far too low there.

You bring up two things which I have also often wondered about: the unsteady component of the drag, and the wave drag. I think the unsteady component is significantly larger than some people realize, as when you accelerate the boat, you need to pump the additional water in the boundary layer around it. Doing a very ballpark estimate of the boundary layer, approximating it as a boundary layer on a 4 meter long plate, I get a boundary Reynolds number of 10^7 and a boundary layer thickness of a few centimeters at the end of the boat. So that is a lot of additional water which needs to be moved around. You cannot estimate this effect from stitching two solutions from different instances together in a simulation.

It has been commonly accepted that wave drag is very small for rowing boats, but it might be a bit larger than some thought before. The reason why I think this, is because you actually do see some kind of bow-wave on a boat, e.g. http://SportToday.org/
It also gets stronger if you look at unsteadiness during the rowing stroke.

A second reason, is because some boat builders are trying to build "longer" boats; for instance, if you see the new Empacher single design, they try to elongate the bow deeper into the water for longer that they did a few years ago. This would be a typical change if they thought that wave drag plays a role and they try to minimize this by elongating the boat.

In the simulation framework in the video you showed, these effects are all neglected; a single-phase steady RANS calculation. I wouldn't use it to design a boat.

If I can get my hands on a good CAD drawing of a rowing single and a good student willing to work with me, I am happy to do some more state-of-the-art calculations.

Sincerely,

Berend van Wachem.

### drag computation

Quote:
> Dear John,

> It's actually a very good question. I think there is a lot of "confusion" or possibly ignorance when it comes to drag calculations on boats. I am not an expert on boats either; my area is fluid mechanics and computational fluid dynamics.

> I am not sure if I agree with everything stated in the video that you mention. For instance, they estimate "Aero drag" at 3% of the total drag, which I think is, for typical rowing conditions, an underestimation. Also, the "viscous drag" is mentioned as "the opportunity", where probably all fast rowing boats know how to deal with this: minimize the surface cross sectional area, thus a U shape under your seat. So it is more of a standard than an opportunity.

> If the viscous drag (I usually say friction drag) is really so ***, then the RANS and potential code they refer to is really not suitable - or at least not what does the calculation in a code. It's all boundary layer dynamics, for which a RANS code uses an empirical "log-layer" profile. But this is, strictly speaking, only valid for a 1D fully developed steady boundary layer. The RANS model far away from the boundary layer is also inappropriate, as the Reynolds number is far too low there.

> You bring up two things which I have also often wondered about: the unsteady component of the drag, and the wave drag. I think the unsteady component is significantly larger than some people realize, as when you accelerate the boat, you need to pump the additional water in the boundary layer around it. Doing a very ballpark estimate of the boundary layer, approximating it as a boundary layer on a 4 meter long plate, I get a boundary Reynolds number of 10^7 and a boundary layer thickness of a few centimeters at the end of the boat. So that is a lot of additional water which needs to be moved around. You cannot estimate this effect from stitching two solutions from different instances together in a simulation.

> It has been commonly accepted that wave drag is very small for rowing boats, but it might be a bit larger than some thought before. The reason why I think this, is because you actually do see some kind of bow-wave on a boat, e.g. http://SportToday.org/
> It also gets stronger if you look at unsteadiness during the rowing stroke.

> A second reason, is because some boat builders are trying to build "longer" boats; for instance, if you see the new Empacher single design, they try to elongate the bow deeper into the water for longer that they did a few years ago. This would be a typical change if they thought that wave drag plays a role and they try to minimize this by elongating the boat.

> In the simulation framework in the video you showed, these effects are all neglected; a single-phase steady RANS calculation. I wouldn't use it to design a boat.

> If I can get my hands on a good CAD drawing of a rowing single and a good student willing to work with me, I am happy to do some more state-of-the-art calculations.

> Sincerely,

> Berend van Wachem.

Good information, Berend.  When I first saw that video over a year back
I was puzzled by some of the percentages advanced.  Like you, I found
the wave drag attibution very low, especially since they seemed to be
advocating a shorter shell for reduced wetted surface.  And the wind
drag element again seemed unduly low - first define your conditions.  It
was all rather hand-wavy, but I've no idea of the audience.

Dirk Kramers used to be associated with Resolute, IIRC.  Also I think
with Custom Sailboats, the builders of USA-53 & USA-58 America's Cup
yachts, and more recently with the very beautiful but ultimately beaten
Alinghi.

I might have to take you up on your final statement/request....

Cheers -
Carl

--
Carl Douglas Racing Shells        -
Write:   Harris Boatyard, Laleham Reach, Chertsey KT16 8RP, UK
Find:    tinyurl.com/2tqujf

### drag computation

Quote:
> Dear John,

> It's actually a very good question. I think there is a lot of "confusion" or possibly ignorance when it comes to drag calculations on boats. I am not an expert on boats either; my area is fluid mechanics and computational fluid dynamics.

> I am not sure if I agree with everything stated in the video that you mention. For instance, they estimate "Aero drag" at 3% of the total drag, which I think is, for typical rowing conditions, an underestimation. Also, the "viscous drag" is mentioned as "the opportunity", where probably all fast rowing boats know how to deal with this: minimize the surface cross sectional area, thus a U shape under your seat. So it is more of a standard than an opportunity.

> If the viscous drag (I usually say friction drag) is really so ***, then the RANS and potential code they refer to is really not suitable - or at least not what does the calculation in a code. It's all boundary layer dynamics, for which a RANS code uses an empirical "log-layer" profile. But this is, strictly speaking, only valid for a 1D fully developed steady boundary layer. The RANS model far away from the boundary layer is also inappropriate, as the Reynolds number is far too low there.

> You bring up two things which I have also often wondered about: the unsteady component of the drag, and the wave drag. I think the unsteady component is significantly larger than some people realize, as when you accelerate the boat, you need to pump the additional water in the boundary layer around it. Doing a very ballpark estimate of the boundary layer, approximating it as a boundary layer on a 4 meter long plate, I get a boundary Reynolds number of 10^7 and a boundary layer thickness of a few centimeters at the end of the boat. So that is a lot of additional water which needs to be moved around. You cannot estimate this effect from stitching two solutions from different instances together in a simulation.

> It has been commonly accepted that wave drag is very small for rowing boats, but it might be a bit larger than some thought before. The reason why I think this, is because you actually do see some kind of bow-wave on a boat, e.g. http://SportToday.org/

> It also gets stronger if you look at unsteadiness during the rowing stroke.

> A second reason, is because some boat builders are trying to build "longer" boats; for instance, if you see the new Empacher single design, they try to elongate the bow deeper into the water for longer that they did a few years ago. This would be a typical change if they thought that wave drag plays a role and they try to minimize this by elongating the boat.

> In the simulation framework in the video you showed, these effects are all neglected; a single-phase steady RANS calculation. I wouldn't use it to design a boat.

> If I can get my hands on a good CAD drawing of a rowing single and a good student willing to work with me, I am happy to do some more state-of-the-art calculations.

> Sincerely,

> Berend van Wachem.

Splendid observations, Berend!

I'm all in favour of more fluid dynamic research, so I hope
you find the information and the good student. It can be an
enormous amount of work to collect and collate all the velocity
and accelration data, especially in 6dof and with varing forces
applied by crew members during different stages of a race.

Just a quick question...

If you calculate the drag of the hull and crew from a stationary
start, and for an entire (let's say) 2000m race while they are
moving with 6 dof, how will you know your predictions are as
good as, or better, than predictions made with other methods?

"All models are wrong, but some are useful." - George Box.

### drag computation

Quote:
> If you calculate the drag of the hull and crew from a stationary
> start, and for an entire (let's say) 2000m race while they are
> moving with 6 dof, how will you know your predictions are as
> good as, or better, than predictions made with other methods?

Which 6 dof are you referring to? A typical fluid mechanics simulation for this case will involve 3 velocity components, pressure, and a parameter to describe the free interface. So 5 unknowns, but multiply this with about 10^7 mesh cells.
Maybe you also mean the dof that control the position/rotation of the boat itself?

For the unsteadiness, there are actually 2 factors: the first is the acceleration/deceleration of the boat, as I described above. The second factor is that because the rower moves relatively in the boat, the orientation of the boat (the pitch angle) will also change, which affects both steady and unsteady drag. The latter will especially contribute to wave drag.

An important item when one performs a calculation is the validation, consisting of several parts: first of all the assumptions that go into the model (e.g. if you use a RANS model, how steady is the boundary layer actually?). The second one concerns the numerical error one makes, which can be verified by refining the mesh around the boat and see how much the solution changes. Thirdly is to test the model on a known solution, preferably an experiment. The latter is easily doable for a steadily moving boat, or a boat in a tow-tank, as there is a lot of empirical data for this. For a moving boat, I guess we could do our own experiment: if we create a velocity curve of a rowing boat (with an impeller) and measure the force exerted on the pins as a function of time, this could serve as a good validation. I know many of the measurements of the force on the pin or oar are referred to as "power" curves. But unless I am mistaken, I think they measure force. Power is force times displacement over time, and I don't see how you can measure this just from the pin, without determining the displacement. Anyway, lot's of interesting details to work out!

Best wishes,
Berend.

### drag computation

Quote:
> > If you calculate the drag of the hull and crew from a stationary

> > start, and for an entire (let's say) 2000m race while they are

> > moving with 6 dof, how will you know your predictions are as

> > good as, or better, than predictions made with other methods?

> Which 6 dof are you referring to? A typical fluid mechanics simulation for this case will involve 3 velocity components, pressure, and a parameter to describe the free interface. So 5 unknowns, but multiply this with about 10^7 mesh cells.

> Maybe you also mean the dof that control the position/rotation of the boat itself?

> For the unsteadiness, there are actually 2 factors: the first is the acceleration/deceleration of the boat, as I described above. The second factor is that because the rower moves relatively in the boat, the orientation of the boat (the pitch angle) will also change, which affects both steady and unsteady drag. The latter will especially contribute to wave drag.

> An important item when one performs a calculation is the validation, consisting of several parts: first of all the assumptions that go into the model (e.g. if you use a RANS model, how steady is the boundary layer actually?). The second one concerns the numerical error one makes, which can be verified by refining the mesh around the boat and see how much the solution changes. Thirdly is to test the model on a known solution, preferably an experiment. The latter is easily doable for a steadily moving boat, or a boat in a tow-tank, as there is a lot of empirical data for this. For a moving boat, I guess we could do our own experiment: if we create a velocity curve of a rowing boat (with an impeller) and measure the force exerted on the pins as a function of time, this could serve as a good validation. I know many of the measurements of the force on the pin or oar are referred to as "power" curves. But unless I am mistaken, I think they measure force. Power is force times displacement over time, and I don't see how you can measure this just from the pin, without determining the displacement. Anyway, lot's of interesting details to work out!

Sorry, for the confusion. I only have time for a few minutes here
every so often.

By 6dof I mean surge, heave and sway, yaw, pitch and roll, for both
the boat and the crew.

You are correct that force, not power, is measured at the pins.

How the oar-blade behaves in the water will certainly fully test your
RANS code, especially when combined with the hull and the moving crew.

Quads and eights are particularly difficult because there are so
many blades in the water at the same time, and so many moving human
bodies. It is quite tedious to just get the anthropometry of all the
crew, e.g. weights, heights, lengths and weights of body segments and
their centres of mass and inertia etc.

I use a completely different approach, in that I start with the simplest
useful model and then add complexity as needed. See:

almost complete, model of the hydrodynamics and aerodynamics of the shell and
crew.

Validation is going to be a very difficult exercise. The steady case of
a DTMB 5415 hull presented enormous logistical and other problems for
the ITTC when it had that hull tested in about 30 different towing tanks
around the world. See:
The Final Report of the Resistance Committee,
http://SportToday.org/

I think you should start with that hull before trying a rowing model because the entire exercise was designed to provide a validation set for CFD codes.
Prof. Frederick Stern deserves the highest praise for continuing to push for proper validation of CFD codes because there are, simply put, so many bullshit claims for the accuracy of the codes. Many practitioners tweak grids and empirical constants in order to promote better concordance with experiments. Not many are game to try "blind" validation exercises, for reasons on which we could speculate!

Here are some recent references to start you off on your quest:

Alexander Day, Ian Campbell, David Clelland, Lawrence J. Doctors,
and Jakub Cichowicz,
"Realistic evaluation of hull performance for rowing shells, canoes,
J. Sports Science, July 2011.

Doctors, L.J., Day, A.H. and Clelland, D.,
"Unsteady Effects During Resistance Tests on a Ship Model in a
Towing Tank", J. Ship Research, Vol 52, No. 4, Dec. 2008.

I wish you well in your research project. I'll be fascinated to hear how
long a simulation of a complete race will take using RANS.
The curves in my newsletter took about 2 secs to produce and I can simulate
an entire 2000m race in less than 1 minute on a PC. If you can get similar
accuracy for the velocity and acceleration in under an hour on a PC I will be
amazed, astounded and very impressed!

All the best,
Leo.

### drag computation

Quote:

>>> If you calculate the drag of the hull and crew from a stationary

>>> start, and for an entire (let's say) 2000m race while they are

>>> moving with 6 dof, how will you know your predictions are as

>>> good as, or better, than predictions made with other methods?

>> Which 6 dof are you referring to? A typical fluid mechanics simulation for this case will involve 3 velocity components, pressure, and a parameter to describe the free interface. So 5 unknowns, but multiply this with about 10^7 mesh cells.

>> Maybe you also mean the dof that control the position/rotation of the boat itself?

>> For the unsteadiness, there are actually 2 factors: the first is the acceleration/deceleration of the boat, as I described above. The second factor is that because the rower moves relatively in the boat, the orientation of the boat (the pitch angle) will also change, which affects both steady and unsteady drag. The latter will especially contribute to wave drag.

>> An important item when one performs a calculation is the validation, consisting of several parts: first of all the assumptions that go into the model (e.g. if you use a RANS model, how steady is the boundary layer actually?). The second one concerns the numerical error one makes, which can be verified by refining the mesh around the boat and see how much the solution changes. Thirdly is to test the model on a known solution, preferably an experiment. The latter is easily doable for a steadily moving boat, or a boat in a tow-tank, as there is a lot of empirical data for this. For a moving boat, I guess we could do our own experiment: if we create a velocity curve of a rowing boat (with an impeller) and measure the force exerted on the pins as a function of time, this could serve as a good validation. I know many of the measurements of the force on the pin or oar are referred to as "power" curves. But unless I am mistaken, I think they measure force. Power is force times dis

placement over time, and I don't see how you can measure this just from the pin, without determining the displacement. Anyway, lot's of interesting details to work out!

- Show quoted text -

Quote:

> Sorry, for the confusion. I only have time for a few minutes here
> every so often.

> By 6dof I mean surge, heave and sway, yaw, pitch and roll, for both
> the boat and the crew.

> You are correct that force, not power, is measured at the pins.

> How the oar-blade behaves in the water will certainly fully test your
> RANS code, especially when combined with the hull and the moving crew.

> Quads and eights are particularly difficult because there are so
> many blades in the water at the same time, and so many moving human
> bodies. It is quite tedious to just get the anthropometry of all the
> crew, e.g. weights, heights, lengths and weights of body segments and
> their centres of mass and inertia etc.

> I use a completely different approach, in that I start with the simplest
> useful model and then add complexity as needed. See:

> almost complete, model of the hydrodynamics and aerodynamics of the shell and
> crew.

> Validation is going to be a very difficult exercise. The steady case of
> a DTMB 5415 hull presented enormous logistical and other problems for
> the ITTC when it had that hull tested in about 30 different towing tanks
> around the world. See:
> The Final Report of the Resistance Committee,
> http://SportToday.org/

> I think you should start with that hull before trying a rowing model because the entire exercise was designed to provide a validation set for CFD codes.
> Prof. Frederick Stern deserves the highest praise for continuing to push for proper validation of CFD codes because there are, simply put, so many bullshit claims for the accuracy of the codes. Many practitioners tweak grids and empirical constants in order to promote better concordance with experiments. Not many are game to try "blind" validation exercises, for reasons on which we could speculate!

> Here are some recent references to start you off on your quest:

> Alexander Day, Ian Campbell, David Clelland, Lawrence J. Doctors,
> and Jakub Cichowicz,
> "Realistic evaluation of hull performance for rowing shells, canoes,
> and kayaks in unsteady flow",
> J. Sports Science, July 2011.

> Doctors, L.J., Day, A.H. and Clelland, D.,
> "Unsteady Effects During Resistance Tests on a Ship Model in a
> Towing Tank", J. Ship Research, Vol 52, No. 4, Dec. 2008.

> I wish you well in your research project. I'll be fascinated to hear how
> long a simulation of a complete race will take using RANS.
> The curves in my newsletter took about 2 secs to produce and I can simulate
> an entire 2000m race in less than 1 minute on a PC. If you can get similar
> accuracy for the velocity and acceleration in under an hour on a PC I will be
> amazed, astounded and very impressed!

> All the best,
> Leo.

Why would you calculate drag for an entire 2000 meter race? Could you
not just calculate boat drag for a set of different cases of boat
motions and use those calculations to make improvements in hull design
or rowing technique. I believe it should be possible to make decisions
based on those computations only. No need to do a full race and
incorporate calculations like on 8 rowing blades.

### drag computation

Quote:

> Why would you calculate drag for an entire 2000 meter race? Could you

> not just calculate boat drag for a set of different cases of boat

> motions and use those calculations to make improvements in hull design

> or rowing technique. I believe it should be possible to make decisions

> based on those computations only. No need to do a full race and

> incorporate calculations like on 8 rowing blades.

Thanks, Tinus.
Your suggestion is perfectly Ok if you are trying to design a
hull that is suitable for a fairly wide range of rowers. As you
say, all you need is a set of different cases. I'm not so sure
it is the best way to go for investigating technique and suggesting
changes.

IMO validating predictions is best done with a specific crew.
There are very few good sets of towing tank data on rowing shells
difficulty, particularly in small, short shallow tanks where the hulls
might not reach a steady state.
On-water experiments, of course, are very difficult to collect and
analyse because there are so many factors. To be useful, you must know
some specifics of who is in the boat at the time. I'm lucky that I have
lots of very good on-water data to play with.

Every computer modeller/analyst has their own preferences and
techniques. And it depends on what they are trying to achieve.

I prefer to look at the forces applied at different phases of the
race by different rowers. Each rower has a different force-time
profile during a race. Some are able to sustain high output from
the start to quite a way into the race; others tend to fade a little
more. Similarly, some rowers move their bodies in different ways at
the middle of a race to how they move at the end when they are
tiring. That affects the attitude of the boat.
Which commercially-available boats best suit that particular crew?
Some boats will give a particular crew an advantage at the start of
the race, when power output is high and speeds are also fairly high.
Other boats will be at a slight disadvantage at the start, but will
make up for that during the middle phases when their drag is a little
lower than other boats (given the changes in hull attitude during
that phase). Your proposed method would not be able to answer that
question.

Obviously that approach is not the best way to design a hull that is
applicable to a wide range of rowers, but that is not what I am
trying to do. I use different methods for analysing hull performance
with a "generic" crew. In that case it is important to know the
expected speed range of the class of rower you are designing for and
how long they can sustain the pace at various stages of the race. Giving
schoolkids a shell designed for an elite crew is just plain dumb.

Remember, too, that I do a lot of this sort of analysis for my own
entertainment. What can I say? I'm an inveterate nerd!

All the best,
Leo.

### drag computation

http://www.biorow.com/Quotes.htm

I wonder if there is a potential for boat manufacturers to offer a "tailoring service" for crews/scullers when they are picking a boat? It wold probably be more useful for scullers but imagine a system where a sculler could go and test row in a "standard" single set up with biomechanics and then the boat manufacturer could use that to recommend a boat shell shape. This would be especially useful for companies like empacher or Filippi who have a wide range of U shaped and V shaped hulls, and hulls with varying amount of buoyancy in the bows for scullers who "bounce" the boat a lot

### drag computation

Quote:

> http://SportToday.org/

> I wonder if there is a potential for boat manufacturers to offer a "tailoring service" for crews/scullers when they are picking a boat? It wold probably be more useful for scullers but imagine a system where a sculler could go and test row in a "standard" single set up with biomechanics and then the boat manufacturer could use that to recommend a boat shell shape. This would be especially useful for companies like empacher or Filippi who have a wide range of U shaped and V shaped hulls, and hulls with varying amount of buoyancy in the bows for scullers who "bounce" the boat a lot

Sounds like a perfectly valid way of selecting a boat that suits.
Of course, there are then a lot of other factors to consider, like how does that boat and crew handle waves and all the real stuff that makes rowing interesting and a ***y great challenge to analyse!

### drag computation

Quote:

>> http://SportToday.org/

>> I wonder if there is a potential for boat manufacturers to offer a "tailoring service" for crews/scullers when they are picking a boat? It wold probably be more useful for scullers but imagine a system where a sculler could go and test row in a "standard" single set up with biomechanics and then the boat manufacturer could use that to recommend a boat shell shape. This would be especially useful for companies like empacher or Filippi who have a wide range of U shaped and V shaped hulls, and hulls with varying amount of buoyancy in the bows for scullers who "bounce" the boat a lot

> Sounds like a perfectly valid way of selecting a boat that suits.
> Of course, there are then a lot of other factors to consider, like how does that boat and crew handle waves and all the real stuff that makes rowing interesting and a ***y great challenge to analyse!

I've been watching with interest this exchange between Berend & Leo.

I would echo Leo's comments on the problem of getting a CFD model to
replicate, without significant tweaking, the performance of a real boat.
This does not decry the merits of computational methods but does
reflect the fiendish complexity of the analytical problem.  And I note
in passing that the F1 teams still need costly wind-tunnels, even though

Rowing is even harder to fully replicate, I would suggest, than certain
aspects of race cars since we have 2 fluids where they have only air,
with a free-surface interface between them.  And, as both Leo and Berend
have noted these darned boats don't just go straight or at constant
speed.  They work in a real world with pitch, wind, waves & constantly
varying aero-drag - we think of ourselves as rowing through water only,
but we all know how much slower we go into a brisk headwind.

So why does rowing merrily discount all these relevant parameters, as
did the guys in the presentation in that Vespoli film?  If you go 20 sec
slower in an Olympic eight due to a headwind, this means that while we
assume (as a first approximation) that your total power output remains
constant, the power consumed in overcoming water drag is down by maybe
more than 16%, so 1/6th of what went into overcoming water resistance in
all its forms is now diverted into overcoming wind resistance.  Races
are often won by fractions of a second, so why aren't we giving more
thought to something which can cost each of us 20+ seconds per 2k on a
bad day?  Especially when there are some very simple and cheap ways to
reduce wind resistance, & excellent ways to cut leeway in crosswinds?

While some of us do have a serious appreciation of the fluid dynamics of
our sport, & Berend & Leo stand at the head of that group, it truly
astounds me that the sport at large has almost none & frankly couldn't
give a damn.  Too often we hear the old saw, "It's the same for everyone"?

The reality is that we want boats for all seasons & conditions, which
means accepting compromise solutions.  But it does not mean we have no
brains & that we cannot learn to adapt our equipment & technique to the
different conditions in which we may have to race.  Thus I have my
doubts about Tom Carter's quote from biorow.com .  The sad fact is that
very few rowers ever attempt to properly evaluate the range of available
boats but work on assumptions & whim.

Now suppose we made a clear understanding of how an oar actually works
in water a part of any advanced coaching course?  I'd suggest that might
be more useful than telling everyone the "correct" postures & styles to
follow, that they shouldn't immerse the oar-shaft & all the other
supposedly "right" things to do, things for which there is often a
dearth of technical substance.

Greg Searle was right - in the UK, anyway, rowing needs to learn from
the cyclists' adoption of their policy of 'the aggregation of marginal
gains' - a concept which I believe dates right back into the C19th, BTW.
That should apply to equipment & technique, & can only come from the
acceptance by the system that rowing is not just about physical
performance.  Real winners use their brains (& those of others) as well.

Cheers -
Carl

--
Carl Douglas Racing Shells        -
Write:   Harris Boatyard, Laleham Reach, Chertsey KT16 8RP, UK
Find:    tinyurl.com/2tqujf

### drag computation

Quote:

>> Dear John, It's actually a very good question. I think there is a lot of "confusion" or possibly ignorance when it comes to drag calculations on boats. I am not an expert on boats either; my area is fluid mechanics and computational fluid dynamics. I am not sure if I agree with everything stated in the video that you mention. For instance, they estimate "Aero drag" at 3% of the total drag, which I think is, for typical rowing conditions, an underestimation. Also, the "viscous drag" is mentioned as "the opportunity", where probably all fast rowing boats know how to deal with this: minimize the surface cross sectional area, thus a U shape under your seat. So it is more of a standard than an opportunity. If the viscous drag (I usually say friction drag) is really so ***, then the RANS and potential code they refer to is really not suitable - or at least not what does the calculation in a code. It's all boundary layer dynamics, for which a RANS code uses an empirical "log-layer"

profile. But this is, strictly speaking, only valid for a 1D fully developed steady boundary layer. The RANS model far away from the boundary layer is also inappropriate, as the Reynolds number is far too low there. You bring up two things which I have also often wondered about: the unsteady component of the drag, and the wave drag. I think the unsteady component is significantly larger than some people realize, as when you accelerate the boat, you need to pump the additional water in the boundary layer around it. Doing a very ballpark estimate of the boundary layer, approximating it as a boundary layer on a 4 meter long plate, I get a boundary Reynolds number of 10^7 and a boundary layer thickness of a few centimeters at the end of the boat. So that is a lot of additional water which needs to be moved around. You cannot estimate this effect from stitching two solutions from different instances together in a simulation. It has been commonly accepted that wave drag is very small for
rowing boats, but it might be a bit larger than some thought before. The reason why I think this, is because you actually do see some kind of bow-wave on a boat, e.g. http://SportToday.org/, is because some boat builders are trying to build "longer" boats; for instance, if you see the new Empacher single design, they try to elongate the bow deeper into the water for longer that they did a few years ago. This would be a typical change if they thought that wave drag plays a role and they try to minimize this by elongating the boat. In the simulation framework in the video you showed, these effects are all neglected; a single-phase steady RANS calculation. I wouldn't use it to design a boat. If I can get my hands on a good CAD drawing of a rowing single and a good student willing to work with me, I am happy to do some more state-of-the-art calculations. Sincere
ly, Berend van Wachem.

Quote:

> Berend,

> Thank you very much for starting off a fascinating conversation! I've been enjoying watching it develop.

<Good stuff snipped to address just 1 point>

Quote:
> How about this for an experiment:  the propulsive power and boat speed over the stroke cycle can be gotten with proper instrumentation for force at the pin and speed as a function of time for real rowing.  You would like to compare the stroke-averaged power with a steady-state towing power measurement at the same steady speed as the RMS (root-mean-squared) speed of the real rowing.  The difference would be the effects of unsteady flow.  I say RMS speed rather than the straight average speed because we would like to separate out the lowest-order effect of drag varying with about the square of the speed.  The RMS speed takes into account this basic dependence that would be included in a "stitched-together" steady-state model of the varying drag through the stroke.  So if the power needed in real rowing is higher than what would be needed to produce a steady speed equal to the RMS rowing speed (which is faster than the straight average), then the extra drag is due to other effects: u

nsteady flow effects.  Do you with real experience in such measurements think this is possible-  or, for that matter, has it already been done?
I'd venture to suggest, John, that whatever you might deduce from the
forces & velocities measured at the pins will get sorely muddied by the
fact that the blades, not the pins, are the point of propulsive
interaction of boat with water, & in that interaction there are also
substantial losses.

Further, there are interactions between rower & boat which continuously
affect the flow of energy between rower & boat, & between boat & water.

Nothing is impossible, everything should be tried, but the models - real
or virtual - aren't going to be easy to design or manage.

Maybe we should measure the actual motions in/through the water of a
boat that is being rowed, then impose exactly those motions in a ship
tank & thereby derive the real energy consumption?  And, taking another
of Leo's points, let that boat run for some distance to observe how long
it really takes to reach (or approach) a stable level of power
consumption.  That done, run the same boat with a corresponding crew
load at the same average speed to see how the power consumed then
differs.  But you then have to decide at what trim the boat should be run.

Cheers -
Carl

--
Carl Douglas Racing Shells        -
Write:   Harris Boatyard, Laleham Reach, Chertsey KT16 8RP, UK
Find:    tinyurl.com/2tqujf

### drag computation

Quote:

> I'd venture to suggest, John, that whatever you might deduce from the

> forces & velocities measured at the pins will get sorely muddied by the

> fact that the blades, not the pins, are the point of propulsive

> interaction of boat with water, & in that interaction there are also

> substantial losses.

> Further, there are interactions between rower & boat which continuously

> affect the flow of energy between rower & boat, & between boat & water.

> Nothing is impossible, everything should be tried, but the models - real

> or virtual - aren't going to be easy to design or manage.

> Maybe we should measure the actual motions in/through the water of a

> boat that is being rowed, then impose exactly those motions in a ship

> tank & thereby derive the real energy consumption?  And, taking another

> of Leo's points, let that boat run for some distance to observe how long

> it really takes to reach (or approach) a stable level of power

> consumption.  That done, run the same boat with a corresponding crew

> load at the same average speed to see how the power consumed then

> differs.  But you then have to decide at what trim the boat should be run.

> Cheers -

> Carl

Carl,  thanks for your post, I apologize for not being clear in mine.  Please tell me if I'm wrong, but it seems to me that the forces that propel the boat are applied at only two points:  the pin(s) and the footplate  (I'm assuming the force due to the seat rolling on the tracks is small enough to neglect).  It's only the component of force on the pins in the fore-and-aft direction that matters, so you would have to resolve that component with the instrumentation.  Likewise with the footplate force, only the fore-and-aft component contributes.  Sorry I left out the footplate force by mistake in my earlier post.  Then the actual power that goes into driving the hull is just the net force applied at those points multiplied by the speed of the boat through the water.  Of course, generally during the drive the footplate force is backward and subtracts from the pin force, while during the recovery the footplate force is forward and the pin force is small, with the blades out of the water. So the forces vary greatly through the stroke, but at any instant the net force is what drives the boat.

This point of view may seem odd, since the rower and her mass and motion do not appear explicitly in this measurement, but the forces she exerts on the boat certainly do appear, and that's all that matters to the boat  (for instance, during acceleration, say at the start of a race, the footplate force will be larger backwards, because the rower's mass has to be accelerated-  this all is accounted for by this measurement).   Anyway, this propulsive power has to balance the power dissipated by the drag forces, averaged over the stroke cycle, if you are going along at a steady pace.  So measuring the average propulsive power would give you directly- it is equal to- the average power dissipated by the drag.  This actual propulsive power is of course only part of what the rower has to put out-  there are all the various losses in the rower's body, and the loss due to the blades' inefficient coupling to the water,  friction at the pins, etc., that the rower also must supply, and which don't contribute to the forward motion of the boat.  But if you just want to know how much power is actually driving the boat, this is it -yes?  It has to be, there's nothing else pushing on the hull  (unless it's Nietzsche's famous demon... sorry, I couldn't resist).   So, if you could instrument to measure those pin and footplate forces, and measure the boat speed through the water, then you could do my experiment.   I thought from what has been said here that the pin and footplate force measurement techniques have already been worked out, and the speed is certainly available.  Yes??

I defer to you with experience in these measurements as to whether this is at all practical.

Your suggestion is certainly another possible method-  recording the speed and trim of the boat throughout the stroke cycle and then reproducing that while towing to measure the force.

Cheers,

John G

### drag computation

Quote:

>> I'd venture to suggest, John, that whatever you might deduce from the

>> forces & velocities measured at the pins will get sorely muddied by the

>> fact that the blades, not the pins, are the point of propulsive

>> interaction of boat with water, & in that interaction there are also

>> substantial losses.

>> Further, there are interactions between rower & boat which continuously

>> affect the flow of energy between rower & boat, & between boat & water.

>> Nothing is impossible, everything should be tried, but the models - real

>> or virtual - aren't going to be easy to design or manage.

>> Maybe we should measure the actual motions in/through the water of a

>> boat that is being rowed, then impose exactly those motions in a ship

>> tank & thereby derive the real energy consumption?  And, taking another

>> of Leo's points, let that boat run for some distance to observe how long

>> it really takes to reach (or approach) a stable level of power

>> consumption.  That done, run the same boat with a corresponding crew

>> load at the same average speed to see how the power consumed then

>> differs.  But you then have to decide at what trim the boat should be run.

>> Cheers -

>> Carl

> Carl,  thanks for your post, I apologize for not being clear in mine.  Please tell me if I'm wrong, but it seems to me that the forces that propel the boat are applied at only two points:  the pin(s) and the footplate  (I'm assuming the force due to the seat rolling on the tracks is small enough to neglect).  It's only the component of force on the pins in the fore-and-aft direction that matters, so you would have to resolve that component with the instrumentation.  Likewise with the footplate force, only the fore-and-aft component contributes.  Sorry I left out the footplate force by mistake in my earlier post.  Then the actual power that goes into driving the hull is just the net force applied at those points multiplied by the speed of the boat through the water.  Of course, generally during the drive the footplate force is backward and subtracts from the pin force, while during the recovery the footplate force is forward and the pin force is small, with the blades out of the wa

ter. So the forces vary greatly through the stroke, but at any instant the net force is what drives the boat.
Quote:

> This point of view may seem odd, since the rower and her mass and motion do not appear explicitly in this measurement, but the forces she exerts on the boat certainly do appear, and that's all that matters to the boat  (for instance, during acceleration, say at the start of a race, the footplate force will be larger backwards, because the rower's mass has to be accelerated-  this all is accounted for by this measurement).   Anyway, this propulsive power has to balance the power dissipated by the drag forces, averaged over the stroke cycle, if you are going along at a steady pace.  So measuring the average propulsive power would give you directly- it is equal to- the average power dissipated by the drag.  This actual propulsive power is of course only part of what the rower has to put out-  there are all the various losses in the rower's body, and the loss due to the blades' inefficient coupling to the water,  friction at the pins, etc., that the rower also must supply, and which d

on't contribute to the forward motion of the boat.  But if you just want to know how much power is actually driving the boat, this is it -yes?  It has to be, there's nothing else pushing on the hull  (unless it's Nietzsche's famous demon... sorry, I couldn't resist).   So, if you could instrument to measure those pin and footplate forces, and measure the boat speed through the water, then you could do my experiment.   I thought from what has been said here that the pin and footplate force measurement techniques have already been worked out, and the speed is certainly available.  Yes??

Quote:

> I defer to you with experience in these measurements as to whether this is at all practical.

> Your suggestion is certainly another possible method-  recording the speed and trim of the boat throughout the stroke cycle and then reproducing that while towing to measure the force.

> Cheers,

> John G

No apologies needed, John.  Like you, I'm always struggling towards the
light.

If we take the analogy of a racing car, would we deduce its aerodynamic
drag losses by using a dynamometer to measure the power delivered at the
drive to the prop shafts?  I think we know enough to say that there are
significant further energy costs in the generation of smoke, noise &
heat as a result of wheel spin & of tyre & suspension hysteresis.

It's the same in rowing, I think.....

Also, rowing carries its own KERS system.  It's not quite the same as in
F1, but it is still a cyclic energy storage & return system, using a
mobile & quite plastic system of interlinked body masses which we move
quasi-independently of the boat & at some energetic cost.  Over the
first part of the stroke we can't help giving a disproportionate amount
of energy to the rower, who accelerates faster over the water than the
boat does (or they wouldn't move from frontstops!).  In the rest of the
stroke, up to the finish, the boat is belatedly catching up with the
crew.  And during a chunk of the recovery the human KERS system is
moving relative to the boat to generate the force needed to sustain its
run despite hull friction.

It is extremely difficult to properly model the inertial dynamics &
energy flows, because we ain't simply towing the boat.  And I'd add as
an afterthought that, if ship-tank methods were to be applied, these
should separately measure the amount of energy spent on generating the
pitching motions.

Cheers -
Carl

--
Carl Douglas Racing Shells        -
Write:   Harris Boatyard, Laleham Reach, Chertsey KT16 8RP, UK
Find:    tinyurl.com/2tqujf