Sunday, April 12, 2009

How much energy does sailing in chop take?




Another perfect sailing day, but still on the DL for another 4 weeks. G. mentioned wanting to filter out the chop input to the flap, so I added chop to the stability model to see the effect. The plot above is the work ( drag times distance ) for the foils in the case of a 3 ft wave and a 3ft wave with 0.5 ft chop. Vertical axis in ft-lbs (1 ft-lb = 1.35 Joules), horizontal in feet. The spike represent the extra work due to response to the chop by the flap.




It does seem like filtering out the high-frequency could be a good idea....all the extra work has to be slowing the boat down. But any filter will slow the response (e.g. to sharp waves) down so there will be some optimization.

There was one question about the work in the chop ( green line) case, so I plotted with a better scale. The wave and chop start at 200 ft. Before that, the work in the green/blue cases are the same. After the chop starts, the work by the flap responding to the chop is taking more energy out of the boat.
This model has some simplification so it is really not good for predicting absolute performance like the one Alan posted in Doug Culnane's website. I just ran this case to get an idea of the energy going into the response to chop, which ideally the control system would ignore.
Just need to figure out a way to implement filters without op-amps and batteries.

Monday, April 6, 2009

If one is good, are two better?


The poor performance of elevator control led me to try some other schemes. Leaving out the ones that did not work, the above graph compares the response of a flap only control scheme to a scheme that moves the elevator in the same direction as the flap, but at half the gain.


It is not clear that this is a better scheme but it does get all the control forces acting in the direction that the boat wants to move when the wand senses the trough.


The trough is an isolated half-sine of depth 3 ft and width 30 ft (1 m by 10 m).

Friday, April 3, 2009

Findlay/Turnock VPP from Uni. of Southampton

Just in case you have not seen it, this is a link to a paper from Findlay/Turnock on a velocity prediction program written in Basic/Excel. It does a good job explaining the inputs to a VPP and has a nice flow diagram for the logic of the program.

http://eprints.soton.ac.uk/52462/01/Findlay_Turnock_Foil_VPP.pdf

Thursday, April 2, 2009

Accuracy of inputs


We had some off-line discussion about how to get a good moment of inertia for the dynamics model. Bifilar pendulum is a standard way. The Finn class uses the Lamboley swing method which can be found here ( www.lamboleyetudes.net). I decided to run the model with +/- 50% pitch inertia change from Alan's original estimate of 20 slug-ft-ft to see if it is sensitive to inertia changes.

Result is shown above for the case of elevator control. Although the high inertia case (30 slug-ft-ft) shows the expected slower response, the basic character of the elevator control ( height goes up before it goes down) does not change.

Still, if someone wants to measure inertia (Lamboley seems the easiest method) it would be interesting to know a more accurate value.

Sunday, March 29, 2009

Elevator or Flap Control?




For height control on a foiling Moth the usual practice to to control the AOA and CL of the main foil with a flap connected to a wand that senses the height of the boat.
One question is whether elevator control, i.e. use the rear foil to change the pitch of the main foil, is viable alternative. This is how airplanes control the AOA and maintain level flight.
I used the stability model (from alans of AUS) to compare elevator control to flap control.
The result is that elevator control seems fundamentally slower than flap control. The graph shows the response of the two different control systems to a artificial step in the water. Step response is the typical way to look at the dynamic response of a control system.
The red curve is the step in the water. The blue and green curves are the response of the elevator control and flap control models as represented by the depth of the main foil.
The flap control model starts dropping as soon as the wand (5 ft in front of the foil) sees the step. On the other hand, the elevator control model rises a little before it starts to drop.
I interpret this result this way: when the wand senses the step, the elevator is pitched up to make the main foil AOA go down. The pitch up of the elevator increases the lift contributed by the elevator which raises the whole boat. Later, the pitch down of the main foil reduces the main foil lift and the boat drops. But the end result is that elevator controlled boat reacts slower than the flap controlled boat.
(note: graph units are feet in both vertical and horizontal axes).

Friday, March 27, 2009

What does wand lead do (AKA midship wand)?


An internet friend gave me his stability model, which is based on using Runge-Kutta integration on the equations of motion (i.e. F= ma, etc).
I used it to do some parameter studies for the response of a flying Moth to holes in the water.

The first plot is the response of an elevator ( tail foil ) controlled model, for several different wand leads. Wand lead is the distance from the main foil to the point that the wand intersects the water. Distances in vertical and horizontal axes are in feet ( = 0.3 m ) and the boat is moving at about 30 f/s.
First conclusion is that more wand lead is good.