True wind with leeway.

[ieb]

Hi,
I see the True wind doesn't account for leeway. Not sure if it should, but here is some code (originally in C++ running on an ArduinoDue) that does the calc. It uses a slightly different approach to the current True Wind calc. Has been checked and tested, but might have bugs.

function calcTrueWind(awa, aws, stw, leeway) {
  if ( leeway === undefined) {
    leeway = 0;
  }
  awa  = fixAngle(awa);
  var stw_lee = stw*Math.cos(leeway);
  var awa_lee = awa;
  if ( awa_lee > 0 ) {
    awa_lee = awa_lee +  leeway;
  } else {
    awa_lee = awa_lee +  leeway;         
  }
  // this should be a noop, but just in case the leeway downwind caused something wierd.
  awa_lee = fixAngle(awa_lee);

  var ctws = Math.sqrt((stw_lee*stw_lee+aws*aws)-(2*stw_lee*aws*Math.cos(awa_lee)));
  var ctwa = 0.0;
  if ( ctws > 1.0E-3 ) {
    ctwa = (aws*Math.cos(awa_lee)-stw_lee)/ctws;
    if ( ctwa > 0.9999 || ctwa < -0.9999) {
      ctwa = 0.0;
    } else {
      ctwa = Math.acos(ctwa);
    }
  }
  if ( awa_lee < 0.0) {
    ctwa = -ctwa;
  }
  return {
    angle : ctwa,
    speed : ctws
  };
}
function fixAngle(d) {
  if ( d > Math.PI ) d = d - Math.PI;
  if ( d < -Math.PI) d = d + Math.PI;
  return d;
}

[sbender9]

How are you getting leeway?

[joabakk]

Leeway could be transverse speed as given from the newer logs. If not defined, it defaults to 0

[ieb]

Airmar do have some 2D ultrasonic logs. The earlier generation of these was not particually reliable. The new generation are better but cost around 1000 USD last time I looked. If I had one, I would use that source of data.

I am using leeway estimated from heel, derived from a low cost ($5 ebay) I2C 10Dof board . For Arduino the library support is not great and lacks callibration, for Node.js there is an excellent npm based on RTIMULib2 which allows a calibration procedure and is simple to use.

leewayAngle = Kfactor * roll / ( waterspeed ^2)

KFactor depends on the hull shape, the units of roll and waterspeed The estimate is from Pedrick's paper published in 1981, which is fairly standard. There are more complex ways for what I need this good enough.

[joabakk]

I see from https://arvelgentry.jimdo.com/app/download/9157993883/Arvel+Gentry+-+Sailboat_Performance_Testing_Techniques.pdf?t=1485748085 that the constant varies from 9 to 16. That's another caclulation to add!