[m_zanichelli]

Hello everybody,
I am an aerospace Engineer, I am currently working in an airfoil optimization project. I want to create an airfoil parametrization starting from a RAE2822 of which I know the coordinates.
At the beginning I was trying to get a unique B-spline representation for the whole airfoil curve, i.e. to find a unique curve that fits the airfoil curve. I have not been able to do so (does anybody have an idea?) . Anyway, I found the spap2 function of Matlab and I've been using it to find the B-spline best-fit to the coordinates I have, keeping separate the lower and upper part of the airfoil. Trying some different knot sequences and found some that seem promising, but I am getting a weird behaviour at the Leading Edge. The B-spline curves follow really accurately the original curve but show a "pointy" behaviour. I am now attaching my matlab script.

filename = 'RAE2822.txt'; % attached here
coords = readtable(filename);
x_0 = coords.Var1;
y_0 = coords.Var2;
% separation of upper and lower parts of the airfoil
xaxis = sort(x_0(1:65));
x_1 = x_0(1:65);
y_1 = y_0(1:65);
x_2 = [0; x_0(66:end)];
y_2 = y_0(65:end);
% plot of the original coordinates
plot(x_1, y_1, x_2, y_2); hold on;
% definition of the knot sequence
x = [0, 0.025, 0.05, 0.2, 0.55, 1];
degree = 3 %b - spline degree
knots = augknt(x,degree+1);
%weights
w1 = ones(size(x_1));                               %
w1(end) = 1000;
w2 = ones(size(x_2));
w2(1) = 1000;
%least squares b-spline fit of the suction side SS
BS_SS = spap2(knots, degree+1, x_1, y_1,w1);
fnplt(BS_SS); 
%least squares b-spline fit of the pressure side PS
BS_PS = spap2(knots, degree+1, x_2, y_2, w2);
fnplt(BS_PS); axis equal
xlim([-0.01, 0.05]) % comment if you want to see the whole parametrization. Here you can see the problem

I am guessing this problem comes from the large derivative I have at the beginning of the curve, but I am not really sure about that. Can somebody help ?

[mluckyw]

Have you tried to use camber line as your line of reference and not coordinates? It will better to resemble the airfoil curves but for exchange of added variables of thickness for each point in the curve.