Troubleshooting Airfoil Calculation Using Joukowski Transformation

AI Thread Summary
The discussion focuses on troubleshooting a fluid mechanics homework assignment involving airfoil pressure coefficient distribution using Joukowski transformation. The method includes defining parameters, transforming coordinates, and applying Bernoulli's equation to calculate pressures. Initially, the user encountered discontinuities in results when adding camber, but later realized the need to calculate pressures directly in the Zeta plane before transforming back to X-Y space. The user seeks confirmation on the appropriateness of this revised approach and requests assistance with MATLAB code modifications. The conversation highlights the importance of correct coordinate transformations in fluid dynamics calculations.
danja
Messages
13
Reaction score
0
I'm having a bit of trouble with a homework assignment for a fluid mechanics course. I'd like to ask if my solution method is appropriate. The goal of the assignment is to determine the pressure coefficient distribution along the upper and lower airfoil surfaces and plot it (I'm using MATLAB for this). To do so, we're supposed to use the joukowski transformation.

Allow me to define quantities of interest:

a = circle radius
L = airfoil chord length
c = L/4
h = camber
t = airfoil thickness
m*exp(i*delta) = -0.77*t*c/L + i*h/2 (parameter in the complex potential function).

Transforming from Zeta plane (circle) to Z plane (normal shape).

In order to obtain the pressure coefficient, I am using the following procedure:

1) Determine values for a, c, h, m*exp(i*delta) for the desired conditions
2) Calculate X and Y locations along upper and lower surfaces
3) Transform X and Y data into the Zeta plane using the quadratic solution to Z(zeta)
4) Take the derivative of F(zeta) and use the data from steps 1 and 3 to
determine W(zeta) for all X and Y
5) Transform W(zeta) into W(z) by applying the operation: W(z) =
W(zeta)/dZ/dZeta
6) Apply Bernoulli's equation while using the real parts of W(z) as the
local velocity to find P(x)
7) Calculate Cp(x) based on P(x).

Results appear to be correct (in trend anyways) for the case when camber is
zero. But when I add camber, I am getting a discontinuity along the lower
surface. I have already set the appropriate roots of the quadratic equation
used in step 3 so the plots are using the right pieces. I can post screen shots if it helps.

I'd really appreciate a little help on this. Is this an appropriate method?
Or is it better to try and do everything in the Z-plane instead?

I've already been through this thread:
https://www.physicsforums.com/showthread.php?t=365978&highlight=joukowski+pressure

But it doesn't help me too much.

Matlab code:
Code:
close all; clear; clc;

camb_per = 0.0002;
t_per = 0.12;
at = 3;

L = .5;
U = 50;
rho = 1.2;
Pinf = 101325;

h = camb_per*L;
t = t_per*L;
c = L/4;
M = 1i*h/2 - 0.77*t*c/L;
a = c + 0.77*t*c/L;
at = at*pi/180;

% Circulation
Gam = pi*U*L*(1 + 0.77*t/L)*sin(at + 2*h/L);
K = Gam/(2*pi);

airshapeU = @(x) sqrt((L^2/4)*(1 + L^2/(16*h^2) ) - x^2) - L^2/(8*h) + 0.385*t*(1 - 2*x/L)*sqrt(1 - (2*x/L)^2);
airshapeL = @(x) sqrt((L^2/4)*(1 + L^2/(16*h^2) ) - x^2) - L^2/(8*h) - 0.385*t*(1 - 2*x/L)*sqrt(1 - (2*x/L)^2);

%fplot(airshapeU,[-L/2 L/2]);
%hold on;
%fplot(airshapeL,[-L/2 L/2]);

[XU YU] = fplot(airshapeU,[-L/2 L/2],800);
[XL YL] = fplot(airshapeL,[-L/2 L/2],800);

% Determine complex coordinates
ZU = double(XU + 1i*YU);
ZL = double(XL + 1i*YL);
Z = ZU;

for i = 1 : 2

    if i == 2
        Z = ZL;
    end

% Map to Zeta space
JP = 0.5*Z + ((Z.^2)/4 - c^2).^0.5;
JN = 0.5*Z - ((Z.^2)/4 - c^2).^0.5;

% Evaluate Complex Potential
WTP = U*exp(-1i*at) - U*a^2*exp(1i*at)./(JP - M).^2 + 1i*K./(JP - M);
WTN = U*exp(-1i*at) - U*a^2*exp(1i*at)./(JN - M).^2 + 1i*K./(JN - M);

% Map back to Z space
WP = WTP./(1 - c^2./JP.^2);
WN = WTN./(1 - c^2./JN.^2);

% Calculate Pressure Distribution
PP = Pinf + 0.5*rho*(U^2 - real(WP.^2));
CPP = (PP - Pinf)/(0.5*rho*U^2);

PN = Pinf + 0.5*rho*(U^2 - real(WN.^2));
CPN = (PN - Pinf)/(0.5*rho*U^2);


% Set plotting information
for k = 1 : size(XU,1)
    if XU(k) < 0
        WU(k) = WN(k);
        XN(k) = (XU(k)+L/2)/L;
    else
        WU(k) = WP(k);
        XN(k) = (XU(k)+L/2)/L;
    end
end

P = Pinf + 0.5*rho*(U^2 - real(WU.^2));
CP(:,i) = (P - Pinf)/(0.5*rho*U^2);

end

plot(XN,CP(:,1));
hold all;
plot(XN,CP(:,2));
set(gca,'YDir','reverse')
axis([0 1 -1 1]);
 
Engineering news on Phys.org
Screen shots of what's happening:

With no camber, 6 deg angle of attack:
11jowhy.png


With camber of 4%, 6 deg angle of attack:
szza82.png
 
Well I might just be an idiot. I was trying to map X-Y coordinates into the Zeta plane to determine the pressures at those coordinates. What I should've been doing was just getting the pressures along the curve of the circle in the Zeta plane and then translating them back to X-Y space. I think it's working now. Will report back.
 
Does this look more correct?

With no camber, 6 deg angle of attack:
2u4ix6p.png


With 4% camber, 6 deg angle of attack:
2elfhh2.png
 
hi danja,
Can you please post the modified correct code?
thanks in advance.
 
I need some assistance with calculating hp requirements for moving a load. - The 4000lb load is resting on ball bearing rails so friction is effectively zero and will be covered by my added power contingencies. Load: 4000lbs Distance to travel: 10 meters. Time to Travel: 7.5 seconds Need to accelerate the load from a stop to a nominal speed then decelerate coming to a stop. My power delivery method will be a gearmotor driving a gear rack. - I suspect the pinion gear to be about 3-4in in...
Thread 'Calculate minimum RPM to self-balance a CMG on two legs'
Here is a photo of a rough drawing of my apparatus that I have built many times and works. I would like to have a formula to give me the RPM necessary for the gyroscope to balance itself on the two legs (screws). I asked Claude to give me a formula and it gave me the following: Let me calculate the required RPM foreffective stabilization. I'll use the principles of gyroscopicprecession and the moment of inertia. First, let's calculate the keyparameters: 1. Moment of inertia of...
Thread 'Turbocharging carbureted petrol 2 stroke engines'
Hi everyone, online I ve seen some images about 2 stroke carbureted turbo (motorcycle derivation engine). Now.. In the past in this forum some members spoke about turbocharging 2 stroke but not in sufficient detail. The intake and the exhaust are open at the same time and there are no valves like a 4 stroke. But if you search online you can find carbureted 2stroke turbo sled or the Am6 turbo. The question is: Is really possible turbocharge a 2 stroke carburated(NOT EFI)petrol engine and...
Back
Top