User:EAS4200C.F08.WIKI.A/Homework3

Curved Panels
Below is an image of a curved panel in the (x,y,z) coordinates.



Note: Sometimes (y,z) coordinates are used for the axes in plane of the cross section of the panel instead of (x,y) coordinates.

Below is an image of another curved panel with thickness t. A force will be applied to the panel which can be seen in the next image below.



Below is a cross sectional view of the same curved panel with force F acting on it at an angle above the horizontal (y-axis). The cross section goes from point A to point B. It's height is b and it's width is a.



$$d\overrightarrow{F} = q\overrightarrow{dl} = q(dl_yj + dl_zk)$$

$$d\overrightarrow{F} = q(dl\cos\theta \overrightarrow{j} + dl\sin \theta \overrightarrow{k})$$

$$\displaystyle\ dl\cos\theta = dy$$

$$\displaystyle\ dl\sin\theta = dz$$

$$\displaystyle\ q = \tau t $$ = shear flow

Here's a closer look at the panel where the force is acting. You can see it's acting on and parallel with section dl.



Closed Thin-Walled Cross Section
$$\overrightarrow{T} = T\overrightarrow{i} $$

Below is the cross section of a closed thin-walled area. There is a force acting on section dl.

$$\overrightarrow{r}$$ = the distance from the origin to section dl

$$\rho$$ = the distance from the origin to the force that is perpendicular to the force vector.



$$ d\overrightarrow{T} = \overrightarrow{r}xd\overrightarrow{F} $$

$$ dT = \rho dF $$

Take a closer look where the force acts on on section dl.

$$dA = \frac{1}{2}\rho dl $$

$$T = \oint_{}^{}{dT} = q\oint_{}^{}{\rho dl} $$


 * $$ \rho dl = 2dA $$

$$ T = 2q\int_{\overline{A}}^{}{dA}$$

$$T = 2q\overline{A}$$

$$\overline{A}$$ = Average Area

Below is an image of the average area of the cross section of a curved panel:



=Open Thin Walled Cross-Sections=



The torque of the depicted thin walled structure is

 $$\displaystyle T = 2q\bar{A}$$

This is also equivalent to the location of the resultant force acting parallel to the line connecting the two ends of the object.

=Uniform Circular Cylinder: Non-Warping=

We will now take a look at a uniform circular cylinder cross-section. This cross-section behaves as a rigid disk.



We will look at the infinitesimal area $$\displaystyle dA $$ at a distance $$\displaystyle r $$ away from the origin. The torque here can be described by the following equation;

$$\displaystyle T = \int\int_{A} r\tau dA $$

It must be noted that;

$$\displaystyle \tau = G\gamma $$, $$\displaystyle \tau dA = F $$

where $$\displaystyle G $$ is the strain modulus and $$\displaystyle \gamma $$ is the shear strain. Together they comprise Hooks Law which is used to describe the deformation of the disk. The variable $$\displaystyle F $$ is the force and when multiplied by the displacement produces the moment around the origin.The shear strain is described by the following equation;

 $$\displaystyle \gamma = r \frac{d\alpha}{dx}$$

where $$\displaystyle \frac{d\alpha}{dx} = \theta $$ which is defined as the rate of twist. Since $$\displaystyle \tau $$ is proportional to $$\displaystyle \gamma $$ which is also proportional to $$\displaystyle r $$, the shear stress $$\displaystyle \tau $$ gets smaller in the material closer to the core because the distance away from the core $$\displaystyle r $$ gets smaller. Therefore it is more efficient and cost effective to remove material to make the cylinder lighter. The limits of the material will be tested but if calculations are done correctly then there shouldn't be anything to worry about.

The torque can be represented by the following;



$$\displaystyle T = \int _A r G (r\theta) dA = G\theta\int _Ar^2dA $$

where $$\displaystyle dA $$ is a function of $$\displaystyle dy $$ and $$\displaystyle dz $$.



$$\displaystyle dA = dy * dz = r(dr)(dw)$$

The strain modulus and rate of twist is not a function of $$\displaystyle z $$ or $$\displaystyle y $$ and is why they come out of the integral. The resulting integral is defined as the second polar moment of inertia.

$$\displaystyle J = \int _Ar^2dA $$

=Hollow Circular Cross-Section: Thin Walled=

The analysis of a hollow circular cross section with thin walls (ie. t << a), is conducted as follows:

$$r_{i} = a$$ (inner radius)  $$r_{o} = b$$ (outer radius)

The equation for the polar moment is done by superimposing 'empty' circle on top of one solid circle, and the resulting equation is as follows:

 $$J = \frac{1}{2}\pi (b^{4} - a^{4})$$

This can be expanded as:

 $$J = \frac{1}{2}\pi (b - a)(b + a)(b^{2} + a^{2}$$

If an average radius of the cross section is then defined as:

 $$\bar{r} = \frac{a + b}{2}$$

And it is noted that, since t is very small,

$$b \cong a$$

and

$$ b \cong \bar{r}$$

and

<p style="text-align:center;">$$ a \cong \bar{r}$$

Therefore,

<p style="text-align:center;">$$ a^{2} \cong \bar{r}^{2}$$

$$ b^{2} \cong \bar{r}^{2}$$

and noting that $$b - a = t$$

J can then be expressed as:

<p style="text-align:center;"> $$J = 2\pi t\bar{r}^{3}$$

J can then be expressed in terms of the average radius by noting that:

<p style="text-align:center;">$$J = (2\pi ^{-\frac{1}{2}}t)(\pi \bar{r}^{2})^{3/2}$$

With $$\pi \bar{r}^{2} = \bar{A}$$ it's easily seen that J is proportional to $$\bar{A}^{3/2}$$, with $$(2\pi ^{-\frac{1}{2}}t)$$ being a proportionality factor.

=Solid Circular Cross-Section vs Hollow Thin Walled Cross-Section=

In order to compare these two cross-sections, we will be using the figure from the text Mechanics of Aircraft Structures by C.T. Sun. The figure is depicted below with all dimensions labeled on the figure.

<p style="text-align:center;">

We will first compute the area for each cross-section.

<p style="text-align:center;">$$\displaystyle  A_a = \pi r^2 = \pi (1cm)^2 = 3.1416 cm^2 $$

$$\displaystyle A_b = \pi r_o^2 - \pi r_i^2 = \pi (5.1cm)^2 - \pi (5cm)^2= 3.173cm^2$$

As the computtion shows, the areas are similar despite the size difference. We will next compute the polar moment of inertia for each cross section.

<p style="text-align:center;">

$$\displaystyle J_a = \frac{\pi r^4}{2} = \frac{\pi (1cm)^2}{2} = 1.5707cm^4 $$

$$\displaystyle J_b = \frac{\pi (r_0^4 - r_i^4)}{2} = \frac{\pi ((5.1cm)^4-(5cm)^4)}{2} = 80.927cm^4$$

Comparing the two Polar moment of inertia shows that the thin walled cross-section is a better torsional member than the solid cross-section. Since its polar moment of inertia (measure of its resistance to torsion) is much larger, it is a more desirable cross-section to use. The polar moment of inertia for the solid cross-section is only $$\displaystyle 2\% $$ of the polar moment of inertia for the hollow thin walled cross-section.

<p style="text-align:center;">$$\displaystyle \frac{J_a}{J_b} = \frac{1.5707cm^4}{80.927cm^4} = .0194 $$

If the thickness to equal $$\displaystyle .02 r_i^{(c)}$$, where $$\displaystyle c $$ represents a new hollow thin walled cross-section, what would this inner radious $$\displaystyle r_i^{(c)}$$ equal such that the polar moment of inertia for figure (a) equals that of the new figure (c)?

<p style="text-align:center;">

Matlab Project Discription
For HW 3, the team was asked to analyze a NACA 4-digit airfoil series. The discription of the problem can be seen here. A diagram showing the shape of the airfoil in this series can be found on this website. The group is asked to break up the airfoil into a number (ns) of discrete sections in order to analyze the geometry of the airfoil.

In order to complete this problem the student must first understand how to discretize an object into triangular segments in order to find the area of the object. A diagram of how this is done is provided below.

Using this, the following equations can be used in order to find the area of the shaded region.

$$d\vec{T} = \vec{r} \times d\vec{F} = q\vec{r} \times \vec{PQ}$$

Here, $$d\vec{F} = \vec{PQ}$$ and $$q\vec{r} \times \vec{PQ} = (2dA)\vec{i}$$

With

$$||\vec{PQ}|| = d\mathit{l} $$

Now that the basic foundation of the analysis has been discussed, the MATLAB code along with its execution is displayed below.

{| class="toccolours collapsible collapsed" style="width:100%" ! MATLAB Code

Airfoil Function
function Airfoil(m,p,ta,c,P_0)

m = m/100; p = p/10; ta = ta/100; check = 1; j = 1;

while check ==1 ns = 20*j; y = linspace(0,1,ns);

for i = 1:ns if y(i)<=p z_c(i) = (m/p^2)*(2*p*y(i)-y(i)^2); theta(i) = atan((m/p^2)*(2*p-2*y(i))); else z_c(i) = (m/(1-p)^2)*((1-2*p)+2*p*y(i)-y(i)^2); theta(i) = atan((m/(1-p)^2)*(2*p-2*y(i))); end z_t(i) = 5*ta*(.2969*y(i)^.5-.1260*y(i)-.3516*y(i)^2+.2843*y(i)^3-.1015*y(i)^4);

y_u(i) = (y(i) - z_t(i)*sin(theta(i)))*c; z_u(i) = (z_c(i) + z_t(i)*cos(theta(i)))*c;

y_l(i) = (y(i) + z_t(i)*sin(theta(i)))*c; z_l(i) = (z_c(i) - z_t(i)*cos(theta(i)))*c; end

Area = 0;

for i = 1:ns-1 PQ = [0 y_u(ns-i)-y_u(ns-i+1) z_u(ns-i)-z_u(ns-i+1)]; r = [0 y_u(ns-i+1)-P_0(2) z_u(ns-i+1)-P_0(3)]; a = cross(r,PQ)/2; Area = Area + a;   end

for i = 1:ns-1 PQ = [0 y_l(i+1)-y_l(i) z_l(i+1)-z_l(i)]; r = [0 y_l(i)-P_0(2) z_l(i)-P_0(3)]; a = cross(r,PQ)/2; Area = Area + a;   end

NS(j) = ns; A_BAR(j) = Area(1);

if j ~= 1 if abs((A_BAR(j)-A_BAR(j-1))/A_BAR(j-1))*100<.01 check = 0; end end j = j+1; end

A_bar = A_BAR(j-1)

subplot(2,1,1), plot(y_u,z_u,y_l,z_l), xlabel('y'), ylabel('z'), title('NACA 2415 Airfoil')

subplot(2,1,2), plot(NS,A_BAR), xlabel('ns'), ylabel('A_b_a_r'), title('ns vs. A_b_a_r')

NACA 2415 Airfoil in MATLAB command window
EDU>> Airfoil(2,4,15,.5,[0 0 0])

A_bar =

0.0253

EDU>> Airfoil(2,4,15,.5,[0 0.19 0.01])

A_bar =

0.0255

EDU>> Airfoil(2,4,15,.5,[0 0.25 -1])

A_bar =

0.0256


 * }

Torsion
Torsion of uniform, non circular bars leads to warping of the cross section, which is defined as an axial displacement along the x-axis (ie, along the bars length due to the conventions used in class) of a point on the deformed, or rotated, cross section. A schematic representation is shown below.

$$u_{y}$$ = the y-component of the displacement vector $$\vec{PP^{'}}$$

and

$$u_{z}$$ = the z-component of the displacement vector $$\vec{PP^{'}}$$

It should be noted (and easily seen) from the figure that $$\vec{PP^{'}}$$ is not on a circle centered at the origin with radius $$\vec{OP^{'}}$$ and $$\vec{OP^{'}}$$ in fact has a larger magnitude than $$\vec{OP}$$

This is due to the line $$\vec{PP^{'}}$$ having a vertical and horizontal displacement from $$\vec{PP^{'}}$$

However, since the angles associated with twist in torsion are likely to be relatively small, one can use small angle approximations and assume that the magnitude of $$\vec{OP} = \vec{OP^{'}} = R$$

$$u_{z} = Rsin\alpha \cong R\alpha$$

$$u_{y} = R(1 - cos\alpha )$$

With $$\alpha = 0$$, this makes $$u_{y} = 0$$ and $$u_{z} = \alpha R$$

The rate of twist angle, $$\displaystyle \theta $$, is defined as


 * $$\displaystyle \theta = {\alpha \over x} $$

Where $$\displaystyle \alpha $$ is the angle of rotation. The horizontal displacement (displacement in the y-direction) is


 * (1) $$\displaystyle u_y = -\theta x z $$

The vertical displacement of a point of the cross section (displacement in the z-direction) can be determined from


 * $$\displaystyle u_z = +(PP') \cos \beta = +\alpha \zeta_p $$

Where $$\displaystyle (OP)\alpha $$ can be substituted in place of $$\displaystyle PP' $$ and $$\displaystyle \theta_x $$ can be substituted in place of $$\displaystyle \alpha $$, thus resulting in the following


 * (2) $$\displaystyle u_z = +\theta x y $$

The displacement due to warping along the x-axis is defined as


 * (3) $$\displaystyle u_x = \theta \psi (y,z) $$

Equations 1, 2, and 3 make up the kinematic assumptions.

=Roadmap for Torsional Analysis=

The following list lays out a general step by step roadmap of equations to apply for the torsional analysis of an aircraft wing:


 * A. Kinematic assumptions (equations 1, 2, and 3 above)
 * B. Strain-displacement relation given by the two equations:
 * $$\displaystyle \gamma_{yz} = {\partial u_x \over \partial y} + {\partial u_y \over \partial x}$$
 * $$\displaystyle \gamma_{zx} = {\partial u_x \over \partial z} + {\partial u_z \over \partial x}$$
 * C. Equilibrium equations for stresses:
 * $$\displaystyle {\partial \sigma_{yy} \over \partial y} + {\partial \tau_{zy} \over \partial z} + {\partial \tau_{xy} \over \partial x} = 0 $$
 * $$\displaystyle {\partial \tau_{yz} \over \partial y} + {\partial \sigma_{zz} \over \partial z} + {\partial \tau_{xz} \over \partial x} = 0 $$
 * $$\displaystyle {\partial \tau_{yx} \over \partial y} + {\partial \tau_{zx} \over \partial z} + {\partial \sigma_{xx} \over \partial x} = 0 $$
 * But, since $$\displaystyle \tau_{yx} $$ and $$\displaystyle \tau_{zx} $$ are independent of x in the displacement field, the equations reduce to the following:
 * $$\displaystyle {\partial \tau_{yx} \over \partial y} + {\partial \tau_{zx} \over \partial z} = 0 $$
 * D. Prandt stress funtion $$\displaystyle \phi (y,z) $$:
 * $$\displaystyle \tau_{yx} = {\partial \phi \over \partial z} $$
 * $$\displaystyle \tau_{zx} = -{\partial \phi \over \partial y} $$
 * E. Strain compatibility equation:
 * $$\displaystyle {\partial \gamma_{zx} \over \partial y} - {\partial \gamma_{yx} \over \partial z} = 2 \theta $$
 * F. Equation for $$\displaystyle \phi $$
 * $$\displaystyle {\partial^2 \phi \over \partial y^2} - {\partial^2 \phi \over \partial z^2} = -2G \theta $$
 * G. Boundary conditions for $$\displaystyle \phi $$:
 * The traction free boundary condition $$\displaystyle t_x = 0 $$ on a lateral surface yields the following:
 * $$\displaystyle {d \phi \over ds} = 0 $$
 * thus, $$\displaystyle \phi $$ is a constant. But, this constant becomes arbitrary since it is a solid section with only one contour boundary, thus
 * $$\displaystyle \phi = 0 $$
 * H. Resultant torque:
 * $$\displaystyle T = 2 \iint_A \phi dA $$
 * I. $$\displaystyle T = G J \theta $$ where
 * $$\displaystyle J = -{4 \over \triangledown^2 \phi} \iint_A \phi dA $$
 * $$\displaystyle T = 2q \bar A $$
 * J. Twist angle
 * $$\displaystyle \theta = {1 \over 2G \bar A} \oint {q \over t}ds $$
 * where $$\displaystyle s $$ is the curvilinear coordinate along a thin wall


 * K. 1) Multicelled Section


 * For cells where $$i = 1, ..., n_{cell}$$


 * [[Image:3cellsection.JPG|500px]]


 * $$\tau = 2\sum_{i=1}^{N_c}{q_i \bar{A_i}} $$


 * $$q_i \;$$ = Shear flow in cell i.


 * $$\bar{A_i}$$ = "average" area in cell i.


 * Define: $$T_i = 2q_i\bar{A_i}$$ torque generated by one cell.


 * $$\tau = 2\sum_{i=1}^{N_c}{q_i \bar{A_i}} $$


 * $$T_i = \sum_{i=1}^{N_c}{T_i} $$


 * 2) Shape of airfoil is "rigid" in the plane $$(X,Y)$$ however it can warp out of the plane.


 * $$\theta = \theta _1 = \theta _2= ...=\theta _{N_c}$$

The twist angle equation applied to cell i.

$$\theta_i =\frac{1}{2 G_i \bar{A_i}}\oint_{}^{}{\frac{q_i}{t_i}ds}$$


 * $$G_i\;$$ = Shear Modulus of cell i


 * $$t_i(s)\;$$ = thickness of walls on cell i. This is a linear curve coordinate along the cell wall.

''' Pb 1.1: Rectangular single cell section with modification. '''

Now more general with a single cell section:

$$t_i=0.08 \; m$$

$$t_2= t_3=0.01 \; m$$



$$\bar{A}=\frac{1}{2} \pi (\frac{b}{2})^2 +\frac{1}{2}ba= 5.5708$$

Shear flow comes from the torque equation: $$T=2q\bar{A}$$

$$q= \frac{T}{2\bar{A}}$$

Twist angle: $$\theta_i =\frac{1}{2 G_i \bar{A_i}}\sum_{j=1}^{3}{\frac{q_jl_j}{t_j}}$$ $$j=1,2,3$$ index for the segment number.

There is only one cell for $$\theta_i\;$$ so it reduces to $$\theta\;$$ in the equation above. You would normally have two indexes for multicelled section $$(i,j)\;$$.

$$\theta =\frac{1}{2 G \bar{A}}\sum_{j=1}^{3}{\frac{q_jl_j}{t_j}}$$

Note: $$\int$$ is and elongated S, standing for continuous summation.

$$\sum{}$$ is a discrete sum.

=Individual Contributions= Christopher PergolaEAS4200C.F08.WIKI.A 18:14, 8 October 2008 (UTC)

Braden Barnes Eas4200c.f08.wiki.b 19:54, 7 October 2008 (UTC)

Eric Viale Eas4200c.f08.wiki.c 22:15, 7 October 2008 (UTC)

Michael Rolle Eas4200c.f08.wiki.d 18:29, 8 October 2008 (UTC)

Jeff Huenink Eas4200c.f08.wiki.f 03:10, 8 October 2008 (UTC)

Chris Fontana Eas4200c.f08.WIKI.E 13:11, 8 October 2008 (UTC)