Semi-Monocoque Theory: corrective solutions



In [1]:

from pint import UnitRegistry
import sympy
import networkx as nx
import numpy as np
import matplotlib.pyplot as plt
import sys
%matplotlib inline
from IPython.display import display



Import Section class, which contains all calculations



In [2]:

from Section import Section



Initialization of sympy symbolic tool and pint for dimension analysis (not really implemented rn as not directly compatible with sympy)



In [3]:

ureg = UnitRegistry()
sympy.init_printing()



Define sympy parameters used for geometric description of sections



In [4]:

A, A0, t, t0, a, b, h, L, E, G = sympy.symbols('A A_0 t t_0 a b h L E G', positive=True)



We also define numerical values for each symbol in order to plot scaled section and perform calculations



In [5]:

values = [(A, 150 * ureg.millimeter**2),(A0, 250  * ureg.millimeter**2),(a, 80 * ureg.millimeter), \
(b, 20 * ureg.millimeter),(h, 200 * ureg.millimeter),(L, 2000 * ureg.millimeter), \
(t, 1.3 *ureg.millimeter),(E, 72e3 * ureg.MPa), (G, 27e3 * ureg.MPa)]
datav = [(v[0],v[1].magnitude) for v in values]



First example: Simple rectangular symmetric section

Define graph describing the section:

1) stringers are nodes with parameters:

• x coordinate
• y coordinate
• Area

2) panels are oriented edges with parameters:

• thickness
• lenght which is automatically calculated


In [6]:

stringers = {1:[(a+2*b,h),A],
2:[(b+sympy.Rational(1,2)*a,h),A],
3:[(sympy.Integer(0),h),A],
4:[(b,sympy.Integer(0)),A],
5:[(b+a,sympy.Integer(0)),A]}
#5:[(sympy.Rational(1,2)*a,h),A]}

panels = {(1,2):t,
(2,3):t,
(3,4):t,
(4,5):t,
(5,1):t}



Define section and perform first calculations



In [7]:

S1 = Section(stringers, panels)




In [8]:

S1.cycles




Out[8]:

$$\left [ \left [ 2, \quad 3, \quad 4, \quad 5, \quad 1, \quad 2\right ]\right ]$$



Plot of S1 section in original reference frame

Define a dictionary of coordinates used by Networkx to plot section as a Directed graph. Note that arrows are actually just thicker stubs



In [9]:

start_pos={ii: [float(S1.g.node[ii]['ip'][i].subs(datav)) for i in range(2)] for ii in S1.g.nodes() }




In [10]:

plt.figure(figsize=(12,8),dpi=300)
nx.draw(S1.g,with_labels=True, arrows= True, pos=start_pos)
plt.arrow(0,0,20,0)
plt.arrow(0,0,0,20)
#plt.text(0,0, 'CG', fontsize=24)
plt.axis('equal')
plt.title("Section in starting reference Frame",fontsize=16);






Plot of S1 section in inertial reference Frame

Section is plotted wrt center of gravity and rotated (if necessary) so that x and y are principal axes. Center of Gravity and Shear Center are drawn



In [11]:

positions={ii: [float(S1.g.node[ii]['pos'][i].subs(datav)) for i in range(2)] for ii in S1.g.nodes() }




In [12]:

x_ct, y_ct = S1.ct.subs(datav)

plt.figure(figsize=(12,8),dpi=300)
nx.draw(S1.g,with_labels=True, pos=positions)
plt.plot([0],[0],'o',ms=12,label='CG')
plt.plot([x_ct],[y_ct],'^',ms=12, label='SC')
#plt.text(0,0, 'CG', fontsize=24)
#plt.text(x_ct,y_ct, 'SC', fontsize=24)
plt.axis('equal')
plt.title("Section in pricipal reference Frame",fontsize=16);






Standard Solution



In [13]:

Tx, Ty, Nz, Mx, My, Mz, F, ry, ry, mz = sympy.symbols('T_x T_y N_z M_x M_y M_z F r_y r_x m_z')




In [14]:

S1.set_loads(_Tx=0, _Ty=Ty, _Nz=0, _Mx=Mx, _My=0, _Mz=Mz)





In [15]:

S1.compute_stringer_actions()
S1.compute_panel_fluxes();




In [16]:

S1.N




Out[16]:

$$\left \{ 1 : \frac{M_{x}}{3 h}, \quad 2 : \frac{M_{x}}{3 h}, \quad 3 : \frac{M_{x}}{3 h}, \quad 4 : - \frac{M_{x}}{2 h}, \quad 5 : - \frac{M_{x}}{2 h}\right \}$$




In [17]:

S1.q




Out[17]:

$$\left \{ \left ( 1, \quad 2\right ) : \frac{3 M_{z} + T_{y} a + T_{y} b}{6 h \left(a + b\right)}, \quad \left ( 2, \quad 3\right ) : - \frac{- 3 M_{z} + T_{y} a + T_{y} b}{6 h \left(a + b\right)}, \quad \left ( 3, \quad 4\right ) : - \frac{- M_{z} + T_{y} a + T_{y} b}{2 h \left(a + b\right)}, \quad \left ( 4, \quad 5\right ) : \frac{M_{z}}{2 h \left(a + b\right)}, \quad \left ( 5, \quad 1\right ) : \frac{M_{z} + T_{y} a + T_{y} b}{2 h \left(a + b\right)}\right \}$$



Compute L matrix: with 5 nodes we expect 2 dofs, one with symmetric load and one with antisymmetric load



In [18]:

S1.compute_L()




In [19]:

S1.L.subs(datav)




Out[19]:

$$\left[\begin{matrix}- \frac{1}{2} & \frac{2}{3}\\1 & 0\\- \frac{1}{2} & - \frac{2}{3}\\0 & 1\\0 & -1\end{matrix}\right]$$



Compute H matrix



In [20]:

S1.compute_H()




In [23]:

S1.H.subs(datav)




Out[23]:

$$\left[\begin{matrix}\frac{1}{2} & - \frac{4}{15}\\- \frac{1}{2} & - \frac{4}{15}\\0 & \frac{2}{5}\\0 & - \frac{3}{5}\\0 & \frac{2}{5}\end{matrix}\right]$$



Compute $\tilde{K}$ and $\tilde{M}$ as:

$$\tilde{K} = L^T \cdot \left[ \frac{A}{A_0} \right] \cdot L$$$$\tilde{M} = H^T \cdot \left[ \frac{l}{l_0}\frac{t_0}{t} \right] \cdot L$$


In [24]:

S1.compute_KM(A,b,t)




In [25]:

S1.Ktilde.subs(datav)




Out[25]:

$$\left[\begin{matrix}\frac{3}{2} & 0\\0 & \frac{26}{9}\end{matrix}\right]$$




In [26]:

S1.Mtilde.subs(datav)




Out[26]:

$$\left[\begin{matrix}\frac{3}{2} & 0\\0 & \frac{28}{15} + \frac{8 \sqrt{101}}{25}\end{matrix}\right]$$



Compute eigenvalues and eigenvectors as:

$$\left| \mathbf{I} \cdot \beta^2 - \mathbf{\tilde{K}}^{-1} \cdot \mathbf{\tilde{M}} \right| = 0$$

We substitute some numerical values to simplify the expressions



In [28]:

sol_data = (sympy.N(S1.Ktilde.subs(datav).inv())*(sympy.N(S1.Mtilde.subs(datav)))).eigenvects()



Eigenvalues correspond to $\beta^2$



In [29]:

β2 = [sol[0] for sol in sol_data]
β2




Out[29]:

$$\left [ 1.0, \quad 1.75937083803185\right ]$$



Eigenvectors are orthogonal as expected



In [30]:

X = [sol[2][0] for sol in sol_data]
X




Out[30]:

$$\left [ \left[\begin{matrix}1.0\\0\end{matrix}\right], \quad \left[\begin{matrix}0\\1.0\end{matrix}\right]\right ]$$



From $\beta_i^2$ we compute: $$\lambda_i = \sqrt{\frac{E A_0 l_0}{G t_0} \beta_i^2}$$

substuting numerical values



In [32]:

λ = [sympy.N(sympy.sqrt(E*A*b/(G*t)*βi).subs(datav)) for βi in β2]
λ




Out[32]:

$$\left [ 78.4464540552736, \quad 104.052378467824\right ]$$




In [ ]: