1
$\begingroup$

I have a cantilever beam with lots of elements. I am trying to find the sheer force and bending moment diagrams of the beam.

EVERY SINGLE TUTORIAL AND EXAMPLE I SAW ONLINE SHOWS HOW TO DO IT FOR A CANTILEVER BEAM WITH 2 ELEMENTS

But I want it for multiple elements.

I am taking n elements in my beam. I have calculated the global stiffness matrix = K and global mass matrix = M each of size (2n, 2n)

I was able to find the deformation by passing the static loads as a vecor of size (2n, 1) in the form

F = [f1, m1, f2, m2, f3, m3 ... ]

Where

fn is the point load at nth node and mn is the moment at the nth node

I find the transverse deflection and angular displacement at each node by

X = inv(K).F


Is there any such way to find the shear force and bending moment values for this case? Where n is very large (600 in my case)

I am doing it in python.

Thanks
$\endgroup$

1 Answer 1

0
$\begingroup$

Unfortunately space derivatives of displacement are not continuous the boundaries between two finite elements, so that you can't get a continuous distribution of strain, stress and internal actions across the beam simply evaluating the space derivatives of the FEM solutions.

Euler Beam - bending. On the other hand, it's possible to evaluate stress and internal actions in a structure using other methods. Here, I'll show you a way to recover reactions at constraints and internal actions for a cantilever Euler beam, of length $\ell$, clamped in $x=0$, with free end in $x=\ell$, whose transverse displacement $w(x,t)$ is governed

$m \ddot{w} - T' = f(x,t) \quad , \qquad x \in [0,\ell]$
$M' + T =0$
$w(0,t) = 0, \quad w'(0,t) = 0, \quad M(\ell,t) = 0, \quad T(\ell,t) = 0$

being $M(x,t) = (EJ(x,t)w(x,t))''$ the constitutive law for the bending moment, so that you can write the indefinite equilibrium equation and the boundary conditions at the free ends as

$m \ddot{w} + (EJ w'')'' = f(x,t) \quad , \qquad x \in [0,\ell]$
$EJw''(\ell,t) = 0, \quad (EJw''(\ell,t))' = 0$

Euler Beam - bending: Weak formulation. Now we can write the weak form of the equations

  • dynamical equation in terms of $w(x,t)$ only:

    $\displaystyle \int_0^\ell \delta w m \ddot{w} + \int_0^\ell \delta w'' EJ w'' = \int_0^\ell \delta w f$

  • equations for internal actions (I'm not 100% that integration by part is needed here, but you can try): $\displaystyle \int_0^\ell \delta \phi m \ddot{w} -\int_0^\ell \delta \phi T' = \int_0^\ell \delta \phi f \qquad \rightarrow \qquad \displaystyle \int_0^\ell \delta \phi m \ddot{w} - \underbrace{\delta \phi(\ell) T(\ell,t)}_{=0, \quad b.c.} + \delta \phi(0) T(0,t) + \int_0^\ell \delta \phi' T = \int_0^\ell \delta \phi f $
    $\displaystyle \int_0^\ell \delta \varphi M' + \int_0^\ell \delta \varphi T = 0 \qquad \rightarrow \qquad \underbrace{\delta \varphi(\ell)M(\ell,t)}_{=0, \quad b.c.} - \delta \varphi(0)M(0,t) -\displaystyle \int_0^\ell \delta \varphi' M + \int_0^\ell \delta \varphi T = 0$

having introduced test functions $\delta \phi$, $\delta \varphi$ for the equations of the internal actions.

Finite element problem. The equation for the displacement $w(x,t)$ can be solved introducing the finite element approximation $w(x,t) = \mathbf{N}_w^T(x) \mathbf{w}(t)$ and using the same test functions as the base functions $\mathbf{N}_w(x)$, i.e. the equation

$\displaystyle \int_0^\ell \mathbf{N}_w(x) m(x) \mathbf{N}_w^T(x) \mathbf{\ddot{w}}(t) + \int_0^\ell \mathbf{N}^{''}_w(x) EJ(x) \mathbf{N}_w^{''T}(x) \mathbf{w}(t) = \int_0^\ell \mathbf{N}_w f(x,t) $

$\mathbf{M} \mathbf{\ddot{w}} + \mathbf{K} \mathbf{w} = \mathbf{f}$

We can solve the equations of the internal actions introducing the approximation of the internal actions
$T(x,t) = \mathbf{N}^T_T(x) \mathbf{t}(t)$,
$M(x,t) = \mathbf{N}^T_M(x) \mathbf{m}(t)$

to get the discrete version of the weak formulation of the problem writeen above, that can be written as

$\mathbf{M}_\phi \mathbf{\ddot{w}} + \mathbf{C}_\phi \mathbf{t} = \mathbf{f}_\phi$
$\mathbf{M}_\varphi \mathbf{t} + \mathbf{C}_\varphi \mathbf{m} = \mathbf{0}$,

being matrices with the subscripts $\phi$ or $\varphi$ the finite element matrices computed using test functions $\phi$, $\varphi$.

These two equations, that should be the equations you need to evaluate internal actions, can be solved after the evaluation of the displacement field $w$, to recover internal actions.

Pay attention that $\mathbf{C}_\phi$ and $\mathbf{C}_\varphi$ must be invertible, to solve the system formally as

$\mathbf{t} = \mathbf{C}_\phi ^{-1} \left[ \mathbf{f}_\phi - \mathbf{M}_\phi \mathbf{\ddot{w}} \right] \qquad $ and thus
$\mathbf{m} = - \mathbf{C}^{-1} _\varphi\mathbf{M}_\varphi \mathbf{t} $,

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.