0
$\begingroup$

Using the Ritz method to find the displacement field in structural analysis can be done as follows. $U$ and $V$ are recpectively the internal and external energy components of a given structural element:

$$U+V=W$$

Expressing $U$ in terms of the strains $\varepsilon$ and the material consitutive matrix $F$: $$\frac12\int_V{\varepsilon^TF\varepsilon}dV + V=W$$ Representing the strains by $\varepsilon=Bu$, where $B$ is a matrix containing the kinematics derivatives and $u$ is the displacement vector: $$\frac12\int_V{u^TB^TFBu}dV + V=W$$ The exact displacement field $u$ can be approximated aa a set of Ritz trial functions:

$$u=g\{c\}+u_0$$

where $c$ are the Ritz constants and $u_0$ is a vector containing the prescribed displacements. Assuming that $V=K_2\{c\}$, the energy equation becomes:

$$U_0 + \left(\frac12\{c\}^TK_1 + K_2\right)\{c\}=W$$

with:

$$U_0=\frac12\int_V{u_0^TB^TFBu_0}dV$$ $$K_1=\int_V{g^TB^TFBg}dV$$

Applying the principle of total virtual work, which states that when a virtual change in the displacement field takes place there will be a corresponding change in the strain state (internal energy) so that the virtual change in the total work is zero:

$$\delta\left(U_0 + \left(\frac12\{c\}^TK_1 + K_2\right)\{c\} \right) =\delta W$$

$$\delta U_0 + \delta\left(\left(\frac12\{c\}^TK_1 + K_2\right)\{c\}\right) = 0$$

Part of this equation has the same format as this other answer. Using the same steps described there, this equation becomes:

$$\delta U_0 + K_1\{c\} + K_2 = 0$$

The question is: what to do with $U_0$ ? I want to solve $\{c\}$ considering the prescribed boundary conditions included in $U_0$, but from the equation above it seems that they will vanish in the variational statement. $U_0$ is a scalar value.

$\endgroup$
1
$\begingroup$

There was a simple mistake with catastrophic effects in my derivation.

From the question, recall that the exact displacement field $u$ can be approximated aa a set of Ritz trial functions:

$$u=g\{c\}+u_0$$(1)

where $c$ are the Ritz constants and $u_0$ is a vector containing the prescribed displacements. Assuming that $V=K_2\{c\}$, I made a false jump to this:

$$U_0 + \left(\frac12\{c\}^TK_1 + K_2\right)\{c\}=W$$(2)

Let's go back some steps and compare the results. The total potential work before substituting the approximated displacement field of Eq. (1) is:

$$W=\frac12\int_V{u^TB^TFBu}dV + V$$

Using the approximated displacement field of Eq. (1), and assuming that $V=K_2\{c\}$:

$$W=\frac12\int_V{ \left( \{c\}^Tg^T + u_0^T \right) B^TFB \left( \{c\}g + u_0 \right) }dV + K_2\{c\} $$

Here was the pitfall, I did not apply the distributive, which would give:

$$W = \frac12\{c\}^T \left( \int_V{ g^TB^TFBg }dV \right) \{c\} + \frac12 \{c\}^T \left( \int_V{ g^TB^TFBu_0 }dV \right) + \frac12 \left( \int_V{ u_0^TB^TTFBg }dV \right) \{c\} + \frac12 \int_V{ u_0^TB^TFBu_0 }dV + K_2\{c\} $$(3)

defining:

$$K_1 = \int_V{ g^TB^TFBg }dV$$ $$K_0 = \int_V{ u_0BTFBg }dV $$ $$U_0 = \frac12\int_V{ u_0^TB^TFBu_0 }dV $$

Rewritting Eq. (3):

$$W-U_0 = \frac12 \{c\}^T K_1 \{c\} + \frac12\{c\}^TK_0^T + \frac12K_0\{c\} + K_2\{c\}$$

Grouping the terms, recalling that $K_0\{c\}=\{c\}^TK_0^T$:

$$W-U_0 = \left( \frac12 \{c\}^T K_1 + K_0 + K_2 \right) \{c\}$$

Applying the principle of virtual work:

$$\delta( W-U_0 ) = 0 = \delta \left( \left( \frac12 \{c\}^T K_1 + K_0 + K_2 \right) \{c\} \right)$$

Note that $\delta W =0$ and $\delta U_0 = 0$, because the variational statement is build for the Ritz constants $\{c\}$. Following the same approach described in this other answer, the following equation can be obtained:

$$K_1\{c\} + K_0 + K_2 = 0$$

Which can be solved for $\{c\}$. Here the prescribed boundary conditions are included in $K_0$. compare this solution with the solution presented here, without prescribed boundary conditions.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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