For a hyperelastic problem, I understand the variational formulation can be written as the minimisation of $\Pi$ with

$\Pi = \int_{\Omega} \psi( \pmb{u} )dx - \int_{\partial\Omega} \pmb{T}\cdot \pmb{u} ds$

With $\Omega$ the reference configuration, $\pmb{u}$ the displacement, and $\pmb T$ the traction force on the reference configuration, i.e $\pmb T = \pmb P \cdot \pmb N$ with $\pmb P$ the first Piola Kirchoff Stress and $\pmb N$ the outward normal.

Now, in my problem, the prescribed forces $\pmb t$ are given in the current configuration and not the reference. In order to keep the same weak form like the one used in this FEniCS tutorial, I understand I have to perform a transformation of the $\int_{\partial\Omega} \pmb{T}\cdot \pmb{u} ds$ term. Seeing different references, many refer to the Nanson formula.

The problem is that the formula gives the transformation from the $da \cdot \pmb n$ term, whether in our case, we have the integral defined over $dA \cdot \pmb N$ (the right hand side of the formula).

However, seeing this result in this book : $\pmb P \cdot \pmb N = J |\pmb F^{-T} \pmb N | \hat{\pmb t} $ with $\hat{\pmb t} = \pmb t \circ \phi$ where $\phi$ is the deformation mapping. Is the following derivation correct ?

  • $\hat{\pmb t} = \pmb t$ Since it is prescribed vector load (exterior forces)
  • the term $\int_{\partial\Omega} \pmb{T}\cdot \pmb{u} ds$ becomes $\int_{\partial\Omega} J |\pmb F^{-T} \pmb N | {\pmb t}\cdot \pmb u ds $ which is still defined on the reference configuration.
  • We thus just need to multiply dot(T, u)*ds by $J |\pmb F^{-T} \pmb N |$ in the mentioned tutorial, with -ofcourse- T becoming my new exterior pressure.

Is this reasoning correct ? Otherwise, how can we derive a proper modification to the weak form for FEniCS ?

I didn't find a similar problem/derivation out there, I would love some help. Thank you very much !


1 Answer 1


As far as I understand, you basically want to express your traction vector from deformed (current) configuration to undeformed (reference) configuration in your integral. Would it be possible to just do a pullback operation on your traction vector and directly use it in the integral?

Holzapfel, G. A., 2001, Nonlinear Solid Mechanics, John Wiley, Chichester. (GAH), Pg 82


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.