Introduction and request for feedback

[Marco Nawijn]

Dear All,

Let me first quickly introduce myself. My name is Marco Nawijn, I am 49 years
old and I hold a masters degree in aerospace engineering. I have many years of
practical experience in the application of the finite element method in the
field of mechanics (plasticity, buckling, damage initiation and progression in
composites and so on). I did have a fair amount of theoretical background back
in the day, but this certainly needs some refreshing. I do have a pretty solid
background in Python/C/C++ programming although I don't have a formal education
in these areas. I am currently the manager of the engineering department of
Rondal (https://rondal.com) where we design sailing systems for super yachts
using commercial FSI (SailPack) and FE (Strand7) packages.

Since a long time however, I feel the urge to open the FE black box so to speak
and get more control/insight/understanding of the FE field (theory, limitations,
practical issues, solvers and so on). After googling around every now and than,
I stumbled across Deal II. And although, a lot of the mathematics is still
daunting for me, I really like the quality of the code itself, the available
reference documentation, the lecture series and the volume of examples and their
in-depth explanations. Up to this point in time, I have been lurking around, but
I decided to get my hands and feet dirty and dive in. Before I do so however, I
would like to kindly ask for some feedback on what my plans are. The reason I
ask is because I am afraid that what I want may not exactly align with the Deal
II objectives.

What I basically would like to do is, gradually (months/years) develop the
necessary functionality to analyse and optimize our sailing systems. The list of
topics I am interested in includes (but is not limited to):
 - Linear elastic static analysis of the rig (mast, spreaders, standing rigging)
   using 1D elements (beam, truss);
 - Linear buckling analysis of the rig;
 - Non-linear buckling analysis (account for effect of cable tensioning on
   buckling behaviour);
 - Linear elastic static analysis of the mast using 2D/3D orthotropic elements;
 - Linear buckling analysis of the mast using 2D/3D orthotropic elements;
 - Detailed stress analysis of composites (ply-by-ply modelling of composite,
   visco-elastic modelling of matrix materials);
 - Calculation of the pressure distribution of a sail using a potential flow
   solver.
 - FSI analysis of rig (flow generates pressure on the sail which deforms the
   mast which alters the shape of the sail which alters the pressure
  distribution and so on)

So basically what I would like to do is use as much functionality as possible
from Deal II, implement what is missing and build practical analysis tools out
of it.

Does this make sense?

In terms of a timeline, I will probably start by myself in my own spare time
untill I am able to do the beam analysis (first 3 bullet points). This allows me
to get enough experience in working with Deal II to also guide master/PHD
students. From that point on, some of the additional work will than
probably/hopefully be done in cooperation with master/PHD students.

As a final note, I am a big fan of open source software. Therefore, I am fully
committed to contribute whatever I develop back to the Deal II community
including examples/tests/documentation and so on. For example, one of the first
things on my list is to expand on the Abaqus input file reader. I have some
ideas in this area that I would like to pitch if there is any interest. Another
practical thing I can contribute to is testing the compilation of Deal II with
different versions of libraries. For example, Deal II will not compile out of the
box with newer TBB versions. But I also get a fair amount of warnings of array
bounds that (the compiler thinks) are not correct. So if there is an interest to
test different configurations (accessible to mortals, because I do not have
access to a supercomputing cluster) I am more than happy to get my hands dirty.

Any thoughts/ideas/feedback?

Marco

[Timo Heister]

Hi Macro,

Thanks for getting in touch. I don't have much to say about your plan,
but I can comment on the following:

> As a final note, I am a big fan of open source software. Therefore, I am fully
> committed to contribute whatever I develop back to the Deal II community
> including examples/tests/documentation and so on. For example, one of the first
> things on my list is to expand on the Abaqus input file reader. I have some
> ideas in this area that I would like to pitch if there is any interest.

Great, we are always happy to get contributions! See
https://github.com/dealii/dealii/wiki/Contributing to get started and
feel free to ask for help.

> Another
> practical thing I can contribute to is testing the compilation of Deal II with
> different versions of libraries. For example, Deal II will not compile out of the
> box with newer TBB versions.

There has been progress on supporting the new oneTBB in the last 2 weeks, see
https://github.com/dealii/dealii/pull/13031
https://github.com/dealii/dealii/pull/12958
and things should work already or soon, I think.

> But I also get a fair amount of warnings of array
> bounds that (the compiler thinks) are not correct.

If you see them on deal.II master, feel free to create an issue at
https://github.com/dealii/dealii/issues or try to fix them and submit
a pull request.

Best,
Timo

--
Timo Heister
http://www.math.clemson.edu/~heister/

[Wolfgang Bangerth]

Marco:
Timo has already answered some of your questions. Let me answer a couple more,
while also saying Welcome to this community!

> What I basically would like to do is, gradually (months/years) develop the
> necessary functionality to analyse and optimize our sailing systems. The list of
> topics I am interested in includes (but is not limited to):
>  - Linear elastic static analysis of the rig (mast, spreaders, standing rigging)
>    using 1D elements (beam, truss);
>  - Linear buckling analysis of the rig;
>  - Non-linear buckling analysis (account for effect of cable tensioning on
>    buckling behaviour);
>  - Linear elastic static analysis of the mast using 2D/3D orthotropic elements;
>  - Linear buckling analysis of the mast using 2D/3D orthotropic elements;
>  - Detailed stress analysis of composites (ply-by-ply modelling of composite,
>    visco-elastic modelling of matrix materials);
>  - Calculation of the pressure distribution of a sail using a potential flow
>    solver.
>  - FSI analysis of rig (flow generates pressure on the sail which deforms the
>    mast which alters the shape of the sail which alters the pressure
>   distribution and so on)

All of these have been done in one way or another with deal.II. I will caution
that deal.II does not have beam or truss (or plate) elements as have
traditionally been used in the engineering community. Rather, we have been
using more modern approaches to solving the underlying partial differential
equations that allow for more systematic integration into how we think about
solving partial differential equations today. If you're curious about the
general ideas, take a look at the introduction to the plate example step-47:
https://dealii.org/developer/doxygen/deal.II/step_47.html

For all of the points you have above, I will mention that while that is not
always obvious on the mailing list for obvious reasons, several of us have
extensive experience applying deal.II in industrial contexts. As such, I am
not overly concerned about the general feasibility of what you propose; it is
more a question of having/finding the time and expertise necessary to do so.

> In terms of a timeline, I will probably start by myself in my own spare time
> untill I am able to do the beam analysis (first 3 bullet points). This allows me
> to get enough experience in working with Deal II to also guide master/PHD
> students. From that point on, some of the additional work will than
> probably/hopefully be done in cooperation with master/PHD students.

These sound like interesting student projects. I'm sure I'm not the only one
who would look forward to hearing about the outcome!

> As a final note, I am a big fan of open source software. Therefore, I am fully
> committed to contribute whatever I develop back to the Deal II community
> including examples/tests/documentation and so on.>

That would of course be much appreciated! We are always willing to assist.

> For example, one of the first
> things on my list is to expand on the Abaqus input file reader. I have some
> ideas in this area that I would like to pitch if there is any interest. Another
> practical thing I can contribute to is testing the compilation of Deal II with
> different versions of libraries. For example, Deal II will not compile out of the
> box with newer TBB versions. But I also get a fair amount of warnings of array
> bounds that (the compiler thinks) are not correct. So if there is an interest to
> test different configurations (accessible to mortals, because I do not have
> access to a supercomputing cluster) I am more than happy to get my hands dirty.

Yes, any of these would be useful. As Timo already mentioned, we are working
on supporting newer TBB versions. Invalid array indices is something that
compilers like to warn about (generally in code that is guarded by conditions
that make sure the invalid access is never actually executed), and we
frequently apply patches that silence these warnings every time a new compiler
version comes out. In any case, anything you have to contribute would be very
much appreciated by the community!

Best
Wolfgang


--
------------------------------------------------------------------------
Wolfgang Bangerth email: bang...@colostate.edu
www: http://www.math.colostate.edu/~bangerth/