Heat Flux Engineering Analysis Tool

[Kunda1]

FreeCAD is used in this project

The Heat flux Engineering Analysis Toolkit (HEAT) is a suite of tools for predicting the heat flux incident upon PFCs in tokamaks. The toolkit connects CAD, FEM, MHD, Plasma Physics, Visualization, HPC, and more, in one streamlined package. The objective is to enable engineers and physicists to quickly ascertain heat loads given specific magnetic configurations and geometric configurations.

If a user wants to use HEAT without building it on his/her local machine, then they can directly access HEAT from a web browser inside the Princeton Plasma Physics Lab (PPPL) VPN. For instructions on where to point your web browser email Tom Looby (email below).

If a user wants to set up HEAT on a local machine there are a variety of configuration steps that need to be completed (ie compiling MAFOT, building OF modules from source, installing FLASK, network configuration and proxy mapping, ParaViewWeb install, etc.). We are currently working on an appImage (or VM) that will enable users to install locally without any setup (fall 2020). For more info, again, contact Tom.

The author engineer is Tom Looby, a PhD candidate on assignment at NSTX-U for Oak Ridge National Lab. This project is an Oak Ridge National Lab tool built by the Fusion Energy Division, but it is also openSource under the MIT license.

We are currently looking for collaborators, so if you feel like you could use HEAT, reach out!

Source: https://github.com/plasmapotential/HEAT

[thschrader]

Interesting!
At a first glance I would say it works like the cfdOF-wb, translating the CAD-model
to openFoam usable text-files. But HEAT seems very special, plasmaphysics "only".
Snapshot from presentation-slides:

HEAT_presentation_slide.JPG (135.03 KiB) Viewed 2233 times

[HoWil]

Maybe this should be moved to FEM subforum.

[plasmapotential]

I am the author of HEAT, and I thought I would post some information on the code. This might be long but I figure an explanation that is too long is better than one that is too short. Maybe someone else trying to create custom applications will find this useful. I wanted to start by saying that I am extremely grateful for all the hard work of the developers of FreeCAD. Without the power of this tool I would have been unable to create HEAT. So thank you all. I am especially grateful for the python interface to freecad. This is what HEAT uses to do complicated tasks.

For examples and movies of HEAT in action, visit the github page: https://github.com/plasmapotential/HEAT. The FreeCAD function calls all live in the CADClass.py module. There will also be a publication coming out in the journal Fusion Science and Technology giving an overview of HEAT in the next few months. I will try to remember to update this post when that happens.

The goal of HEAT is to provide an integrated modelling suite to engineers and physicists for analysis of tokamak plasma facing components (PFCs). PFCs are the tiles that line the wall of the tokamak, and take a beating from the plasma. These PFC tiles will see incredible heat loads . Power producing tokamaks are expected to have heat fluxes in the 20-50 MW/m^2 range. For comparison, the average heat load seen by a rocket nose cone is ~1MW/m^2. This creates a big challenge for PFC engineers.

The only way get truly accurate heat flux profiles onto the PFC tiles in the machine is to directly couple the physicist's plasma physics to the engineer's CAD. This is the intent of HEAT; to provide an interface between tokamak physics (MHD), CAD, and FVM.

This slide gives an overview of the connection between these different engineering/physics domains in HEAT:

archSlide.png (266.85 KiB) Viewed 1923 times

HEAT uses FreeCAD to handle the engineer's CAD. I use the FreeCAD STP file reader, and then I do some basic filtering to remove nuts, bolts, and other stuff that I dont want to waste calculation time on. Once I have a list of PFC parts (no bolts/nuts/others), I mesh each PFC part. I use the mefisto mesher because that enables me to choose the "resolution" of each triangle. I choose the maximum edge length to correspond to the resolution of HEAT fluxes that I want to calculate. I calculate the mesh center for each triangle, and then in a series of steps I calculate the heat load at each mesh center using the magneto-hydro-dynamics (MHD) provided by the physicist and some empirical scalings (ie plasma physics). The last step in the heat flux calculation is to find triangles that are magnetically shadowed (remember that in a plasma heat flows along magnetic field lines, so different electromagnet configurations produce different shadows), and set the heat flux on them to 0. For those who care, this is done by integrating up the magnetic field line and checking for intersections with a list of possible intersection triangles (also generated using freecad) using a code called MAFOT.

This slide shows the an example freecad mesh and corresponding mesh centers for a PFC:

meshCtrs.png (751.57 KiB) Viewed 1923 times

And this slide shows an example heat flux calculated on these mesh centers (with magnetic shadows included):

heatfluxCtrs.png (390.82 KiB) Viewed 1923 times

Once I have a heat flux calculated across the tile surface, I then use openFOAM to solve for the temperature (and stress if needed) through the tile. The PFCs are limited by temperature and stress limits of the material (usually graphite or tungsten), so this step will tell me if the PFC violates its thermal limit. To do this, I created a custom openFOAM solver called heatFOAM, which uses the heat flux that I just calculated as a boundary condition to the temperature dependent thermal diffusion equation.

In openFOAM I take the mesh (STL) generated by freecad and then create a volume mesh using blockMesh and snappyHexMesh. Next I map the heat flux that I just calculated to the volume mesh surface (boundary) using timevaryingMappedFixedValue to interpolate and a groovyBC that calculates the gradient on the boundary using Fourier's Law. Once the boundary condition is set, I solve the heat diffusion equation throughout the tile using temperature dependent thermal properties (I interpolate at each iteration step). The final product is the temperature distribution through the tile. A similar process is performed for stress.

Here is an example of the magnetic equilibrium, heat flux, and temperature (left to right), that are generated using this process overlaid onto real engineering CAD from an engineer:

example1.png (698.34 KiB) Viewed 1923 times

Anyways I hope this gives you an idea of how HEAT works. Feel free to ask any questions.

[HoWil]

Very nice work and presentation.... and a bit of woooow

[Kunda1]

Very nice work and presentation.... and a bit of woooow

Yea, it's a bit dumbfounding.