FEniTop

FEniTop (pronounced as /ˈfi nɪ tɒp/) is short for FEniCSx-based topology optimization.

It is an open-source topology optimization software inheriting several advantages of FEniCSx including

• seamless transitions to varying spatial dimensions (2D and 3D), mesh geometries (structured and unstructured), element geometries (e.g., triangles, quadrilaterals), element types (e.g., Lagrange, Crouzeix–Raviart), element orders, and quadrature degrees;
• the expression of PDEs in weak form, bypassing the need for tedious matrix manipulation;
• the automatic differentiation that can mitigate the intricacies of chain rules in FEA and sensitivity analysis;
• the access to readily available linear and nonlinear solvers and preconditioners through the PETSc backend via petsc4py;
• scalable parallel computing support compatible with various platforms from laptops to distributed computing clusters.

Additionally, it features

• a Helmholtz-type PDE filter tailored for large-scale problems in parallel computing, particularly emphasizing its efficacy for large filter radii;
• parallel implementations for both the Optimality Criteria (OC) optimizer and the Method of Moving Asymptotes (MMA) to update the design variables;
• exporting optimized designs into XDMF files, enabling seamless importation into open-source ParaView software for post-processing.

Examples with performance tests

2D cantilever beam with a structured mesh

2D disk with an unstructured mesh

3D cantilever beam with a structured mesh

3D shell with an unstructured mesh

2D compliant mechanism design

How to access FEniTop

Installation of the coding environment

FEniTop relys on the coding evironment of FEniCSx and a few other dependencies. To install the coding environment, we can follow the procedure below.

Short version

• Install FEniCSx of version 0.7.3.
• Install PyVista.
• Install Xvfb with apt-get -qq update && apt-get -y install libgl1-mesa-dev xvfb.

Detailed version

To install the coding environment in one shot, we recommend to use Docker containers. Take Windows 11 as an example, we can follow the steps below.

• Run wsl --install in Windows PowerShell, which automatically enables the features necessary to run Windows Subsystem for Linux (WSL) and installs the Ubuntu distribution of Linux.
• Install Docker Desktop and Visual Studio Code.
• Start Docker Desktop, select settings-Resources-WSL integration, and activate Ubuntu we just installed if not activated.
• Start Ubuntu and run docker pull jiayingqi/dolfinx-fenitop, which pulls an image containing all required packages (FEniCSx+PyVista+Xvfb).
• Use the pulled image to create a Docker container by running docker run -it --name fenitop_container --shm-size=1024m -v /mnt/d/windows_folder_name:/shared jiayingqi/dolfinx-fenitop in Ubuntu. Change d to the actual disk and change windows_folder_name to the acutual folder.
• Start Visual Studio Code and select Extensions (Ctrl+Shift+X). Search for and install Docker and Remote Development extensions.
• Select Docker extension we just installed, and we should see a container named as fenitop_container. Right click on this container and select Start and then Attach Visual Studio Code.
• A new window will pop up. In this window, select File-Open Folder (Ctrl+K and then Ctrl+O) and open the shared folder.
• Now we have successfully created the coding environment inside this folder. A good thing is whatever changes you make in this shared folder will be synchronized to windows_folder_name and vice versa.

Installation of FEniTop

To install FEniTop, simply execute the command in your terminal (such as the terminal of Visual Studio Code).

git clone https://github.com/missionlab/fenitop

How to run FEniTop scripts

To run FEniTop scripts, simply execute the following commands in your terminal.

• 2D cantilever beam with a structured mesh: mpirun -n 8 python3 scripts/beam_2d.py
• 2D disk with an unstructured mesh: mpirun -n 8 python3 scripts/disk_2d.py
• 3D cantilever beam with a structured mesh: mpirun -n 8 python3 scripts/beam_3d.py
• 3D shell with an unstructured mesh: mpirun -n 8 python3 scripts/shell_3d.py
• 2D compliant mechanism design: mpirun -n 8 python3 scripts/mechanism_2d.py

Authors, sponsors, and citation

Authors

• Yingqi Jia (yingqij2@illinois.edu)
• Chao Wang (chaow4@illinois.edu)
• Xiaojia Shelly Zhang (zhangxs@illinois.edu)

Sponsors

• U.S. National Science Foundation (NSF) EAGER Award CMMI-2127134
• U.S. Defense Advanced Research Projects Agency (DARPA) Young Faculty Award (N660012314013)
• NSF CAREER Award CMMI-2047692
• NSF Award CMMI-2245251

Citation

• Jia, Y., Wang, C. & Zhang, X.S. FEniTop: a simple FEniCSx implementation for 2D and 3D topology optimization supporting parallel computing. Struct Multidisc Optim 67, 140 (2024). https://doi.org/10.1007/s00158-024-03818-7

@article{jia_fenitop_2024,
  title = {{FEniTop}: A Simple {FEniCSx} Implementation for {2D} and {3D} Topology Optimization Supporting Parallel Computing},
  shorttitle = {FEniTop},
  author = {Jia, Yingqi and Wang, Chao and Zhang, Xiaojia Shelly},
  year = {2024},
  month = aug,
  journal = {Structural and Multidisciplinary Optimization},
  volume = {67},
  number = {140},
  issn = {1615-1488},
  doi = {10.1007/s00158-024-03818-7},
}