Posture Optimization Using CFD

A computational pipeline that integrates Python-based data manipulation scripts, MPI parallel processing architectures, and the SimVascular hemodynamic solver (svsolver). The primary objective is to execute parameter sweeps simulating varying human postures by systematically altering numerical boundary conditions (specifically, vascular resistance values) across segmented 3D vascular models.

---

Simulation Pipeline

graph TD
    A[PostureConfig DataClass] -->|Angle Bounds| B(Calculate Multiplier Matrix)
    B --> C{VesselConfig Resistance}
    C -->|Output .flow & .inp| D[subprocess: mpiexec.exe -n 8]
    D --> E[svsolver-msmpi-bin.exe]
    E --> F[Extract Logs via Regex]
    F -->|Compile ndarray| G(Pandas DataFrame)
    G -->|Serialize| H[(run_summary.pkl)]
    H --> I[Jupyter Notebook Matplotlib]

Loading

1. Configuration and Initialization

• The PostureConfig dataclass defines the geometric constraints. Variables neck, torso, and legs denote rotational angles ( \theta ) in degrees.
• These variables are bounded by physical constraints via a __post_init__ method (e.g., neck angle is clamped between -30° and 45°, torso between -15° and 60°).

2. Boundary Condition Calculation

• The script uses a VesselConfig class containing nested dictionaries AORTA_RESISTANCE, ABDOMINAL_RESISTANCE, and CORONARY_RESISTANCE. These dictionaries map distinct surface IDs (e.g., surface 2: btrunk, surface 3: carotid) to their baseline resistance magnitudes.
• The input posture angles compute a scalar multiplier matrix. This matrix modifies the baseline resistances, simulating the physiological constriction or dilation of the vascular beds caused by gravitational or biomechanical factors.

3. Solver Execution

• The pipeline relies on the subprocess module to interface with the operating system shell.
• It initiates the MPI executable (mpiexec.exe) mapped to the binary svsolver-msmpi-bin.exe. The execution defines the number of parallel threads using the -n 8 argument, distributing the computational matrix operations across logical cores to accelerate the Navier-Stokes equations resolution.

4. Post-Processing and Data Extraction

• Following completion, the script automatically parses the generated log files and output dat files.
• Using Regular Expressions (re module), it identifies and slices specific time steps (e.g., ITERATION_DURATION_MIN = 30) and extracts nodal pressure and velocity vectors.
• Extracted vectors are compiled into multi-dimensional NumPy arrays (np.ndarray) and converted into pandas DataFrames for structured indexing.
• The final aggregated data matrix is serialized via the pickle module into run_summary.pkl to optimize read I/O during subsequent visualization phases.

---

Setup & Run Instructions

Caution

Configured paths pointing to the execution binaries (mpiexec.exe, svsolver) must be strictly set within the global variables of run_summary_extract.py before initiating the Jupyter kernel.

1. System Prerequisites Ensure that Microsoft MPI (or OpenMPI) and the SimVascular suite are installed on your host system.

2. Environment Setup Install the core data science stack required for parsing and visualization:

   pip install numpy pandas matplotlib jupyter

3. Execute the Pipeline Launch the Jupyter Notebook server:

   jupyter notebook

   Open main.ipynb or new.ipynb in your browser. Configure your desired PostureConfig angles in the first cell and execute the notebook sequentially to trigger the mpiexec solver instances.

---

Data Analysis & Jupyter Integration

The repository utilizes Jupyter Notebooks (main.ipynb, new.ipynb) as the frontend controller for the pipeline. These environments allow for iterative configuration of the PostureConfig parameters and use matplotlib.pyplot to generate flow waveform graphs comparing pressure gradients across the defined surface IDs under varying geometric states.