Search papers, labs, and topics across Lattice.
This paper compares POD-Galerkin and POD-Reservoir Computing (RC) for model order reduction (MOR) of cerebrovascular hemodynamics, aiming to accelerate simulations while maintaining accuracy. They trained both methods on CFD snapshots of an idealized basilar artery bifurcation, using a multi-harmonic training signal to enhance efficiency. Results show that both POD-G and POD-RC achieve 100x-1000x speedups compared to full CFD simulations, suggesting their utility as surrogates for predicting flow quantities like wall shear stress.
Forget waiting hours for CFD simulations: this work shows you can get accurate cerebrovascular hemodynamics predictions 100x-1000x faster using reduced-order models based on POD and reservoir computing.
We investigate model order reduction (MOR) strategies for simulating unsteady hemodynamics within cerebrovascular systems, contrasting a physics-based intrusive approach with a data-driven non-intrusive framework. High-fidelity 3D Computational Fluid Dynamics (CFD) snapshots of an idealised basilar artery bifurcation are first compressed into a low-dimensional latent space using Proper Orthogonal Decomposition (POD). We evaluate the performance of a POD-Galerkin (POD-G) model, which projects the Navier-Stokes equations onto the reduced basis, against a POD-Reservoir Computing (POD-RC) model that learns the temporal evolution of coefficients through a recurrent architecture. A multi-harmonic and multi-amplitude training signal is introduced to improve training efficiency. Both methodologies achieve computational speed-ups on the order of 10^2 to 10^3 compared to full-order simulations, demonstrating their potential as efficient and accurate surrogates for predicting flow quantities such as wall shear stress.
