Reduced-order modeling of flow in the circulatory system

Lumped parameter networks (LPN) coupled with 3D CFD simulations provide a mean to simulate the flow in the circulatory system at a reasonable accuracy and affordable cost. In this method, the flow in major vessels is fully resolved, whereas it is modeled in the rest of the circulatory system using lumped components. Despite these attractive properties, the usage of 3D-LPN coupling method is hindered by need for reverse-engineering the LPN components through a tuning process, which itself requires performing many CFD simulations. To reduce the cost of such expensive calculations, a reduced-order model is introduced to replace the 3D model that is orders of magnitude less expensive to solve. The introduced reduced-order model is constructed using a circuit network analogy that can capture flow inertial and viscous losses in the 3D fluid domain. Additionally, we show that the reduced-order model can be constructed directly through the Jacobian matrix in the CFD solver. Reasonable accuracy and computational efficiency are demonstrated by comparing the reduced-order model result with the reference CFD result.