Simulation of the flow profile.

We considered a 30 × 10 × 13.8cm (length × height × width) parallelepiped, in which we solve static, incompressible, laminar Navier-Stokes equations through the “Laminar Flow” environment in COMSOL Multiphysics. We set non-slip wall conditions on the bottom, left, and right boundaries, and an open boundary on the top surface, where no viscous stress is generated. Between the inlet and outlet, we imposed a pressure difference. For all our simulations, we used COMSOL built-in water material properties and a “Normal” size mesh. We first estimated the required pressure difference to generate the maximum experimental fluid speed at the center of the section, assuming a Poiseuille flow within a circular pipe, with a diameter equal to the hydraulic diameter of the rectangular section, yielding 0.038Pa. By imposing this pressure, we obtained a larger speed for all the points but the central one. To identify the correct value of the pressure drop to reconstruct the experimental profile, we carried out a parametric analysis in which we varied the pressure drop between 0.0021Pa and 0.038Pa in 20 steps. We find that the sum of squared errors of fluid speed at the measurement positions is minimized for a pressure difference of 0.0172Pa. From these simulations, we conclude that the average fluid speed within the section is not equal to the maximum velocity within the plane passing through the center of the test section, that is, the plane in which fluid speed was experimentally measured. By computing the average speed on the plane parallel to passing through the section of the test section, we obtain a value of 2.3 cm/s, which is approximately 70% of the maximum speed at the center of the section, 3.2 cm/s. (TIF)