This work aims to develop a method for quantitatively analyzing red blood cell (RBC) aggregates under controlled flow conditions. Images of experiments are captured and processed in order to quantify aggregation. The experimental setup consists of RBC suspensions in blood plasma entrained by a phosphate-buffered saline solution in a 110 × 60 µm polydimethylsiloxane microchannel. The experiments are performed by varying the hematocrit (5, 10, and 15%) and the flow rate (Q = 5 and 10 µl/hr) in order to observe the effect of shear rate on RBC aggregation. Microchannel dimensions as well as fluid flow rates are determined using numerical simulations. The flow is visualized using a highspeed camera coupled to a micro particle image velocimetry system. Videos obtained with the high-speed camera are processed using a MATLAB program, with each frame analyzed separately. RBC aggregates are detected based on the image intensities and the connectivity between RBCs, using image processing techniques. The average aggregate size and distribution of RBCs for various aggregate sizes are determined for each of the shear rates and hematocrits. These aggregates are shown to be larger at low flow rates where the shear rate is small. Results from tests performed at high hematocrits also show larger RBC aggregates.

The Navier-Stokes-Fourier (NSF) equations are conventionally used to model continuum flow near local thermodynamic equilibrium. In the presence of more rarefied flows, there exists a transitional regime in which the NSF equations no longer hold, and where particle-based methods become too expensive for practical problems. To close this gap, moment closure techniques having the potential of being both valid and computationally tractable for these applications are sought. In this study, a number of five-moment closures for a model one-dimensional kinetic equation are assessed and compared. In particular, four different moment closures are applied to the solution of stationary shocks. The first of these is a Grad-type moment closure, which is known to fail for moderate departures from equilibrium. The second is an interpolative closure based on maximization of thermodynamic entropy which has previously been shown to provide excellent results for 1D gaskinetic theory. Additionally, two quadrature methods of moments (QMOM) are considered. One method is based on the representation of the distribution function in terms of a combination of three Dirac delta functions. The second method, an extended QMOM (EQMOM), extends the quadrature-based approach by assuming a bi-Maxwellian representation of the distribution function. The closing fluxes are analyzed in each case and the region of physical realizability is examined for the closures. Numerical simulations of stationary shock structures as predicted by each moment closure are compared to reference kinetic and the corresponding NSF-like equation solutions. It is shown that the bi-Maxwellian and interpolative maximum-entropy-based moment closures are able to closely reproduce the results of the true maximum-entropy distribution closure for this case very well, whereas the other methods do not. For moderate departures from local thermodynamic equilibrium, the Grad-type and QMOM closures produced unphysical subshocks and were unable to provide converged solutions at high Mach number shocks. Conversely, the bi-Maxwellian and interpolative maximum-entropy-based closures are able to provide smooth solutions with no subshocks that agree extremely well with the kinetic solutions. Moreover, the EQMOM bi-Maxwellian closure would seem to readily allow the extension to fully three-dimensional kinetic descriptions, with the advantage of possessing a closed-form expression for the distribution function, unlike its interpolative counterpart.