Three parallel algorithms for classical molecular dynamics are presented. The first assigns each processor a fixed subset of atoms; the second assigns each a fixed subset of inter-atomic forces to compute; the third assigns each a fixed spatial region. The algorithms are suitable for molecular dynamics models which can be difficult to parallelize efficiently—those with short-range forces where the neighbors of each atom change rapidly. They can be implemented on any distributed-memory parallel machine which allows for message-passing of data between independently executing processors. The algorithms are tested on a standard Lennard-Jones benchmark problem for system sizes ranging from 500 to 100,000,000 atoms on several parallel supercomputers--the nCUBE 2, Intel iPSC/860 and Paragon, and Cray T3D. Comparing the results to the fastest reported vectorized Cray Y-MP and C90 algorithm shows that the current generation of parallel machines is competitive with conventional vector supercomputers even for small problems. For large problems, the spatial algorithm achieves parallel efficiencies of 90% and a 1840-node Intel Paragon performs up to 165 faster than a single Cray C9O processor. Trade-offs between the three algorithms and guidelines for adapting them to more complex molecular dynamics simulations are also discussed.

Fast parallel algorithms for short-range molecular dynamics

To improve the performance of particle image velocimetry in measuring instantaneous velocity fields, direct cross-correlation of image fields can be used in place of auto-correlation methods of interrogation of double- or multiple-exposure recordings. With improved speed of photographic recording and increased resolution of video array detectors, cross-correlation methods of interrogation of successive single-exposure frames can be used to measure the separation of pairs of particle images between successive frames. By knowing the extent of image shifting used in a multiple-exposure and by a priori knowledge of the mean flow-field, the cross-correlation of different sized interrogation spots with known separation can be optimized in terms of spatial resolution, detection rate, accuracy and reliability.

Theory of cross-correlation analysis of PIV images : Image analysis as measuring technique in flows

Fluid dynamic turbulence is one of the most challenging computational physics problems because of the extremely wide range of time and space scales involved, the strong nonlinearity of the governing equations, and the many practical and important applications. While most linear fluid instabilities are well understood, the nonlinear interactions among them makes even the relatively simple limit of homogeneous isotropic turbulence difficult to treat physically, mathematically, and computationally. Turbulence is modeled computationally by a two-stage bootstrap process. The first stage, direct numerical simulation, attempts to resolve the relevant physical time and space scales but its application is limited to diffusive flows with a relatively small Reynolds number (Re). Using direct numerical simulation to provide a database, in turn, allows calibration of phenomenological turbulence models for engineering applications. Large eddy simulation incorporates a form of turbulence modeling applicable when the large-scale flows of interest are intrinsically time dependent, thus throwing common statistical models into question. A promising approach to large eddy simulation involves the use of high-resolution monotone computational fluid dynamics algorithms such as flux-corrected transport or the piecewise parabolic method which have intrinsic subgrid turbulence models coupled naturally to the resolved scales in the computed flow. The physical considerations underlying and evidence supporting this monotone integrated large eddy simulation approach are discussed.

New insights into large eddy simulation

ReviewFunction of alternative splicing

Hypoplastic Left Heart Syndrome : Current Considerations and Expectations

FLOW CHARACTERISTICS AND WALL SHEAR AND THEIR CORRELATION WITH CLOT FORMATION IN THE 50cc PENN STATE ARTIFICIAL HEART

Computational fluid dynamics (CFD) is widely used to predict mechanical hemolysis in medical devices. The most popular hemolysis model is the stress-based power law model that is based on an empirical correlation between hemoglobin release from red blood cells (RBCs) and the magnitude of flow-induced stress and exposure time. Empirical coefficients are traditionally calibrated using data from experiments in simplified Couette-type blood-shearing devices with uniform-shear laminar flow and well-defined exposure times. Use of such idealized coefficients in simulations of real medical devices with complex hemodynamics is thought to be a primary reason for the historical inaccuracy of absolute hemolysis predictions using the power law model. Craven et al. (Biomech Model Mechanobiol 18:1005-1030, 2019) recently developed a CFD-based Kriging surrogate modeling approach for calibrating empirical coefficients in real devices that could potentially be used to more accurately predict absolute hemolysis. In this study, we use the FDA benchmark nozzle to investigate whether utilizing such calibrated coefficients improves the predictive accuracy of the standard Eulerian power law model. We first demonstrate the credibility of our CFD flow simulations by comparing with particle image velocimetry measurements. We then perform hemolysis simulations and compare the results with in vitro experiments. Importantly, the simulations use coefficients calibrated for the flow of a suspension of bovine RBCs through a small capillary tube, which is relatively comparable to the flow of bovine blood through the FDA nozzle. The results show that the CFD predictions of relative hemolysis in the FDA nozzle are reasonably accurate. The absolute predictions are, however, highly inaccurate with modified index of hemolysis values from CFD in error by roughly three orders of magnitude compared with the experiments, despite using calibrated model coefficients from a relatively similar geometry. We rigorously examine the reasons for the inaccuracy that include differences in the flow conditions in the hemolytic regions of each device and the lack of universality of the hemolysis power law model that is entirely empirical. Thus, while the capability to predict relative hemolysis is valuable for product development, further improvements are needed before the power law model can be relied upon to accurately predict the absolute hemolytic potential of a medical device. 

Examining the universality of the hemolysis power law model from simulations of the FDA nozzle using calibrated model coefficients

Develpoment of a pulsatile ventricular assist device for infants and children

While medical options for children born with congenital heart defects include transplantation, the amount of available organs remains limited. This lack of donors led to the development of the National Institute of Health’s National Heart, Lung and Blood Institute Pediatric Circulatory Support Program. Contracts have been awarded to five teams with the task of creating novel support systems for children, ranging from 2 to 25 kg [1]. As part of this program, Penn State has developed a 12 cc pulsatile pediatric ventricular assist device (PVAD), based on the successful 70 cc Pierce-Donachy adult assist device. During the process of reducing the volume of the device for pediatric use, changes were made to the design including altering the angles of the inlet and outlet ports. Previous two-dimensional flow visualization in the PVAD by Manning et al. had shown that these changes led to an increased three-dimensionality of the flow, the influence of which required further investigation [2]. It is important to characterize the fluid dynamics of the flow field inside assist devices such as the PVAD because certain characteristics including high blood residence time, stagnant flow and wall shear rates below 500 s−1 can lead to an increased propensity of thrombus deposition [3,4].© 2008 ASME

A comprehensive flow study of the 12 cc Penn State pulsatile pediatric ventricular assist device

Thrombosis and thromboembolism are deadly risk factors in blood‐contacting biomedical devices, and in‐silico models of thrombosis are attractive tools to understand the mechanics of these processes, though the simulation of thromboembolism remains underdeveloped. The purpose of this study is to modify an existing computational thrombosis model to allow for thromboembolism and to investigate the behavior of the modified model at a range of flow rates. The new and existing models are observed to lead to similar predictions of thrombosis in a canonical backward‐facing step geometry across flow rates, and neither model predicts thrombosis in a turbulent flow. Simulations are performed by increasing flow rates in the case of a clot formed at lower flow to induce embolization. While embolization is observed, most of the clot breakdown is by shear rather than by breakup and subsequent transport of clotted material, and further work is required in the formulation and validation of embolization. This model provides a framework to further investigate thromboembolization.

Keefe B. Manning

Papers

FLOW CHARACTERISTICS AND WALL SHEAR AND THEIR CORRELATION WITH CLOT FORMATION IN THE 50cc PENN STATE ARTIFICIAL HEART

Examining the universality of the hemolysis power law model from simulations of the FDA nozzle using calibrated model coefficients

Develpoment of a pulsatile ventricular assist device for infants and children

A comprehensive flow study of the 12 cc Penn State pulsatile pediatric ventricular assist device

Toward modeling thrombosis and thromboembolism in laminar and turbulent flow regimes