symp

Abstract

The study of flows over spherical bluff bodies is relevant in engineering systems such as nuclear power reactors. These flows exhibit complex spatial and temporal behavior due to the presence of significant streamwise curvature that leads to boundary layer separation and reattachment, and the turbulence phenomena associated with these effects. In this study, we present results for flow over one and two spheres in free-stream conditions at Reynolds number 1,000, which corresponds to a turbulent-flow regime. The goal of the study was to verify our simulation methodology and develop an initial understanding regarding the spatial and temporal effects of additional spheres in close proximity to the canonical single-sphere case. Based on our results, we find that additional spheres significantly alter the behavior of the flow. The addition of a second sphere suppresses vortex formation, while the temporal correlation between points in the wake is also reduced.

Conclusion

We have presented selected results from the simulation of incompressible flow over a single and two spheres at Re = 1000. Several verification and validation results for flow over a single sphere were presented to establish confidence in our numerical results. We consider this work to be an initial extension of the canonical case of flow over a single sphere. Of specific interest is the change in temporal dynamics of the flow when additional spheres are added. Our results show that the addition of a second sphere results in a marked increase in the asymmetry of turbulence statistics, even for long periods of time integration. This is consistent with our results from studies of higher Reynolds number flows through domains consisting of regularly ordered spheres. In terms of results, a non-exhaustive selection of first-second order statistics were shown that illustrate the increased level of asymmetry of the flow statistics resulting from the addition of a second sphere downstream from the first. The difference between the temporal vortex shedding behavior of the two cases was discussed by comparing the temporal auto- and crosscorrelations at selected points in the wakes. Finally, a set of results generated by a POD of the turbulent flow field was discussed to illustrate the qualitative difference in the structure of the highest energy coherent structures between the two simulations. A marked decrease in the formation of a vortex street was observed for the two sphere case. Continued work will focus on refining the quantification of the temporal effects of the additional sphere. Potential methods of analysis include dynamic mode decomposition and spectral POD, which are similar to the POD done in this study, but better quantify the temporal dynamics. A linear stability analysis will also be performed to help identify the laminar to turbulent transition mechanism for the two cases and identify any potential differences between them.