Introduction

Abstract

Water flow through the spillway of the Ohio River Olmsted Dam was studied by performing a finite-element simulation of the flow within a narrow section of the volume.
Since the geometry of the spillway is repetitive and identical inflow conditions apply at each subdomain, periodic boundary conditions were assumed in the computations. To test that assumption, the results were compared with a previous simulation that modeled a wider section of the dam.

Wide vs. Narrow

Finite element simulations can be a very useful tool in the process of testing and modifying designs for dams and waterways. The computational expense of such simulations can often be reduced if the geometry of the domain is repetitive or symmetrical. In that case, the assumption can often be made that the flow within one given subdomain will be the same as the flow within other subdomains with identical geometry and boundary conditions. When this assumption is justified, we can obtain useful information about the flow by modeling just one representative segment of the whole domain, thereby reducing the total processing time. Alternatively, we can use the time saved to increase the resolution of the simulation.
However, nonperiodic flow patterns spanning more than one subdomain width can potentially be lost by using smaller subdomains, especially when significant flow is expected across inter-domain boundaries.

Therefore, in order to develop confidence in the results of a "narrow subdomain" simulation, we must compare it with simulations involving wider subdomains. Significant differences would indicate the need to use the latter.