## Empirical velocity predictions at culvert inlets

##### Abstract

The velocity distribution at the entrance cross section of a culvert is typically diverse, reflecting the nuances of the bed material, debris and other hydraulic factors just upstream of the culvert. These diverse inlet velocity fields have been observed to perpetuate some distance into the culvert, impacting the ability of fish to travel upstream in the culvert barrel. It is important to be able to quantitatively describe the inlet velocity field, especially as this serves as a necessary boundary condition for three-dimensional modeling of fluid flow in culverts. While there are various theory-based models of velocity distributions in open channels, velocity distributions at culvert inlets tend to be chaotic and are not well represented by analytic methods. The goal of this project was to use field data collected at existing culverts to estimate the density at which velocity observations should be collected to adequately describe the nature of the velocity at the culvert inlet. Two methods of data analysis were utilized to determine the required density of velocity observations. The first approach randomly selected velocities to be used as predictors and did not stress the location of the predictors, but instead emphasized the number of velocity observations needed to describe the nature of the velocity at the culvert inlet. The second method employed the idea that the location of the predictors was more important than quantity of predictors used. Results indicate that the pattern of velocity measurements is important - that is, velocities should not be measured at randomly selected positions in the cross section, but should follow a geometric pattern where the measurement density increases in zones having larger velocities. Also, it appears that if one follows the rigorous implementation of the USGS method for measuring stream flow (often referred to as the velocity-area method in texts), velocity predictions can be extrapolated using the inverse-distance-squared technique to adequately describe the inlet velocity field. The implication of this research is that there are steps that can be followed to adequately describe the nature of the velocity at culvert inlet even through the velocity distributions are chaotic in these regions.