Fish Passage in Large Culverts With Low Flow

CHAPTER 7. CONCLUSIONS AND RECOMMENDATIONS

This report documents a series of physical and numerical modeling efforts designed to support the development of a design procedure for characterizing the variation in velocity within non-embedded and embedded culverts. Physical modeling of symmetrical half-section circular culverts was conducted to provide data with which CFD modeling could be validated. Use of CFD allows more extensive test matrices to be evaluated in less time and budget than physical modeling.

The initial CFD modeling featured two-phase numerical computations that successfully reproduced the physical modeling results. To simplify the CFD modeling, single-phase modeling and truncated single-phase modeling were evaluated with good results. For the embedded culvert runs, a successful strategy for representing natural bed material within the culvert was developed.

Once the CFD modeling of the physical data was completed and validated, the CFD modeling was used to analyze the full culvert cross-sections (3-ft CMP) rather than symmetrical half sections. Subsequent test matrices included CFD runs scaled up to larger culvert sizes. One series of runs maintained Froude number-based scaling (6-ft CMP), and one series tested larger sizes without the scaling constraint (8-ft CMP). These full culvert cross-sections were used to develop and test the proposed design method.

The velocity distribution model that formed the basis of the design methodology is described by Chiu and Chiou and then Chiu.^(30,31) Using the 42 CFD runs for a 3-ft diameter culvert, the five parameters necessary for the velocity model (β_i, δ_y, ε, M, δ_i) were estimated. Based on geometric and hydraulic parameters available to a designer, relations were developed to estimate those parameters for design purposes. The approach was successfully validated on CFD runs for 6-ft and 8-ft diameter culvert models.

The proposed design procedure allows a designer to estimate the velocity throughout a cross-section. These data may be depth-averaged to provide a distribution of velocity and depth across the culvert cross-section that may be used to evaluate fish passage. Although developed for circular culverts, the parameters used in the method are such that the procedure should be applicable to rectangular and other shapes. Two design examples are provided to illustrate the method and the required computations.

Future work is recommended to further validate and improve the recommended design methodology. Further CFD runs should capture the following qualities:

• A wider range of Froude numbers. In this study, Froude numbers (based on average velocity and average depth) ranged from 0.13 to 0.44.
• More variation in bed roughness values, both in an absolute sense and in a relative sense. In this study, only one bed roughness value was tested in detail.
• Shapes other than circular. Rectangular shapes will mimic the results of the embedded circular shapes, so these are not a priority. Testing of elliptical, pipe arch, and other shapes may be useful.
• Water depth/embedment scenarios where the water surface elevation is significantly higher than the elevation at maximum culvert width, resulting in decreasing top widths with increases in water surface elevation. The recommendation from this study is that the top width, T, should not decrease more than 2 percent from the maximum value of top width when applying the methods from this study.