Total Dissolved Gas Generation at Hugh L. Keenleyside Dam, B.C., Canada

Abstract

Total dissolved gases generated downstream of hydropower dams pose an environmental risk to fish species. The objective of this study was to understand total dissolved gas generation mechanisms at Hugh L. Keenleyside Dam along the Columbia River near Castlegar, BC. This was in an effort to produce a predictive model for dam operations. Two types of spill structures were studied, spillways, and low-level outlets. Several predictive models from the literature were modified and tested for the structures at the dam to better understand their wider applicability and to bring insight into this dam’s unique ability to dissolve gas. It was found that the southern set of low-level outlets can generate a significant amount of total dissolved gas relative to the northern set of low-level outlets, primarily as a result of greater turbulence in the stilling basin, and as a result, a higher gas-transfer rate. In 2016, for similar flow rates, the operation of three south low-level outlets generated 124% total dissolved gas supersaturation versus 112% total dissolved gas supersaturation generated by three northern low-level outlets. The spillway stilling basins at Hugh L. Keenleyside Dam are very unique relative to most other basins found in the literature, as a result, many models also found in the literature failed to adequately present acceptable total dissolved gas predictions. This was remedied in part by separately considering the stilling basin region’s and river region’s contribution to total dissolved gas generation. Of the models tested for the spillway dataset, only one achieved a root-mean-square-error within instrument uncertainty, the error was 1.5%. The best results were found using a similar approach for the low-level outlets, where the lowest root-mean-square-error was 1.14%. The dynamic nature of the flow within the stilling basin regions of the dam and difference in flow patterns in the downstream river region underscores the importance of numerical modelling and the challenges simpler analytical models have at predicting total dissolved gas generation.