Numerical Model of Flow in the Sectional Model of the Spillway for Garrison Dam

 Project Duration: CFD Aspect of Garrison: April 2024 – June 2024 

 Study Area: Hydraulic Engineering, Computational Fluid Dynamics 

 Location: Missouri River, Central North Dakota, USA 

 Principal Investigators: Dr. Chris Thornton and Jeff Ellis 

 Graduate Student: Josh McMartin

 Project Partners: AECOM and USACE 

 Project Contacts: Jeff Ellis (jeff.ellis@colostate.edu) and Josh McMartin (jmcmartin@usbr.gov) 

Project Description

Garrison Dam is located on the Missouri River as the river flows through central North Dakota. The Colorado State University (CSU) Hydraulics Lab has conducted this study of Garrison Dam to investigate the conditions impacting the stilling basin of the prototype. The objective of this report was to design a 3D computational fluid dynamics (CFD) spillway model in FLOW-3D to accurately match to USACE water surface elevation data approaching the stilling basin to aid the design of the physical model being built in the CSU Hydraulics Lab. In the original design, the concrete slabs in the stilling basin failed when they experienced uplift forces due to seepage flow under the slabs. In the new design, concrete overlays have been added to the stilling basin to prevent this issue. Using numerical modeling, part of the study involved testing between different slopes in the entrance slope section to see if this made a difference on hydraulic flow depths, velocity vector magnitudes and velocity vector angles at different points along the model. It was important to check to see if there were large differences which could cause further erosional problems in the stilling basin. This numerical model was created at 1:24 length scale of the prototype and is 1:1 to the physical model created in the Engineering Research Center (ERC).  

Many simulations were created and run for different design flow rates, having baffles or excluding them, changing the approach slope, and also changing the geometry of the inflection point between the main chute and the approach slope. Each run took approximately 12 hours to complete and a total of 16 simulations were made. Data from the FLOW-3D was compared to the 2D model created by USACE for model validation. After the completion of the physical model in the lab, the collected data will be compared to the numerical model results for further validation. 

Key Highlights

  • In the numerical model, the spillway is a representative four-feet-wide section of the center of the prototype spillway.  
  • Dimensions of the model include the head tank (16.6 ft long), the entrance slope section at about 12 ft long, the main chute at about 35 ft long and the exit section at 8 ft long. 
  • The grid mesh cell size was selected to be two inches uniformly across the model.  
  • Total simulation time was 360 seconds, the time step was selected to be 0.02 seconds, and the time step was given a maximum and minimum value that it can vary between so that stability is maintained.  
  • The numerical model uses first-order approximation and uses general FLOW-3D standard solvers for processes like diffusion and advection. 
  • The FLOW-3D results compare favorably to the 2D CFD model created by USACE when considering water surface elevation at the investigation points along the spillway.  
  • Changing the approach slope between 0.0075 and 0.015 did not make a significant difference on the velocity vector magnitudes or angles when the flow nears the end of the main chute approaching the stilling basin.  
  • Having the inflection point curved or pointed did not make a significant difference to these same flow parameters.  
  • Preliminary tests of the physical model found that the data measured compared well to the CSU CFD and USACE CFD data.