Three-Dimensional Computational Fluid Dynamics Modeling Study Report

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling Study3Study AreaThe study area includes the E.L. Field Powerhouse forebay, tailrace, and fish lift, thebypass reach in the vicinity of the Pawtucket Dam fish ladder entrance, and within thefish ladder.4Model Development4.1Model DescriptionFLOW-3D is a commercially available solver capable of solving the three-dimensional (3D) Unsteady Reynolds Averaged-Navier Stokes (URANS) equations. The softwareutilizes a Volume of Fluid (VOF) method to calculate the free surface within the domain(Hirt & Nichols, 1981). The package contains the meshing module (pre-processor),solver, and post-processor. FlowSight was used to produce the results presented belowand was provided with the FLOW-3D software package.4.1.1Modeling ApproachFLOW-3D is software developed and supported by Flow Science, Inc. The governingequations used in Flow-3D can be found in the user’s guide (Flow Science, Inc, 2019).The software solves fully URANS equations on structured grids. A model fitted mesh wasdeveloped for the proposed auxiliary spillway modification. A known water surfaceelevation (WSE) was applied to the upstream boundary of the CFD model. Applying theknown WSE reduces the complexity of the URANS equations.4.1.1.1Pressure Solver OptionsTwo numerical schemes are available for the pressure solver with multiple options.Explicit and implicit solvers are available. Within the Implicit solver, multiple options areavailable. Limited compressibility models can be toggled to relax the constraints of thepressure solver for cases where solution stability is an issue. The implicit pressure solverwas applied to the model for the results presented below.4.1.1.2Turbulence ModelsVarious one (Prandtl Mixing Length and Turbulent Energy Model) and two equation (k-ε,k-ω, and RNG) turbulence models are available in FLOW-3D. A Large Eddy Simulation(LES) model is also available for selection depending on the type of flow expected anddesired flow feature resolution. The Renormalized Group (RNG) model was selected andis an applicable closure for the CFD study based on anticipated flow patterns (Orszag &Yakhot, 1986).4 February 25, 2021

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling Study4.1.1.3Model LimitationsThe CFD model is limited in the data it can accurately produce. Some hydrodynamicfeatures are not accurately modeled with the selected solver and turbulence closure.Recirculation patterns and vortices are approximate in size and strength.4.2Model GeometryHDR developed a topographic digital elevation model (DEM) using GeographicalInformation Systems (GIS). All geo-processing of the following data sources wasachieved using the ArcMap (Esri) application’s 3D Analyst geo-processing tools.A single DEM was created from the following data sources: Ortho imagery Light Detection and Ranging (LiDAR) based elevation data collectedby the Normandeau Associates provided in coordinate system UTM Zone 19N, NAD83 2011, units meters, North American Vertical Datum of 1988 (NAVD 88), unitsmeters. Bathymetry depth points collected by Normandeau Associates. These areas weresite specific to three areas 1) just upstream of the main dam fish ladder, 2) powercanal, and 3) powerhouse tail race. Data provided in coordinate system UTM Zone19N, NAD 83 2011, units meters, vertical datum NAVD 88, units feet. LiDAR acquired from United States Geological Survey (USGS) LAS format 2011.Data provided in coordinate system Universal Transverse Mercator (UTM) Zone 19N,NAD 83, units meters, vertical datum NAVD 88, units meters.Most of the coverage area required for CFD modeling was comprised of data received byNormandeau Associates as a randomly spaced point cloud xyz file. The xyz file wasderived from remotely sensed survey in NAVD 88 vertical datum. These data weretransformed as a discrete point feature class using "ASCII 3D to Feature Class" geoprocessing tool. Both the horizontal coordinate system (UTM 83 meters) and verticalunits (NAVD 88 meters) were maintained.The vertical units for the bathymetric survey data points were converted from feet tometers in order to match the vertical units of the larger areas of the overall CFD modelarea.In several small overbank areas upstream of the fish ladder and east bank of the powercanal and tailrace, additional terrain data was needed to provide additional detail. USGSLiDAR randomly spaced point cloud data was utilized. These USGS LiDAR derived fileswere converted from LAS to points using geo-processing tools "LAS Dataset to TIN" then"TIN Node".In the areas of the fish ladder and fish elevator, polygon feature class were created todepict an area of flat elevation slightly higher in elevation than the bottom of StandardFebruary 25, 2021 5

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling StudyTessellation Language (STL) in order to exclude water from seeping between the finalterrain DEM and the STL.All components as described above were combined into a triangulated irregular network(TIN) using the "Create TIN" geo-processing tool. This resulting TIN was then convertedto a final DEM raster using "TIN to Raster" (1 X 1-foot postings raster resolution, "ProjectRaster" (convert from UTM to State Plane coordinates), and "Raster Calculator" (convertvertical units from meters to feet". A final step to convert the .tiff raster to ASCII file"Raster to ASCII" for final import into the CFD model.The fish ladder and E.L. Field Powerhouse structures were created in AutoCAD andexported to an STL. The STLs and ground surface were rotated to align to an arbitraryvertical position. The rotation allows the structure location to align with the orthogonalmesh elements required for the solver. The rotated model reduces the number of meshelements necessary to define the features within the model domain.5Mesh DevelopmentThe pre-processor for FLOW-3D works with orthogonal elements. The model topographyand features were rotated to capture significant features with fewer elements. A balancebetween mesh density and computational time was desired. Several iterations of meshdensity were performed, providing the basis for a mesh sensitivity analysis. A multi-blockapproach was utilized to build the model domain. The use of a multi-block domain allowsmultiple mesh sizes to be used throughout the domain. Each mesh block conformed tobest practice guidelines provided by FlowScience (Flow Science, Inc, 2019).5.1Fish Ladder5.1.1Mesh Sensitivity StudyMultiple meshes of the fish ladder were analyzed to determine if the solution was gridindependent. The first grid tested used 0.8-ft spacing in all three axial directions (coarse).After completion of the first simulation, the grid was refined to use a 0.4-ft spacing in allthree axial directions (standard). The 0.4-ft mesh was further refined in a third test to limitthe mesh spacing to 0.2-ft (refined). Figure 5-1 shows the meshed blocks with 0.4-ftspacing.6 February 25, 2021

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling StudyFigure 5-1: Fish Ladder Mesh Blocks (0.4-ft grid spacing)5.1.2Model Development ScenarioDuring the mesh development stage of the CFD study, multiple runs using identicalboundary conditions were completed to determine the sensitivity to mesh parameters.Virtual probes measuring depth and velocity were placed throughout the fish ladder tomeasure the effects of the mesh refinement on model hydraulics. Table 5-1 lists theprobes and corresponding hydraulic measurements. Figure 5-2 shows the virtual probelocations.Table 5-1: Virtual Probe 03.16vs .155.562.641.800.759.839.339.775.11vs cityfps3.405.005.317.863.061.561.999.259.567.59vs std-1.600.314.79-1.51-7.260.314.49February 25, 2021 7

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling 760.710.78vs 073.171.902.075.043.293.4822.140.2215.08vs 2.38vs std0.871.830.200.480.12-21.53-0.05Figure 5-2: Mesh Sensitivity Probe LocationsWater surface elevations were measured along the centerline of the fish ladder andcompared between the three mesh refinement levels. This comparison is shown inFigure 5-3.The standard 0.4’ grid spacing was selected for the final model. Water surface elevationsdifferences were negligible, and depth and velocity differences were minor between therefined and standard grid spacing.8 February 25, 2021

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling StudyFigure 5-3: Water Surface Elevation ComparisonFebruary 25, 2021 9

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling Study5.1.3Adding TerrainFollowing the mesh sensitivity study, the Merrimack River bathymetry and PawtucketDam were added upstream and downstream of the fish ladder. The combinedbathymetry, dam, and fish ladder geometry is shown in Figure 5-4. It should be notedthat the best available LiDAR data was used to capture the downstream bathymetry.Water levels in the Merrimack River at the time of LiDAR collection were such that onlythe five most up-stream weirs were captured in the terrain data.The model extends approximately 300’ upstream of the fish ladder exit and roughly 1000’downstream to the Mammoth Road Bridge.As the fish ladder was the area of interest, the combined model mesh was refined to 0.4ft grid spacing in all three axial directions in the fish ladder vicinity. To facilitatereasonable computation run times, the mesh refinement was decreased from 0.4-ft, to0.8-ft, and finally 1.6-ft grid spacing moving away from the ladder. The three refinementlevels are shown in Figure 5-5 as red (0.4-ft), green (0.8-ft) and blue (1.6-ft grid spacing)areas.10 February 25, 2021

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling StudyFigure 5-4: Combined Fish Ladder, Dam, and Bathymetry GeometryFebruary 25, 2021 11

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling StudyFigure 5-5: Combined Fish Ladder, Dam, and Bathymetry Mesh Refinement Zones12 February 25, 2021

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling Study5.2E.L. Field Powerhouse ModelA model was developed to study the hydraulics of the flow entering the powerhouseforebay. The full model geometry is shown in Figure 5-6.Figure 5-6: Forebay Model Geometry5.3Tailrace ModelA tailrace model was developed to study the hydraulics of the flow leaving thepowerhouse. The full model geometry is shown in Figure 5-7. A detailed view of the E.L.Field exits and flow directions with approximate powerhouse inflow boundaries is shownin Figure 5-8.February 25, 2021 13

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling StudyFigure 5-7:Tailrace Model Geometry14 February 25, 2021

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling StudyFigure 5-8:Tailrace Model Geometry, E.L. Field Exits and Powerhouse Inflow BoundariesFebruary 25, 2021 15

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling Study5.3.1Mesh Sensitivity StudyMultiple meshes were analyzed to determine if the solution was grid independent. Thefirst grid tested used 1.6-ft spacing in all three axial directions. After completion of thefirst simulation, the grid was refined to use a 0.8-ft spacing in all three axial directions. Toensure reasonable computation run times, the model mesh was then refined in the first100’ downstream of the powerhouse to 0.4-ft spacing in all three axial directions. Theremainder of the domain was left at 0.8-ft spacing. Figure 5-9 shows the mesh for the0.8-ft spacing in the vicinity of the powerhouse.Figure 5-9: Typical mesh layout for tailrace model5.4Model Approach and ScenariosBoundary conditions for the CFD model were applied through multiple boundary types.Boundary types for the CFD model are listed below:5.4.1Volume flow inletThe volume flow inlet allows a specified volume of flow to enter the model and was usedat the upstream boundary of the model. A directional vector was applied to the inflow.The water surface was known and was specified for the inflow boundary condition.16 February 25, 2021

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling Study5.4.2Pressure OutletThe pressure outlet specifies a known pressure and/or water surface elevation for thedownstream boundary condition. The pressure was specified as atmospheric for thecases below.5.4.3WallThe boundary type wall applied the no-slip condition at the outer boundary of themesh blocks as well as a zero-velocity condition normal to the boundary.5.5Model EvaluationCompleted model runs were evaluated quantitatively and qualitatively using velocitymagnitude and flow streamlines. Results are presented below in Section 6. The CFDmodel solves the URANS equations and data presented at a single time step maycontain a maximum or minimum value, which may or may not correspond to theanticipated hydraulic characteristics.FLUX SURFACESFlux surfaces were used to monitor the volumetric flow through the spillway and thereservoir.MONITORING POINTSMonitoring points were placed within the model to gather point data for the reservoir andarea upstream of the spillway. The monitoring points were selected based on theirproximity to key model elements.6Study ResultsAs noted above in Section 1, Boott notes that the results presented in this report arepreliminary. The calibrated and validated models will run simulations under various inputoperational scenarios. Boott has refined the suite of potential simulation runs based oninput provided by National Marine Fisheries Service in response to the PSP. Boottanticipates providing additional results of the models by March 30, 2021 after modelshave been further refined. Boott also anticipates conducting a working group meeting todiscuss refinement of the models and scenarios to be simulated.Example preliminary CFD results of the fish ladder, E.L Field Powerhouse, and tailracemodel are presented in Appendix A, Appendix B, and Appendix C, respectively.February 25, 2021 17

Lowell Hydroelectric ProjectThree-Dimensional Computational Fluid Dynamics Modeling Study7Variances from FERC-Approved Study PlanAs previously noted, due to diverse locations and accessibility of the areas surveyed inthe canal, forebay, tailrace, bypass reach and within the fish lift and fish ladderstructures, bathymetric and flow data collection surveys were needed, and separate CFDmodels need to be constructed. Much of the coverage area required for CFD modelingwas comprised of data received from Normandeau Associates while conducting the nowcomplete telemetry studies (Downstream American Eel Passage Study, Juvenile AlosineDownstream Passage Assessment Study, and Upstream and Downstream Adult AlosinePassage Assessment). Consequently, Boott is still working on development andrefinement of the CFD models.Boott anticipates providing additional results of CFD model development by March 30,2021 after CFD models for the fish ladder, E.L. Field Powerhouse, and the tailrace havebeen further refined. Boott anticipates conducting a working group mee

Three-Dimensional Computational Fluid Dynamics Modeling Study February 25, 2021 v List of Acronyms 3-D three dimensional Boott Boott Hydropower, LLC (or Licensee) CFD Computational Fluid Dynamics C.F.R. Code of Federal Regulations DEM digital elevation model FERC Federal Energy Regulatory Commission (or Commission)