A Comparative Study of HEC-RAS 2D, TUFLOW, & Mike 21 Model Benchmark Testing

Transcription

1 A Comparative Study of HEC-RAS 2D, TUFLOW, & Mike 21 Model Benchmark Testing June 2016 Presented by: Murari Paudel, PhD, PE, CFM Soledad B Roman, EIT John Prichard, PE, CFM Wood Rodgers Inc. Sacramento, CA

2 Acknowledgements Co-authors Soledad B Roman and John Prichard Hydrologic Engineering Center (HEC), Davis, CA BMT WBM, Australia DHI USA Water Resources Engineering Dept., Wood Rodgers Inc.

3 Presentation Overview Study Purpose Methodology and Limitations Test Cases Assessment of Results Conclusions

4 Study Purpose Benchmark TUFLOW (2016), MIKE21 (2014) and HECRAS (2016) Test the models for a range of basic hydraulic conditions and model performance measures: 1. Hydraulic losses: Does the model scheme account for losses accurately? 2. Grid convergence: Does the model produce similar result for different cell sizes? 3. Simulation run time: Evaluate the simulation times for the different scenarios in each software

5 Methodology Create 3 simple test cases 1. Channel bend 2. Constriction 3. Flat bed (flood propagation) Run simulations for a range of flow conditions and cell sizes Number of scenarios = 280 Different grid resolution and boundary conditions 1 2 3

6 Test Case Bend Loss Objective: verify models estimate losses resulting from an abrupt bend Laboratory experiment (Malone 2008)

7 Test Case Bend Loss Empirical equations from laboratory results

8 Bend Model Configuration Scaled to match dimensions of laboratory test (Froude Model Law) 00, 45 and 90 degree scenarios 15ft grid resolution Free overfall conditions downstream boundary Same grid resolution across all three programs 12 computational nodes across the X- section Default parameters in all three programs Eddy Viscosity HECRAS 2D Not used M21 and TUFLOW - Smagorinsky No Calibration effort Inflow Boundary 12 grid cells 15ftX15ft Cross section A A Downstream Boundary

9 Bend Model Configuration Base 00 Degree 45 Degree Bend 90 Degree Bend

10 Bend Model Configuration Water Surface Elevation and Flow velocity monitored at two location: Upstream of the bend Point P1 After the channel bends Point P2 Base condition (00- bend) is used to compute friction loss H f-00 = WSEL at point1 WSEL at point 2 Output Location P1 Output Location P2 h b-45 degree = (Head loss in 45 degree Head loss in 00 degree) h b-90 degree = (Head loss in 90 degree Head loss in 00 degree)

11 Velocity Profiles HEC-RAS TUFLOW

12 Result: HECRAS 5.0.1

13 Result: TUFLOW

14 Result: MIKE21

15 Result: HEC-RAS, MIKE21 and TUFLOW

16 Bend Test Result Summary All models produced consistent results for various flows HEC-RAS showed highest head loss among the three programs MIKE21 simulated head loss in between HEC RAS 2D and TUFLOW TUFLOW resulted in smallest head loss among the three All three programs exhibited higher head loss in 90 degree bend Runtime 90 degree 36000cfs HECRAS (FM) = 41 mins 51 sec (8 Processors) MIKE 21 = 35mins 44 sec (8 Processors) TUFLOW = 22 mins 18sec (1 Processor)

17 Test Case Constriction Objective: Verify models estimate losses resulting from a constriction (eg. bridge abutments) Assess grid convergence in model results Validated against Federal Highways Administration Equations (Bradley, 1978)

18 Constriction Model Configuration Sudden contraction and expansion of flow path in a uniform rectangular channel 5ft and 10ft grid resolution. 11 flow scenarios (1000, cfs) Free overfall conditions downstream boundary Inflow Boundary Downstream Boundary

19 Constriction Model Configuration Base No Constriction 5ft grid cells With Constriction 10ft grid cells

20 Constriction Model Configuration Output Location P1 Output Location P2 Output Location P3 Base condition (no constriction) is used to compute friction loss Eddy viscosity HEC-RAS Not used TUFLOW and Mike21 Smagorinsky h L = WSEL at point1 WSEL at point 2 h c degree = (Head loss in constriction Head loss in base) = h L_cons h L_base

21 Velocity Profiles HEC-RAS TUFLOW Mike21

22 Result: HECRAS 5.0.1

23 Result: TUFLOW

24 Result: MIKE21

25 Result: HEC-RAS, MIKE21 and TUFLOW

26 Constriction Test Result Summary HEC-RAS 2D (Full momentum), Mike21 and TUFLOW produced consistent and reasonable results The results from all of these programs are slightly different As compared to the loss computed using FHA equation: HECRAS seems to slightly overestimated losses, results are closer to FHA on 5ft grid Mike21 matched the FHA results quite well with 5ft resolution grid and underestimates on 10ft grid TUFLOW simulated loss quite well with 5ft grid and slightly underestimated with 10 ft grid HECRAS Diffusion wave method is not suggested in this kind of problem Runtime (5ft resolution) HECRAS = 2 hrs 54 minutes (8 Processors) -- Computation Time step = 0.1 s MIKE 21 = 2 hrs 48 mins (8 Processors) -- Computation Time step = 0.1 s TUFLOW = 21 min 31 seconds (1 Processor) -- Computation Time step = 0.4 s

27 Test Case Speed of flood propagation Based on UK EPA benchmark test Objective: test is to assess the package s ability to simulate the celerity of propagation of a flood wave and predict transient velocities and depths. Aim: Compare model simulation run time The model scenario size has been increased from UK version to make a better estimate of the model simulation time (50,000 cells)

28 Flood propagation Model Configuration Flat bed 20m grid resolution 20m 3 /s inflow through a 20m opening P1 4000m P2 24hr simulation period Inflow Boundary 8000m

29 Model Results? Conclusions Channel Bend Loss All three software produce slightly different but consistent results HECRAS produced highest head loss, MIKE21 less than HEC-RAS but slightly more than TUFLOW and TUFLOW produced the least head loss Channel Constriction Consistent results across all the three software HECRAS and MIKE21 produced slightly higher head loss as compared to TUFLOW Difference in head loss computed in three programs Expected?

30 Conclusions Model Simulation Times? Is it an indicator of overall software efficiency? Parallel processing vs parallel simulations? Courant Criteria Other factors? Turbulence simulation (Kinematic viscosity) HEC-RAS Diffusive wave Use with caution GIS data model compatibility Mapping and cartographic options Single simulation vs production job Pretty pictures/animations Which program to use among the three? All

31 Thank You Murari Paudel, PhD, PE, CFM Wood Rodgers Inc. Sacramento, CA (916)