What will you learn? 15/01/2018

Transcription

1 Mekong River Commission Office of the Secretariat in Vientiane 184 Fa Ngoum Road, Ban Sithane Neua, P.O. Box 6101, Vientiane, Lao PDR Tel: (856-21) Fax: (856-21) Office of the Secretariat in Phnom Penh 576 National Road, no. 2, Chok Angre Krom, P.O. Box 623, Phnom Penh, Cambodia Tel: (855-23) Fax: (855-23) ISH Consultancy for the Development of Guidelines for Hydropower Environmental Impact Mitigation and Risk Management in the Lower Mekong Mainstream and Tributaries An Introduction to how Hydro- and Morphodynamic Models can help Test Effectiveness of Mitigation Guidelines and its Measures Kees Sloff and Jenny Pronker What will you learn? Introduction to model types and characteristics for hydrology, hydrodynamics, sediment and geomorphology (morphodynamics) for the Mekong Setting-up a model: data requirements Modelling mitigation in the Mekong (ISH0306) Lessons learned, blessings and cursus, and further usage 2 1

2 Guidelines and Recommendations for Mitigation of Hydrological and Flow Impacts Modelling What and why modelling? 3 Lancang The Mekong pathways Rainfall to runoff of flow and sediment Tributary run off Presence and operation of reservoirs modify flows and sediment transport River runoff hydrograph Evaporation/rain Storage, sedimentation Dams Tributary run off Evaporation/rain modified Hydrograph Sand hunger cascade All elements interact, and mitigation requires considering the entire system Storage, sedimentation Tonle Sap Salinity intrusion 2

3 Quantification needed How does the system work (how much flow can we expect, where does sediment come from, etc.)? How big is the risk and damage due to HPPs? Calculate the indicators, and test them (e.g., safety regulations, trans-boundary agreements) Calculate effectiveness of the mitigation measures (also w.r.t. costs and benefits) How do different schemes interact (cascade, cumulative impacts) and how do they affect the different coupled processes (such as flow and sediment)? How does mitigation affect the turnover of the HPP (next)? Etcetera Models PROBLEM: Physical models (laboratory scale models) Computational models (numerical models) Simple models: empirical, equilibrium states, spreadsheet Advanced models: 1- dimensional models, catchment models Advanced models: high-resolution 2 and 3-dimensional models ANALOG: ANALOG: ANALOG: 3

4 Use of physical scale models Useful for local complex 3D flows (near the dam and turbines) Much work and time consuming Most problematic to do the right scaling: - Use Geometrical scaling: ((keep the same) - For flow use Froude scaling (ratio inertia/gravity): - For flow use Reynold scaling (ratio inertia/viscous): - For sediment use Shields scaling: Impossible to satisfy all scaling rules Pak Beng hydropower project - Overall Hydraulic Physical Model Investigation of Pak Beng HPP Use of computer models Use for various different applications from small-scale to largescale (basin) models Physics represented with mathematical equations (obeying the laws of nature: conservation of mass, energy, momentum, etc.) Software tools require training and experience Easy to run scenarios, vary the settings, rerun simulations Averaging in space: replace the varying conditions on the smaller scale with empirical relations : this usually reduces the accuracy of the models, but increases the aplicability 4

5 Dimensions Full 3D DNS (CFD): Variations in time (unsteady) Spatial varying in 3 dimensions Details turbulence 2DV (width average): Variations in time (unsteady) Spatial varying in 2 dimensions 1D cross-section average: Variations in time (unsteady) Spatial varying in 1 dimension Delft3D Reynolds average 3D: Variations in time (unsteady) Spatial varying in 3 dimensions Averaged turbulence models (LES, eddy viscosity, k- epsilon) 2DH (depth average): Variations in time (unsteady) Spatial varying in 2 dimensions 3D processes parametrized 0D basin model (catchment models, waterbalance models) Hydrology models versus hydrodynamic models Used to calculate rainfall to runoff Usually catchment models and simple flow routing (0-D) Time steps in the order of weeks to months (output monthly average discharge, etc) Does not include a detailed physical description of fluids Used to calculate physics of flow processes (velocities, water levels, ) 1-D to 3-D models Time steps in the order of minutes to hours Also includes modules for sediment, geomorphology and water quality 5

6 Model elements for assessing main-stream dams 0D Lancang Rainfall to runoff Tributary run off 0D 2D/3D River runoff hydrograph Storage Evaporation/rain Dams 1D Tributary run off 2D/3D Evaporation/rain modified hydrograph HP model cascade HP model Storage Tonle Sap Salinity intrusion Environmental impacts and mitigation What to model? Hydrology, sediment yield Reservoir sedimentation, flushing, Erosion, fish habitat change, water levels,

7 Build on existing DSF for LMB, plus specials Additional or alternative components Flood Modeller Free (250 nodes) WUP-FIN (Tonle Sap) 13 Guidelines and Recommendations for Mitigation of Hydrological and Flow Impacts Modelling Setting-up models for the Mekong and its tributaries 14 7

8 Setting-up a model Step 1: numerical grid 1-dimensional: follows thalweg 2(3)-dimensional: cover wet area Grid size: hundreds of meters to kilometers (approx. > river width) Grid size: meters to several tens of meters (approx. > river depth) 15 Setting-up a model Step 2: topography (bed levels) 1-dimensional: follows thalweg 2(3)-dimensional: cover wet area Cross-sections prescribed (e.g perpendicular to axis) Bed level in each grid cel 16 8

9 Setting-up a model Step 3: conditions Initial conditions - Flow situation at time = 0 (start condition) - Bed sediment composition (grain size distribution) Boundary conditions - Water discharge series at inflow(s) - Water levels at outflow(s) - Sediment load at inflows 17 Setting-up a model Step 4: closure 1: Bed roughness - Chezy C or Manning s n - Bed forms - Vegetation 2: Sediment transport - Transport formula - Advection/diffusion 3: etcetera 18 9

10 Baseline data used for the new sediment/flow model (Delft3D) Topography (same baseline as used for DSF): Hydrographic atlas (1998) (soundings navigation program) SRTM data (2000) Bed composition: MRC DSMP (discharge sediment monitoring pro.) results Sediment inflow for calibration: Historic MRC DSMP (discharge sediment monitoring pro.) results Hydrology conditions: Discharges for BDP2030 development (Chinese dams and tributary dams) Follows from hydraulic run Dec.2015 version DSF Mekong Delft3D model Sediment inflow for reservoir runs: MRC DSMP predicted sediment loads for Chiang Saen Tributary loads (70% trap efficiency for future dams) 19 Setting-up a model Step 5: calibration and verification Use historic data to calibrate the model - Water levels and flow velocities at key-locations (for a range of discharges) - Bed-level changes observed in time, sediment loads - Sufficiently long periods Use another data set to do the verification simulation (with the settings obtained during calibration) 20 10

11 Setting-up a model Step 6: Runs run Take care of output (make sure indicators are quantified) Coupling of models for hydrodynamics and hydropower Think about scenarios and simulation period Etc. 21 Example use of model for sediment mitigation: Pushing the sediment balance buttons High velocity and turbulence: much transport Mostly sand, silt and clay Inflow of sediment High at begin of flood season transport Entrainment and deposition High shear stress: net entrainment Low shear stress: net deposition Outflow of sediment Releasing through the main outlets 11

12 River- en reservoir modelling Mekong ISH0306 Main channel indicators for flow and sediment Reservoirs (first basic in 1D, correct with 3D outcomes) Use 1D models as backbone for the more detailed reservoir models (input/output of discharges, water levels, sediment) 3D model Flow + sediment Flow + sediment Reservoir reach Crosssection 1D model 23 Example ISH0306: mitigation of risks/impacts in cascade Delft3D Q out (t) and Qs (t) DSF Tributary Q trib (t), Q trib,s (t) HEC-ResSim Pool level from hydropower model z(t) HEC-ResSim Outflows Q out (t) Upper model boundary Lower model boundary 12

13 Discharge (m 3 /s) Restrictions Changes in river-bed dynamics 15/01/2018 Example ISH0306: Simulating long term dynamics (cascade) 9000 Discharge at Chaing Saen E E E E E E E E+05 Hydrodynamic time (minutes) Pak Beng Luang Prabang Xayaburi Pak Lay (>)7 year Chiang Saen and tributary inflow Sanakham Vientiane Sediment transfer Paksane (or Pakse) Bed composition, grain size Reservoir sedimentation, sand bars 25 Hydrology HP design & operation Input DSF (BDP) hydropower modelling & Detailed reservoir models & Downstream models modelling Simulation without mitigation Simulation with mitigation sediment, fish,... Assessment without mitigation Propose mitigation measures Optimised operation without mitigation Optimised operation with mitigation Power & value loss Modelling set up & tools Optimise for power generation with/without mitigation 13

14 Guidelines and Recommendations for Mitigation of Hydrological and Flow Impacts Modelling Example ISH0306 case study and other examples 27 Setting-up the 3D model: physics Xayaburi reservoir (bridge Luang Prabang to Xayaburi): Lateral sand bars Long (days) 28 14

15 Setting-up the 3D model: physics Luang Prabang reservoir (downstream Pak Beng): turbulence at rock outcrops Long (days) 29 Schematisation Delft3D models ISH0306 Numerical grid Chiang Saen to Pakse: 6 sub models Bed level, velocities, water levels, etc. are computed in each grid cell Highest resolution in main channel 30 15

16 Mekong Delft3D model ISH0306 Min elevation ca 280 m Luang Prabang dam The model is based on the same data as used for the DSF models Operation level ca 275 m (so upper reach is still free flowing) Xayaburi dam Min elevation ca 230 m 31 Mekong River at Xayaburi Rocky patches Inflow from China Sandy beaches 32 16

17 Sand bars in bed-rock channel Eddies form sand bars Delft3D rock Eddy beach 33 Just downstream of Luang Prabang 17

18 Vectors = flow velocity 18

19 Hydraulic roughness (calibration) Bed rock n=0.045 Alluvial n=0.025 Chiang Sean to upstr. Vientiane Upstream Vientiane to Pakse 37 Sediment transport Class Grain size Model Gravel 4 mm Engelund and Hansen Coarse Sand 1 mm Engelund and Hansen Medium Sand 0.35 mm Engelund and Hansen Fine Sand mm Engelund and Hansen Silt&Clay (mud) > 0.06 mm Partheniades and Krone 38 19

20 Sediment load (m3/s) 1-Jan-30 1-Jun-30 1-Nov-30 1-Apr-31 1-Sep-31 1-Feb-32 1-Jul-32 1-Dec-32 1-May-33 1-Oct-33 1-Mar-34 1-Aug-34 1-Jan-35 1-Jun-35 1-Nov-35 1-Apr-36 1-Sep-36 Sediment load (m3/s) 1-Jan-30 1-Jun-30 1-Nov-30 1-Apr-31 1-Sep-31 1-Feb-32 1-Jul-32 1-Dec-32 1-May-33 1-Oct-33 1-Mar-34 1-Aug-34 1-Jan-35 1-Jun-35 1-Nov-35 1-Apr-36 1-Sep-36 15/01/2018 Sediment transport without dam Scenario 0.2 BDP 2030 Scenario 1.1 Run of River with dam 0.8 Luang Prabang, computed reference 0.8 Luang Prabang, computed with dams Silt and Clay Fine Sand Medium Sand Coarse Sand Gravel Silt and Clay Fine Sand Medium Sand Coarse Sand Gravel Annual Sediment Load at Luang Prabang Damsite Km 2038 Input to Xayabouri 10.7 Mt/yr Gr: <1% C Sand:6% M Sand 11% F Sand: 63% Silt+Clay: 20% 2.4 Mt/yr Gr: <1% C Sand:<1% M Sand: <1% F Sand: 2% Silt+Clay: 98% 39 Dam Operation Gates opened/closed automatically to control pool level according to operation rule (HP model) Use of PID controller 40 20

21 Elevation (m + MSL) 15/01/2018 Mitigation measures (some examples) Artificial flood aerator Fish passage Diversion channel Ramping (limit water level change) resettlement Sediment flushing and sluicing Sediment mitigation Progressive flushing and sluicing 360 Pak Beng 350 km 2188 Min water level LLW (MRC) m - 335m Luang Prabang Minimum bed level Mean bed level 330 km 2010 Operating level m Xayaburi dam km m Pak Lay (1) km m Sanakham km m (low) m (full) Chainage (km) 42 21

22 Sediment concentration (mg/l) Bed-level difference (m) Water level at the dam (m) Sediment concentration (mg/l) /01/2018 Some results: bed level at Pak Beng reservoir Pak Beng reservoir (smoothened bed-level sedimentation/erosion, long term) Without dam, 28 years With dam, 10 years With dam, 15 years With dam, 20 years With dam, 28 years Dam Erosion Sedimentation Water level at the dam (m) Distance from dam (km) Some results: flushing (dynamics) Ramp down 0.5 m/hr Pak Lay reservoir, increased draw down flushing Sediment outflow 0.5 m/hr 0.2 m/hr Silt&Clay Fine Sand Medium Sand Coarse Sand Gravel Water level dam , , , , Ramp down >2 m/hr Pak Lay reservoir, unconstrained draw down flushing Sediment outflow Silt&Clay Fine Sand Medium Sand Coarse Sand Gravel Water level dam Date Full draw down (to 225 m) with 0.5 m/hr ramp down; 0.2 m/hr ramp up Full draw down (to 229 m) with no Date constraints to ramping Q max =9743 m 3 /s Duration 10 days Inflow upstream ( 10 3 ton) Outflow dam ( 10 3 ton) Net ( 10 3 ton) Duration 14 days Inflow upstream ( 10 3 ton) Outflow dam ( 10 3 ton) Net ( 10 3 ton) Gravel Coarse Sand Medium Sand Fine sand Silt/Clay , ,076.6 Total , ,309.9 Q max =18520 m 3 /s 44 22

23 Some results: shear stress during flushing During normal operation or after full draw down During a rapid draw down 45 Draw down flushing simulation 0.1 m/hr 0.2 m/hr 46 23

24 Example Taiwan: turbidity current venting Power outlet converted to low level sediment sluice (2010) Release clear water from surface spillways to dilute turbid outflow

25 Shape of the Shihmen Reservoir and (planned) sluices/tunnels Dahan River TDR monitoring 50 25

26 2DV model (Delft3D Open Source) 50 layers Dx = 200 m 51 Sediment concentration 2DV 52 26

27 53 Conclusions from the Taiwan modelling work Sluicing tunnel has highest V eff Location is important Reservoir operations as well Delft3D (RANS) able to capture turbidity currents in Reservoirs Good comparison with TDR (concentration measurements) 54 27

28 Guidelines and Recommendations for Mitigation of Hydrological and Flow Impacts Modelling Final comments, lessons learned and take-home messages 55 Interpretation and uncertainty Principle of least astonishment Model is as good as its input: garbage in = garbage out Ockham s raiser: simple theory is best 56 28

29 Some lessons learned System approach needed (coupled basin model, 1D model and 2D/3D models) for the larger picture and interactions between processes and reservoirs Integrate detailed 2D/3D models with larger-scale 1D and basin models Accuracy of models: - Limited by the data availability - Fit for purpose means that tools are adequate for making the right decisions - Remaining uncertainty translates to adaptive approach and monitoring program 57 Some more lessons learned Monitoring is an essential part of measuring success of the measures Mitigation can cause new problems Modelling is a powerful tool if sufficient data is available to calibrate the model 58 29

30 Concluding remarks For reservoir sediment management it is necessary to consider the full river system (all pathways of sediment, pulses, etc.) Sediment management in a cascade will benefit from joint operation Flushing and sluicing is modeled best with a 2DH or 3D model (flushing channel, dynamics of draw down, etc.) Hard flushing can be damaging for the downstream: use full dynamic model Models can be used to simulate the impacts of sediment augmentation / nourishments, such as the placement of sediment dikes Allow time for analyzing the uncertainties in the model (sensitivity analyses, stochastic modeling) Take home message: use model as a tool Models are not an exact copy of reality, just an approximation Future conditions are uncertain, so they cannot be exact Do not forget their purpose 30

31 Questions? Nam Gnoung Dam (Theun-Hinboun) 31