Available online at www.sciencedirect.com ScienceDirect Procedia Engineering 154 (2016) 536 – 543 12th International Conference on Hydroinformatics, HIC 2016 2D SEDIMENT TRANSPORT MODELLING IN HIGH ENERGY RIVER – APPLICATION TO VAR RIVER, FRANCE Elodie Zavatteroa,*, Mingxuan Dua, Qiang Maa, Olivier Delestreb,c, Philippe Gourbesvillea,b a Innovative CiTy lab Polytech Nice-Sophia / Nice Sophia Antipolis University, 1-3 Boulevard Mtre Maurice Slama, 06200 Nice, France b Polytech Nice-Sophia / Nice Sophia Antipolis University, 930 Route des Colles, 06410 Biot, France c Laboratory J.A Dieudonné - UMR 7351, CNRS Université de Nice - Sophia Antipolis 06108 Nice Cedex 02 Abstract Specific environments such as the French Mediterranean coastline are characterized by steep catchments associated to a climate affected by high intensity rainfall events (over 1000 mm/hour) In such costal catchments, development of human activities and urban developments is very challenging and requests to address properly the issue of vulnerability for both water resources and infrastructures In addition, the coastal area is frequently characterized by complex relationships between underground resources, rivers and sea Due to the geological and the morphological conditions, the sediment management is a main issue for local communities Under specific conditions and with a clear understanding of performance limits, 2D hydraulic modelling is a meaningful approach that can provide an accurate view on the physical processes within the river such as morphological dynamic This paper develops a suitable methodology to build a 2D surface water flow model, including sediment transport in the lower Var valley, France, with MIKE 21 FM (DHI) modelling system A sediment transport model – MIKE 21 Sand Transport (ST) is used to find out the movement of sediments along the river MIKE 21 ST is interfaced with the hydrodynamic module, MIKE 21 Flow Model (FM), which simulates the water level variations and flow in response to a bunch of forcing functions The results allow to understand the sediment transport along the Var river and to analyse the impacts on the river-aquifer exchange areas This approach can lead to carry out a model able to simulate properly the behaviour of physical processes within the river bed © 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license © 2016 The Authors Published by Elsevier Ltd (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-reviewunder under responsibility of organizing the organizing committee HIC 2016 Peer-review responsibility of the committee of HICof2016 Keywords: sediment transport, free surface flows, 2D hydraulic modelling, river-aquifer exchange, MIKE 21 FM, Var river, France * Corresponding author Tel.: +33 651447802 E-mail address: zavattero.elodie@hotmail.fr 1877-7058 © 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the organizing committee of HIC 2016 doi:10.1016/j.proeng.2016.07.549 Elodie Zavattero et al / Procedia Engineering 154 (2016) 536 – 543 Introduction Groundwater is one of the most important water resources for people in the world in both rural and urban areas Sometimes, freshwater sources are extremely vulnerable due to their strong exchanges with the river in terms of quantity and quality With decrease of water resources associated with the climate change and growing demands, scientists need to consider groundwater and surface water as a single resource Indeed, these two entities are interconnected and may have a deep impact between each other depending on their quantity or quality Beyond this phenomenon, rivers are strongly influenced by their morphology [1,2] River morphodynamics is the study of physical processes that control and maintain river environment [3] Hence, understanding these complex morphological changes is a key issue for the knowledge of exchange areas between the river and the aquifer 1.1 The lower Var valley The lower Var valley river is located in the southeast of France (Fig 1) This 22 km section connects the mountainous area and the Mediterranean Sea [4] It drains water of a catchment of 2893 km2 The yearly average discharge is 50 m3/s, while the highest measured instantaneous discharge during flood peak can reach 3750 m3/s Fig (a) North part of the lower Var valley; (b) southern part of the study area The alluvial aquifer is one of the main drinking water resources for the local population Indeed, statistical results of recent years mentioned that 50 million m3 are extracted from this unconfined aquifer [5] However, since several decades, the human activities (sediment extraction, urbanization, etc.) influence the water table behaviour 1.2 Key issue of morphological dynamic in the Var river Agricultural development, urbanization and sediment extraction, in the lower Var valley, caused morphological change Indeed, dykes were built to protect population and agricultural land from floods; 15 million m3 have been extracted from the riverbed (corresponding to 150 years of natural sediment supply [6]) In addition, weirs were 537 538 Elodie Zavattero et al / Procedia Engineering 154 (2016) 536 – 543 constructed to increase the groundwater table Actually, the riverbed extends from 150 to 280 m, this led to an increase of the water velocity Thus erosions happened gradually and have been observed in many places along the river Hence, these physical processes affect exchange areas between river and aquifer According to measurements recorded by ADES Eau since 1970 (Fig 2), the groundwater table decreases after weirs construction and after 2011 flood event; increases after the 1994 flood event which has destroyed weir and Currently, a 3D hydraulic model of the alluvial aquifer in the lower Var valley describes the groundwater table at any location [7] Fig Groundwater table recorded close to weir in downstream part of the Var river from 1970 to 2012 (Source: Portail d’accès aux données sur les eaux souterraines, http://www.ades.eaufrance.fr/) Beyond the morphological change due to human activities, physical dynamics of the riverbed is controlled by flood period which induces a reorganization of the sediments In order to understand the morphological dynamics and its influence on the river-aquifer exchange areas, a numerical model must be set up This paper presents the methodology to achieve a numerical model and the approach used to simulate properly morphological dynamics in the Var river Data and Methodology The modelling of stream evolution requires computational modules to simulate the hydrodynamics and bed evolution The 2D approach allows obtaining an accurate view of the processes if the development process is managed in the proper way 2.1 Data The period from 2009 to 2013 is significant for the Var river Indeed, not only the river morphology has changed (Fig 3), but also it contains two important events: a flood happened on November 2011 followed by a drought on July and August 2012 The return period of the 2011 flood event is equals to 10 years according to the Gumbel law and the peak discharge was 1190 m3/s Actually, a deterministic distributed hydrological model is setting-up [8] to provide discharge which is upstream boundary condition for 2D hydraulic model of the lower Var valley Elodie Zavattero et al / Procedia Engineering 154 (2016) 536 – 543 Fig Comparison between the drainage lines from 2009 to 2014 in the lower Var valley In order to evaluate the morphological dynamic of the Var river two digital elevation models have been used: 2011 and 2013 Such high-resolution and fully dimensional models of channel topography offer real potential to develop new ways of characterizing rivers and understanding braiding mechanics [9] The morphological change resulting from the 2011 flood event can be revealed by subtracting the two surface models (Fig 4) However, while this approach has much to offer in terms of geometrical characterization, the accuracy of the data affects the results and this method presents some limitations [9] Fig Comparison between DEMs to identify sedimentation and erosion during 2011 flood event 2.2 Hydrodynamic model Simulation of hydrodynamics has been carried out by solving 2D shallow water equations of mass and momentum thanks to the HD model of MIKE 21FM To solve these equations using finite volume numerical methods, unstructured grids were used to define the topography of the study area [10] Simulations generate unsteady two-dimensional flows in one layer fluids (vertically homogeneous) Flow and water level variations are described by the Saint-Venant equations (the conservation of mass and momentum integrated over the vertical) which are the following: 539 540 Elodie Zavattero et al / Procedia Engineering 154 (2016) 536 – 543 wh wt wp wx wq w § p² · w § pq · wz wx â h wy â h wx á h Đ wpa à ă wt ă ¸ gh gp p² q² C ² h² ªw w º ( hW xx ) ( hW xy ) ằ U w ôơ wx wy ẳ (2) U w â wx wq (1) w wy ă wt wp wz w êw p q Đ q · w § pq · ( hW yy ) ( hW xy ) ằ ă ă gh gq ô wy â h wx â h wy wx U w wy C h ẳ w Đ wpa à ă U w â wy h (3) where, is the water depth [m], ݓൌ ܴ െ ܫis the net incoming flow rate [m/s], ݖis the surface elevation [m], ሺǡ ݍሻ ൌ ሺ݄ݑǡ ݄ݒሻ are the flux densities in directions x and y respectively [m²/s], ܥis the Chezy resistant [m1/2/s], ݃ is the gravitational gravity [m/s²], is the atmospheric pressure [kg/m/s²], ߩ௪ is the density of water [kg/m3] and ߬௫௫ ǡ ߬௬௬ ǡ ߬௫௬ are the components of effective shear stress The aim of hydrodynamic part is to compare suitable way to build a 2D surface flow model in the lower Var valley with MIKE 21FM Focus is given in the implementation of hydraulic structures such as weirs, which are directly included in the topography or specified by an empirical law 2.3 Sediment transport model Simulation of bed evolution has been carried out with Sand Transport module in Mike 21FM which calculates the sediment transport capacity, the initial rates of bed level changes and the morphological changes for non-cohesive sediment due to currents The sediment transport computation is based on hydrodynamics conditions and sediment properties ST model takes in account bed load and suspended load The first one is controlled by shear stress or stream power per unit and reacts instantaneously with the flow The second one is characterized by a phase-lag in the transport compared to the flow Bed load does not need advection-dispersion modelling, but two important effects must be taken into account: the deviation of the direction of the bed shear stress and the effect of a slopping river bed [9] On the other hand, modelling of non-cohesive suspended sediment in a fluid can be described by a transport equation for the volumetric sediment concentration [11] Van-Rijn model has been used for the sediment transport model [12,13] Sbl 0.053 * T 2.1 0.3 D* ( s 1) * g * d50 f * ca * V * h S sl (4) (5) where, ܵ is the bed load [m²/s], ܵ௦ is the suspended load [m²/s], ܶ is the non-dimensional transport stage depending on critical friction velocity and effective friction velocity, כܦis the non-dimensional particle parameter, ݏ is the relative density of the sediment, ܿ is the volumetric bed concentration [m], ݂ the correction factor for suspended load based on bed load, ܸ is the velocity [m/s] The key parameter to determine the level change is the rate of bed level change at the element cell centers This evolution of the bed is obtained with Exner equation which is the sediment continuity equation: ( n 1) wz wt wS x y 'S wx wy wS (6) Elodie Zavattero et al / Procedia Engineering 154 (2016) 536 – 543 where, n is the bed porosity, z is the bed level [m], ܵ௫ is the bed load or total transport in the x-direction [m²/s], ܵ௬ is the bed load or total transport in the y-direction [m²/s], οܵ is sediment sink or source rate Then, the bed is updated continuously through a morphological simulation based on the estimated bed level change rates The aim of sediment transport part is to compare suitable way to build a model able to represent the morphological dynamics in the lower Var valley with MIKE 21FM and ST module Focus is given in the construction of the riverbed geometry and the equations used to get the morphological changes Results and discussions 3.1 Hydrodynamic results First of all, different geometries have been compared for the simulation of hydrodynamic The strategy is discussed according to mesh resolution and needs regarding operational management In order to obtain an efficient Mike 21FM model, several meshes have been created to simulate the same flood event The study area has been represented by triangular discretization with different resolutions (5 m, 10 m, 15 m, 20 m and 25 m) Afterwards, triangular mesh has been compared with quadrangular mesh Finally, hydraulic structures have been implemented with Villemonte weir formula [14], which is an empirical law The flood event chosen for this set of tests is from 3rd October 2015 to 6th October 2015, because of the availability of observed data The comparison of models and observed data (Fig 5.a) suggests that the 10 m resolution is the most accurate The triangular discretization is more efficient to simulate the flow over the weir (Fig 5.b) Implementation of structures requests many parameters and is not really representative of weir effects (Fig 5.c) Furthermore, regarding to the velocities (Fig 5.d) the use of empirical law for structures is the least suitable way to simulate weirs effects Fig Comparison of water depth for (a) different resolutions; (b) different types of mesh; (c) different implementations of weirs; (d) comparison of velocities for triangular mesh, triangular mesh with structures and quadrangular mesh 541 542 Elodie Zavattero et al / Procedia Engineering 154 (2016) 536 – 543 3.2 General morphological evolution In the lower Var valley, one of the main issues is to represent the weirs effects on the morphological dynamics The sediment transport model takes into account bed load and suspended load Consequently, simulation of the 2011 flood event changes the bed level including the weirs Hydrodynamics results demonstrated that implementation of structures disapprovingly influences the flow fields But, three models have been compared to find an acceptable solution responding to the weirs effects on the morphology: the first one is defined by an uniform grain size, the second one is implemented with structures and the third one is defined by non-uniform grain size in order to stop the bed load which destroyed the weirs The comparison of the bed level change for the three models (Fig 6) shows that implementation of structures is the most appropriated to simulate physical processes in the riverbed for this part of the Var river Fig Comparison of bed level change simulated by (a) ST model with uniform grain size, (b) ST model with structures and (c) ST model with non-uniform grain size The calculation of the bed evolution is based on hydrodynamics and sediment properties Here, only sediment properties are not enough to describe the weirs effects on the morphological dynamics Conclusion The topography resolution is a key parameter and, if properly adjusted, can allow developing a hydrodynamic model able to simulate the behavior of weirs Sediment transport model has been built with discretization extracted from the hydrodynamic model However, while triangular mesh (with a 10 m resolution) seems to be the best way to represent the flow fields over the weirs, it is not a suitable solution to simulate the bed evolution on the lower Var valley Consequently, it requires implementation of structures with empirical law For more details, it should be interesting to analyse the sensitivity of the model to use different sediment transport equations and apply this methodology on meandering parts of the Var river Finally, sedimentation area will be characteristics of sealing area for the exchange between river and aquifer Acknowledgements This research is currently developed within the AquaVar project with the support of Metropole Nice Côte d'Azur, Agence de l'Eau Rhone Mediterranéen, Conseil Départemental 06 and Meteo France This work benefited from the data provided by the Metropole Nice Côte d'Azur and by Conseil Départemental 06 The results were provided by Elodie Zavattero et al / Procedia Engineering 154 (2016) 536 – 543 543 DHI Software (MIKE 21 HD FM) Then, DHI is acknowledged for the sponsored MIKE Powered by DHI licence file References [1] MG Kleinhans, Sorting out river channel patterns, Progress in physical geography 2010, 34: p 287-326 [2] G Seminara, Meanders, Journal of Fluid Mechanics 2006, 551, p 271-297 [3] T Iwasaki, Numerical simulation of bar and bank erosion in a vegetated floodplain: A case study in the Otofuke River, Advances in Water Resources 2015, In Press Corrected Proof [4] C Potot, Etude hydrochimique du système aquifère de la basse vallée du Var, apport des éléments traces et des isotopes (Sr, Pb, δ18O, 226, 228Ra) PhD thesis Université Nice Sophia Antipolis, France, 2011 [5] M Moulin, Nappe de la basse vallée du Var (Alpes-Maritimes), suivis 2006 quantité et qualité BRGM, France, 2009 [6] K Souriguère and C Ceraulo, SAGE Nappe et Basse vallée du Var 2014, https://www.departement06.fr [7] M Du, E Zavattero, Q Ma, O Delestre and P Gourbesville, 3D hydraulic modeling of a complex alluvial aquifer for groundwater resource management, Submitted to International Conference on Hydroinformatics 2016 [8] Q Ma, E Zavattero, M Du, V N Duong and P Gourbesville, Assessment of high resolution topography impacts on deterministic distributed hydrological model for extreme rainfall-runoff simulation, Submitted to International Conference on Hydroinformatics 2016 [9] J Brasington, J Langham and B Rumsby, Methodological sensitivity of morphometric estimates of coarse fluvial sediment transport, Geomorphology 53 (2003), 299-316 [10] M.S Darwish, and F Moukalled, TVD shcemes for unstructured grids, International Journal of Heat and Mass Transfer 2003, 46: p.599- 611 [11] DHI Software (2016), MIKE21 and MIKE Flow Model FM: Sand Transport Module, Scientific Documentation, Danish Hydraulic Institute, Horsholm, Denmark [12] L Van Rijn, Sediment transport, Part I: Bed Load Transport, Journal of Hydraulic Engineering (1984), 110 (10): p 1434-1456 [13] L Van Rijn, Sediment transport, Part II: Suspended Load Transport, Journal of Hydraulic Engineering (1984), 110 (11): p 1613-1641 [14] DHI Software (2016), MIKE21 Flow Model FM, User Guide, Danish Hydraulic Institute, Horsholm, Denmark ... should be interesting to analyse the sensitivity of the model to use different sediment transport equations and apply this methodology on meandering parts of the Var river Finally, sedimentation... for 2D hydraulic model of the lower Var valley Elodie Zavattero et al / Procedia Engineering 154 (2016) 536 – 543 Fig Comparison between the drainage lines from 2009 to 2014 in the lower Var. .. bed load or total transport in the x-direction [m²/s], ܵ௬ is the bed load or total transport in the y-direction [m²/s], οܵ is sediment sink or source rate Then, the bed is updated continuously through