ABSTRACT. The uncertainty in urban drainage water quality modelling is highly relevant in any practical application. Several models are available in the literature for such tasks, and one of the most challenging choices is selecting the most appropriate approach for the specific application. The Bayesian Model Averaging approach attempts to support the modeller in such choices by providing a method to identify and select the best-performing models and average their output response to reduce the related uncertainty. This method dates back to studies conducted by Bates and Granger (1969), whose analysis used model-averaging techniques in economic forecasting and determined that a pooled forecast of competing models outperformed any single model’s prediction. Techniques such as equal weight, Granger-Ramanathan averaging, and Bates-Granger averaging linearly combine the deterministic model outputs into another single-point deterministic forecast. This research applies the Bayesian Model Averaging to an actual catchment and is compared with several single water quality models. The analysis showed that the Bayesian Model Averaging approach outperformed all single-model applications.

ABSTRACT. The Jaunay reservoir is a drinking water reserve of 3.7Mm3 which extends over 114 ha and stretches over 8 km. It results from the construction of a 200 m long dam located in Landevieille (Western France). A 1D MIKE-HYDRO-River model was developed to understand the hydrodynamics of the reservoir over the years 2016 to 2020 and to simulate the dilution of several water quality parameters. The 1D model considers a succession of 70 cross sections and calculates the heights of water and flows every 100 m. The daily flows measured at the Réveillère hydro station located 2 km upstream the reservoir, are injected directly into the upstream boundary condition of the model. A rain-flow model (GR4J) is used to assess lateral flows. Upstream pollutant inputs were determined based on the measures on this station. The water quality modeling was carried out with two approaches: Macropollutants by coupling the EcoLab module to the 1D hydrodynamic model MIKE HYDRO River and dispersion of chlorides and micropollutants with the Advection-Dispersion module of a conservative tracer. The model considers 7 state variables: Biological Oxygen Demand (BOD), Dissolved Oxygen (DO), Chlorophyll a (CHL), Ammonium (NH4), Nitrite (NO2), Nitrate (NO3) and Phosphate (PO4). The model was calibrated over 2019-2020 using in situ measurements. The model was then compared with available measurements for 2019-2020 period and the results present a good correlation for oxygen and nutrients. For Chlorophyll a, the order of magnitude of the base levels and the peaks were well reproduced, but the results do not show the two seasonal peaks (summer bloom and autumn bloom) observed on the measured data. Indeed, the present model was unable to reproduce the autumn bloom correctly, mainly because a remineralization of nutrients and an interaction with sediments which were not considered. Measured and modelized data were then compared for each one of the 7 state variables using a F-test and with 95% confidence they present no significant statistical differences. Despite its limitation, the 1D water quality model could be a simple and efficient solution to estimate the major impacts due to climate change, to the evolution on the activities upstream the reservoir and other projects that may influence on the water quality dynamics.

ABSTRACT. The study of the aquatic habitat requires a holistic analysis, the relationship between hydraulics and biota being considered in all situations, not only during extreme conditions. Several approaches have been proposed to characterise the physical habitat based on geomorphologic, hydrologic or hydraulic criteria. Currently, there are various 1D or 2D numerical hydrodynamic tools that integrate the biological requirements of one or several target species typically of fishes, but not exclusively, that allows for modelling the weight usable area (WUA). However, one of the main limitations of these eco-hydraulic tools is typically related to computational cost, especially in 2D and 3D modelling. The simulation of flow properties in natural conditions, large river stretches, or the domain discretization with millions of elements usually requires high computational effort. The application of high performance computing (HPC) techniques, particularly general-purpose computing on graphics processing unit (GPGPU), instead of the traditional central processing unit (CPU) computing, allows for overcoming this limitation. A GPU-based hydraulic code parallelised version of Iber, called R-Iber, performs as a fully-integrated GPU-parallelised tool to carry out high-resolution and long-term eco-hydraulic numerical modelling. The benefits of using this GPU-parallelised tool are: i) speed-ups of two orders of magnitude in the computational time compared to traditional computing, ii) the use of a specific boundary condition called ‘stepped discharge’ for the application of the Instream Flow Incremental Methodology (IFIM) in a unique model that reduces the computational time to a minimum, and iii) the capability of simulating high-resolution and long-term in long-river stretches –even entire rivers– in a suitable timeframe.

ABSTRACT. Groundwater modelling during geotechnical work phase of a civil engineering project - Case study of construction an underground stormwater network and infiltration gallery for long-term management of runoff from a bridge - Davenport railway grade separation, Toronto (Canada). The project involves the construction of a 570 m long bridge with two access ramps, allowing rail traffic to pass each other without interruption. The hydrogeological component of this project involves the installation of approximately 1,300 m of underground stormwater network and an infiltration gallery. The objective of this paper is to demonstrate the advantages of hydrogeological modeling based on a finite element method (Feflow) in the context of infrastructure projects. The hydrogeological model was developed using public databases and maps and data collected along the railway alignment. The calibration of the model was carried out automatically using the PEST software which made it possible to model the heterogeneity of aquifer permeabilities, including that of a buried channel, and thus to replicate the complex piezometry with vertical gradients of about ten meters. The unsaturated zone was modelled using Richards' equation. The dewatering zones of influence as well as the impacts on surface water were evaluated by combining the method of particle tracking and the simulation of solute transport. The results show that the hydraulic impact of the project on natural water bodies will remain negligible to moderate and under control.

ABSTRACT. Drinking water shortage is a global problem. Microorganisms compromise water safety and can cause waterborne diseases to the consumer. Exposure of water to germicidal UV-C radiation can lead to the inactivation of a wide range of pathogens without the formation of disinfection by-products. The performance of a UV disinfection reactor depends on the accumulated radiation dose received by the microorganisms present in the treated water. It is critical to guarantee that all the microbial particles receive the required threshold level of radiation. For reactors with complex geometries (with preferential routes, some degree of vorticity, or when dead zones are present), it becomes necessary to integrate radiation distribution and microbial trajectory to accurately calculate the radiation dose received by the microbial population. A rigorous calculation of irradiance distribution is also critical when the reactor's geometry is complicated or when the lighting sources generate intricate radiation patterns inside the reactor, such as the non-homogeneous light distribution provided by LED. This work presents the development of a novel methodology for the simulation of photochemical processes for water disinfection using computational fluid dynamics (CFD). A new approach was implemented to calculate and visualize the disinfection performance as the microorganisms move along the photoreactor. Hydrodynamics and microorganisms' statistical trajectories were computed using the discrete phase model. The radiation distribution in the reactor was calculated using the discrete ordinate method. The local values of incident radiation were integrated over each statistical trajectory path to get the accumulated dose received for each microbial particle. The coupling in situ of the cumulative radiation dose with the inactivation kinetics allows monitoring of the disinfection process concurrently with the particle tracking. The developed tool enables the evaluation of photoreactor efficiency in each reactor position, a useful feature for optimizing and scaling up complex geometries.

ABSTRACT. Wastewater-derived contamination is one of the key causes for surface water bodies failing the European Commission requirements for good chemical status. Assessment tools that determine the contamination of surface water are pivotal as they can help to identify the stretches where microcontaminants level are estimated to be high and support monitoring and mitigation strategies optimization. The EU policies led to the production and the centralization of sanitation and hydrological datasets. Notably: (i) WWTP discharge, (ii) river flow conditions, and (iii) monitoring observations (for calibration/validation), are valuable information layers allowing microcontaminants fate and transport (MFT) modelling of wastewater-derived contamination in rivers. wOtter is a recently released fast and open-source MFT code, capable of simulating contamination across the EU hydrographic network fine-tuned for using above mentioned publicly available datasets. The produced MFT model in its early release was calibrated for a lumped contaminant over 273 EU sites. This study presents an improved approach to modelling the attenuation of a multitude of microcontaminants at EU scale, by assuming that the attenuation rates follow a statistical distribution. This results in higher rates of attenuation when highly degradable compounds are present initially, and lower rates of attenuation when only persistent contaminants remain. The new calibrated, model reaches an R^2 of 78.5%. Results exploitation at River Basin District level support micropollutant monitoring campaign adjustments and mitigation strategies based on WWTP upgrades.

ABSTRACT. AquaVar system is a DSS based on deterministic modeling tools that are capable of representing the full hydrological behavior of the Var catchment (about 3000 km2) located in the French Riviera. The architecture of the AquaVar system is based on the coupling of 3 models (hydrology, free surface flows and underground resources) that are operating in real time to produce a 3 days forecast of the conditions for all the 3 models. The system can also be used to reproduce hydrological phenomenon and build scenarios in order to help decision makers or to discuss technical issues. In 2022, a severe drought affected the Var river basin in Alpes Maritimes, France. This caused the lower Var Aquifer to reach such a low piezometric level, which resulted severe issues for the drinkable water extraction and distribution systems. The AquaVar system is also used to reproduce the long-term phenomenon to assess the mid-term situation (3-6 months). This will allow Eau d’Azur to increase the understanding of the drought of 2022 and try to anticipate what could happen in 2023, in the event of a more severe drought.

ABSTRACT. We here report an interesting case study of an evaluation of a tree-based model predictive control solution (TB-MPC) to advice operators on the optimal response to a forecast compound flood event. Tree-based model predictive control was implemented in the decision support system for the North Sea Canal, the Netherlands and tested numerically using a 1D hydrodynamic model and measured data.

The North Sea Canal is connected to the North Sea by the largest pumping station of the Netherlands at IJmuiden and has uncertain inflows coming from ensemble weather forecasts applied with a rainfall-runoff model. While the canal conducts the drained water from the lowlands to the sea, the water levels should be kept within a certain range for safety and shipping, and the cost of pumping should be as low as possible. Deterministic (single forecast) linear MPC is currently used in operations to determine when to use each individual pump of the pumping station and when to spill water under gravity through the sluices next to the pumping station. Now that ensemble inflow forecast has become available for this system, we set to explore the feasibility, advantages, and limitations of using TB-MPC with a linear optimization model adapted from the original model.

This case-study is interesting because: 1) Real data is used on a real decision support system to perform a closed-loop analysis for two flood events using tree-based as well as deterministic MPC. 2) Applied case study for a real rainfall event. 3) It is integrated into a decision support system: in case it is desired, just by pushing a button the pump settings computed by tree-based MPC are automatically executed.

ABSTRACT. Inundation simulation is essential to flood disaster management, providing important information about distribution of water and risk areas, and thus corresponding disaster management strategies against flood can be developed. To analyse the outcomes of disaster management strategies against flood, the cellular automata (CA) concept was applied in an agent-based model (ABM), Netlogo, to perform 2D inundation simulations and agent simulation. The model utilised a simple rule to redistribute the water between grids instead of solving complex equations. The model performance of inundation simulation was evaluated by comparing the model outputs with three benchmark cases and two historical events. Then an agent simulation coupled with inundation simulation was conducted for checking the practicability of the inundation-agent coupled model. The aims of this study are to establish a simple 2D overland flow model in an ABM and to seek the possibility of conduct inundation-agent coupled simulation in the ABM. The results revealed the inundation-agent coupled model capability of conducting inundation simulation and the potential for generating plausible outcomes of the interaction between flood water, response strategies and human behaviour in a complex system during a flood event. The model can be a valuable tool for investigating and improving existing disaster response strategies and for enhancing community-based disaster management.

ABSTRACT. In the southeastern region of France, floods and droughts affect the mountainous catchments including the Paillons and the Var on the west and east of Nice city, France. On the French Riviera, as in many Mediterranean coastal cities, urbanisation and economic development are on the rise and can be deeply affected during extreme convective events such as the Alex storm recorded in October 2020 or the major flood of 2019. This situation motivated the development of the AquaVar decision support system since 2013. The system is based on deterministic modelling and tailored toward modelling runoff and groundwater flow processes for flood risk and water resources management in real time. In the Paillons catchment, a similar tool is expected to address the risk of urban flooding within the east part of the Nice city. The upstream part of the catchment is steep and forested. Meanwhile, a densely urbanized area and covered river are located downstream. A flood monitoring system has been implemented since 1983, but gaps and errors pose a challenge in understanding the water cycle and easing the management of flood and droughts events. The present study shows the ability of the developed rainfall-runoff and groundwater flow models to reproduce the behavior of the catchment. Robust modelling tools and an effective strategy were useful in dissecting how to reproduce flow processes over long periods and fill gaps in missing data. The simulation system orchestrates the MIKE SHE and FEFLOW modelling tools. The needed discharge and groundwater information represent valuable insights for water resources and flood risk management. They also serve as boundary conditions to specific modelling tool that address flood and groundwater mapping. Results show strong variability in rainfall records. Interestingly, most flood events are well reproduced and suggest the tools to be useful for integration into a real time decision support system. The models provide the much-needed information to fill gaps in knowledge of hydrological processes in the Paillons catchment and contribute to improve the understanding of its global water cycle. The models can be used within the DSS and so, provide a real-time support for the decision-making process.

ABSTRACT. Besides being a landlocked country, Armenia has limited access to the water due to its geography and the lack of water resources sharing agreements with its neighbors: the decision to share 50/50 the waters from the Akhuryan river (western border with Turkey) dates back to Soviet age and besides the efforts, the sharing of Arax river’s resources (southern border with Turkey and Iran) still is a sensible topic. Thus, Climate change is a crucial issue for Armenia specially because the exacerbation of droughts and soil degradation, and the potential for water scarcity on the next few years (-4% in 2040 and -10% in 2070 under RCP6.5 scenario and -5% in 2040 and-15% in 2070 under RCP8.5 scenario). Hence it is vital to estimate the evolution of water resources (surface water and groundwater) due to climate change and compare it the progression on demands due to socio-economic dynamics. A model was then designed to estimate in an annual timestamp the amount of unmet water needs, the associated cost of inaction, the expected investments on works to cope with climate change and the net benefits of those works under ceteris paribus analysis. This model showed that the unsatisfied water uses could generate losses of about 330MUS$/year in 2040 and 820MUS$/year in 2070 and this only considering domestic uses and the main Armenian economic activities, i.e., Agriculture (14% of the total GDP and 1/3 of all labor force), Fish farming (85% of Armenia’s total fish production and about 3% of GDP depend directly from a the Ararat Valley, a fertile area enclaved on semi-arid region) and Energy generation (that impacts virtually every economic activity). The hydro-economic tool was built to allow decision-makers to calculate the cost benefit analysis (CBA) of important water infrastructure projects and adaptation measures, study the priorities and phasing of those works and measures, and being easily adjusted to integrate changes on socio-economic parameters (return rates, exchange rates, country risk) on hydrological parameters (RCP scenarios, change of farming cultures, dissemination of weak blue water footprint tools, distribution losses reduction) and legislation (water permits). The main result is that an ensemble of adaptation measures expected to be implemented until 2030 by the Armenian government, which investment amounts to 128.08MUS$, produce a benefit of 49.07MUS$ in 2040 and 152.52MUS$ in 2070 in present values for a discount rate of 9%. Even with an increased country risk such that the discount rate would rise to 14%, the Net Present Value (NPV) would still be positive at 24.82MUS$.

ABSTRACT. This work presents the development and applications of a new, open-source toolbox that aims to provide automatic identification and classification of hydroclimatic patterns by their spatial features, i.e., location, size, orientation, and shape, as well as the physical features, i.e., the areal average, total volume, and spatial distribution. The highlights of this toolbox are: (1) incorporating an efficient algorithm for automatically identifying and classifying the spatial features that are linked to hydroclimatic extremes; (2) use as a frontend for supporting AI-based training in tracking and forecasting extremes; and (3) direct support for short-term nowcasting of extreme rainfall via tracking rainstorm centres and movement. The key design and implementation of the toolbox are discussed alongside three case studies demonstrating the application of the toolbox and its potential in helping build machine learning applications in hydroclimatic sciences. Finally, the availability of the toolbox and its source code is included.

ABSTRACT. This study is led within the CIPRHES project (Perrin et al., 2022) that focuses on the improvement of low-flow forecasting systems. An initial and essential step towards enhancing low flow forecasts is to understand the behaviour and the dynamics of the rainfall-runoff models in simulation mode. Therefore, the aim of this study is to develop a tailored modelling framework to simulate low flows in the French Moselle catchment (10 770 km²), characterized by a warming climate (+0,5°C/decade) and decreasing low flows (-20 %). We propose to apply the daily PRESAGES conceptual rainfall-runoff (RR) model to 40 watersheds located in the study area for the reference period 2001-2020. To compare the different approaches, we use both a numerical metric and a visual inspection of observed and simulated hydrographs over the evaluation periods. Our research focuses on three main modelling issues: – Which input data ? In order to select optimal input data for the RR model, we compared a new daily gridded precipitation dataset based on rainfall observations vs the Météo France SAFRAN reanalysis. For potential evapotranspiration, we used alternatively the thermal-based Oudin formula and the Penman-Monteith formula. The outcomes indicate that the model performs optimally when utilizing the new daily gridded precipitation dataset in combination with the Penman-Monteith formula. – Which optimization algorithm for model calibration ? Three optimization algorithms were tested: Fmincon and Genetics Algorithms (i.e. intrinsic functions implemented in Matlab) and Shuffled Complex Evolution-University of Arizona (SCE-UA). – Which objective function for model calibration ? Visual hydrograph inspection show that high metric values can hide bad low flow simulations. Therefore, in order to identify the most suitable objective function, five functions derived from the widely used metric Kling-Gupta Efficiency (KGE) were tested : single objectives with discharge transformations and multi-objectives combining single objectives. Optimal modelling performance was obtained using Fmincon and a single KGE objective function applied to discharges at power 0.2. As a final step, this modelling framework has been used to parameterize a new semi-distributed version of the RR model in the study area.

Perrin, C. and the CIPRHES Team, 2022. Integrated chain for the hydrometeorological forecasting of low flows and droughts in France. The CIPRHES project. IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-665.

ABSTRACT. Small reservoirs have proliferated in the latest decades in many watersheds, often drawn by more frequent and severe droughts. Reservoirs decrease streamflows and may threaten river ecosystems, while the sustainability of their uses in a changing climate is questioned. At the catchment scale, several models of varying complexity are available to quantify the impact of numerous reservoirs and the efficiency of management rules (e.g., pumping restrictions, instream flows) or incentives (e.g., alternative crops), both now and for different scenarios of global change. Our objective is to develop an open source model which catchment managers could use for long-term planning and daily management. The model is the conceptual semi-distributed model airGRiwrm, a variant of the acknowledged rain-flow model GR. This work investigates how to include in the model small reservoirs used for irrigation, fish farming and leisure, to produce information for decision making. The study site, the Sèvre Nantaise watershed in Western France (2350 km²), has an oceanic climate and no large aquifers. We complemented the existing database of the watershed reservoirs, configured the model with a detailed description, especially for the irrigation demand for each farm dam, and successfully calibrated it over observed 2008-2020 flows on 13 subcatchments. We then assessed the impact on the calibration results for different hypotheses for linking reservoir surface, filling and volume, for aggregating reservoirs, and for the terms of their water balance, especially irrigation demand and evaporation. The results indicate the adequate accuracy for reproducing low-flows, a prerequisite for simulating management or climate scenarios.

ABSTRACT. Catchments convert precipitation into water flows for rivers and other surface water bodies. Transforming rainfall to runoff is nonlinear and complex. Initially, the use of deep neural networks in runoff modeling was limited, but this changed with the successful use of long short-term memory (LSTM) networks in runoff preditiction. This research delves into LSTM-based runoff models, examining them concerning hydrological concepts like catchment timing and regionalization, specifically within the French context. Both time-variant data and time-invariant attributes, spanning 376 French catchments, are utilized. A regime classification based on three hydroclimatic variables, derived from the analysis of interannual monthly regimes of runoff, total precipitation, and temperature, is used and validated. Three LSTM training levels are conducted resulting in three primary model types: SINGLEs (each catchment gets its LSTM model), REGIMEs (an LSTM model trained per hydrologic regime using data from respective catchments), and NATIONAL (trained using data from all catchments at the national level). The study finds that: a. model performance often improves with longer sequence lengths up to a certain point, after which it plateaus or declines, varying by hydrologic regime; b. the Uniform and Nival regimes demonstrate the most sensitivity to LSTM sequence length, indicative of their long-term hydrological dynamics; c. regional LSTMs don't always offer superior performance, and in some cases, local LSTMs can achieve minimal error with adequate high-quality data; d.training at both the regime and national levels produces similar results, but the HYBRID NATIONAL LSTMs, which combine national training with localized hyperparameter tuning, yield the most optimal outcomes.

ABSTRACT. Over the past 30 years, the frequency of natural disasters related to intense rainfall events has increased due to climate change. In this context, hydrodynamic rainfall-runoff modeling is of paramount importance to understand and predict the consequences of climate change. It is therefore essential to make advances in this field, particularly by improving runoff modeling through numerical solution of river hydrodynamics equations, as done by the Telemac 2D model. In this paper, we present a method for integrating spatialized rainfall into the calculation of net rainfall, using the SCS-CN model integrated in Telemac 2D. We have added a module that allows us to take into account spatialized rainfall from radar rainfall products, such as the Antilope products from Météo France, satellite data or even rain gauge data if we have a dense network of rain gauges in the area to be modeled. This module assigns to each node of the Telemac 2D mesh the value of the nearest rain pixel. We have applied this method to calculations on the catchment area of the Real Collobrier in the Var, with a surface area of about 92 km². In an urban environment, most of the watercourses are underground or are cut by crossing structures in many places. To integrate these structures such as culverts into Telemac 2D, we propose two solutions to overcome the problems of numerical noise at their outlet. The first solution is to divide the large structures into several smaller structures, so that the flow arrives at several nodes and not just one. The second solution is to create another module where the structure laws are calculated upstream with Mascaret (1D) and integrated into Telemac 2D. The output variables of the structure are then given in a formatted input file. Those We compared the calculations with spatially homogeneous rainfall to those with spatially variable rainfall, and highlighted the differences, which can be the shape of the flood hydrograph, the time and the value of the peak which strongly depends on the spatial location of the rain epicenters. We also showed the impact of our new method of taking culverts into account. In conclusion, our study has allowed us to develop methods to improve rainfall-hydrodynamic modeling in natural and urban environments. These advances contribute to a better understanding and prediction of the consequences of intense rainfall events.

ABSTRACT. The aim of this study is to compare innovative Polygon Trend Analysis (IPTA) and Innovative Trend Analysis (ITA) with the Significance Test and Mann-Kendall (MK) methods. To achieve this, the monthly total rainfall trends of the stations in the Vu Gia-Thu Bon River Basin of Vietnam have been examined for the period 1979–2016. The analyses show that rainfall tends to increase (decrease) in March (June) at nearly all stations. IPTA and ITA with the Significance Test are more sensitive than MK in determining the trends. While trends were detected in approximately 90% of all months in IPTA and ITA with the Significance Test, this rate was only 23% in the MK test

ABSTRACT. The classical estimation of future floods is using statistics; however, the uncertainty of climate change may lead to underestimation of potential extreme events. Therefore, the development and implementation of risk management plans considering climate change, is an urgent issue that must be addressed by government authorities and supported by the scientific community. For this reason, the present study pro poses a methodology to analyse fluvial flood sensitivity local level using the concept of "downward counterfactuals", it basically consists in tracking potential catastrophic floods using a previous extreme event registered in the a rea and assessing he consequences in terms of economic damages, severely affected people and deaths. The application is demonstrated by evaluating Onyar river flood produced by Glories storm in 2020 in an urban a rea of Girona - Spain. The hydraulic model was built using IBER 3.1 and a set of 49 counterfactual scenarios were defined considering changes in three main parameters of the root hydrograph: peak discharge, number of peaks, and duration; additionally, the risk assessment was performed following the criteria of Catalan Water Agency ACA through GIS tools The results showed that the magnitude of the peak discharge is the main parameter that conditionate the effect of the flood in Girona, while the number of peaks or the durations do not produce significant changes. Finally, six equations that relate the effects of the flood against the peak discharge and the affected area are presented and can be used by decision makers to get metrics of the impacts of a flood produced by Onyar river.

ABSTRACT. Mass movements and delta collapses are significant sources of tsunamis in lacustrine environments, impacting human societies enormously. Palaeotsunamis play an essential role in understanding historical events and their consequences along with their return periods. Here, we focus on a palaeo event that occurred during the Younger Dryas to Early Holocene climatic transition, ca., 12,000 years ago in the Lake Aiguebelette (NW Alps, France). Based on high-resolution seismic and bathymetric surveys and sedimentological, geochemical, and magnetic analyses, a seismically induced large mass transport deposit with an initial volume of 767172 m3 was identified, dated and mapped. To investigate whether this underwater mass transport produced a palaeotsunami in the Lake Aiguebelette, this research combines sedimentary records and numerical models. Numerical simulations of tsunamis are performed using a visco-plastic landslide model for tsunami source generation and two-dimensional depth-averaged nonlinear shallow water equations for tsunami wave propagation and inundation modelling. Our simulations conclude that this sublacustrine landslide produced a tsunami wave with a maximum amplitude of approximately 2 m and run-up heights of up to 3.6 m. The modelled sediment thickness resulting from this mass transport corroborates well with the event deposits mapped in the lake. Based on our results, we suggest that this sublacustrine mass transport generated a significant tsunami wave that has not been reported previously to the best of our knowledge.

ABSTRACT. Scour is a form of erosion induced by the influence of fluid flow on a collection of solid particles. The presence of a structure, such as an offshore wind turbine, a bridge pier, or a sluice gate, can locally alter the hydrodynamics of the flow and lead to additional erosion, referred to as local scour. This phenomenon exposes the foundations of the structure, jeopardizing its stability. The accurate prediction of local scour patterns is a critical concern for designing effective scour protection measures. Computational Fluid Dynamics (CFD) simulations hold the potential to offer greater precision compared to the semi-empirical laws commonly employed in offshore and river engineering, especially in complex scenarios.

In this study, we employ the neptune_cfd code, based on the multi-Eulerian approach where particles are treated as a continuous phase, to simulate scour resulting from the flow generated by a rotating wheel. Within a water tank, a wheel rotates counterclockwise above an initially flat sediment bed. This motion sets the surrounding fluid in motion, causing it to transport sediments away from the wheel and create a depression beneath it.

We compare our simulation results with experimental data from Ndoye et al. (2016), obtained through an acoustic method at various time points. These experiments encompass different combinations of wheel rotation velocity and wheel-to-bed distance. Initially, we validate the rotational motion of the wheel by implementing i) a rotating mesh with non-conforming junctions and ii) a tangential velocity in the boundary conditions of the wheel. Subsequently, we successfully reproduce the overall shape of the scour profile and observe an increase in scour depth when i) the wheel's rotation speed is augmented and ii) the wheel-to-bed distance is reduced.

ABSTRACT. Flow-Accelerated Corrosion, FAC, is one of the most prevalent form of degradation of carbon and low alloy steel pipes exposed to water flow, leading to significant wall thinning, in particular in the primary and secondary coolant systems in nuclear power plants. It is of critical importance to be able to predict regions of local FAC enhancement to avoid any future failure of the pipe systems. CFD alongside experimentations is a relevant tool to determine those locations of a severe attack. The objective of this study is to compare different numerical approaches to estimate the locations and the rates of FAC using CFD. Many theoretical and empirical models have been developed to allow predictions of the FAC rate. The simplest and commonly used prediction of FAC relies on the postulate that the mass transfer is the rate controlling step of the overall corrosion process and that only the hydrodynamic quantities influence locally the rate of degradation. The wall thinning rate in this model is proportional to a mass transfer coefficient and a concentration driving force between the wall and the bulk fluid.1 With this approach, the prediction of both FAC locations and intensity especially relies on the accuracy of calculation of the mass transfer coefficient distribution. However, some authors have already pointed out discrepancies between experimental results of wall thinning and numerical simulations, peculiarly after separation and reattachment positions of the flow in pipe singularities. This is particularly the case of elbows, a geometry investigated in this study.2 Because variations in the local chemistry, and consequently the oxide layer, may have been mistakenly neglected, a model based on a coupling of fluid dynamics, water chemistry as well as local electrochemical reactions is considered. Conclusions will seek to determine which approach is the more suitable and accurate to characterize rate and locations of FAC in different configurations.

1. Dooley B. & Lister D. Flow-Accelerated Corrosion in Steam Generating Plants. PowerPlant Chemistry, 20(4) (2018). 2. Xiadong Si et al. Investigation of corrosion behavior at elbow by array electrode and computational fluid dynamics simulation. Materials and Corrosion, (2020).

ABSTRACT. Air entrainment is a widely occurring phenomenon in air-water flows, it is crucial in many different natural or industrial flows such as rivers, water jets or spillways and breaking waves, thus underlining the importance of being able to understand and model these kinds of events. It can be caused by an important shear (water jets), by high fluid turbulence near the free surface or by the geometric encapsulation of air below the free-surface (breaking waves) and it is characterised by a wide range of bubbles or air pockets being created near the free-surface and then transported below it. The phenomenon can be described with two quantities: the amount of air entrained and its diameter probability density function. As direct numerical simulations (DNS) of this phenomenon, even if very appealing, remain an unachievable feat mainly because of the large number of bubbles and wide range of possible diameters, modelling this process is still a necessity. In the literature, air entrainment models exist but mostly predict the amount of air entrained during the simulation. Very little studies focus on the diameter created by this process. Unfortunately, in most simulation frameworks, the bubble diameter is a crucial physical quantity since it determines the interfacial momentum transfer between the bubbles and the surrounding fluid. It also possibly drives most of the large scale behaviour for a cloud of bubbles. In this paper, we present a new air entrainment model which predicts the bubble diameter for entrained air based on Hinze theory. In this theory, after bubbles are created at the free-surface, they are successively broken down into smaller and smaller structures due to turbulent breakup. When turbulence is no longer intense enough, the bubble diameter stabilises. By assuming this turbulent breakup cascade occurs close to the free surface and rapidly compared with other flow characteristics, it is possible to determine the diameter of stable bubbles when they are entrained below the free-surface by turbulent breakup. This new model is implemented in a multifluid RANS solver with an interfacial area transport equation to account for bubble diameter polydispersion. The diameter calculated from Hinze theory determines how the interfacial area transport equation must be adapted to account for air entrainment in the simulations.

Air entrainment is first generally described before introducing the model which is developed in neptune_cfd, a finite volume RANS solver developed by EDF, CEA, IRSN and Framatome which allows for the numerical resolution of separate Reynolds averaged Eulerian equations (mass, momentum and energy) for N phases coupled by interfacial transfer terms. Results obtained are then compared with experimental data in several cases representative of air entrainment phenomena. A special focus is made on mesh convergence and on the model relation with the mesh.

ABSTRACT. Over the past two decades, numerical modeling of boiling flows in microchannels has gained significant importance across diverse domains encompassing electronics and aerospace devices, where it plays a pivotal role in fostering the creation of compact cooling systems. Furthermore, this research is of critical importance in optimizing the drying process of porous media, especially within the field of nuclear industry applications, particularly in the context of addressing failed fuel rods.

Boiling flows in microchannels exhibit marked distinctions from their counterparts in conventional channels. Typically, the bubble diameters in microchannels exceed or closely match the channel dimensions, thereby introducing notable confinement effects as bubbles expand against the channel walls. Therefore, surface tension and wettability effects assume a pronounced role in this scenario and need to be considered. Finally, microchannels typically witness limited occurrences of boiling at their channel walls, with nucleate boiling being less prevalent, as the primary boiling takes place at the interface between bubbles and the liquid.

Numerical simulations of boiling flows in microchannels were conducted using the industrial software neptune_cfd. Using the pre-existing Large Interface Model, these simulations employed an interface tracking approach that considers surface tension and wettability effects when modeling vapor bubbles. This paper introduces a novel analytical mass transfer model based on a cut-cell method. Notably, this new model exhibits rapid mesh convergence, with validation studies demonstrating its ability to reach convergence with cells three times larger than the previous model. Furthermore, an Adaptive Mesh Refinement (AMR) technique was employed to track the liquid-vapor interface, resulting in a substantial reduction in CPU costs ranging from 10 to 100 times, depending on the specific case.

The validation of this method encompassed a range of 3D scenarios, including the growth of a vapor bubble in an unbounded superheated liquid (the Scriven case) and the buoyant ascent of a bubble within a superheated liquid. Furthermore, this study delves into more intricate scenarios, specifically those characterized by confinement and wettability effects within realistic microchannels.