HORIZON 2020: Pollutant Transport in Urban Floodwaters Fluid Mechanics and Acoustic Laboratory (LMFA-INSA) - €87,874.05

There is an increasing awareness that the risks associated with urban flooding extend to impacts on public health from exposure to contaminated flood water. Surcharge from combined or foul sewer systems during flood events leads to the presence of faecal contamination within flood water and contamination of urban surfaces that can persist for extended periods after floodwater has receded (ten Veldhuis et al., 2010). Han and He (2021) also studied the links between urban floods and Covid transmission (without definite conclusions). The ageing urban drainage infrastructure is unlikely to cope with increasing extreme rainfall intensity in the future (Willems et al., 2012). Understanding the risks posed by contaminated floodwaters is a challenging task as health impacts are affected by numerous variables including the composition of bacteria, concentration, exposure time and vulnerability (Fewtrell et al., 2011). Recent attempts have thus been made to include health risk assessment within urban flood models based on a prediction of wastewater concentration within surface floodwaters (Mark et al., 2018). Understanding the dispersion and mixing of pollutants in floodwaters is a critical aspect for modelling this risk and current modelling developments are hindered by a lack of robust calibration and validation datasets. Recently, various teams in Hannover (Germany, see Sämann et al., 2019), Zaragoza (Spain, see Fernández-Pato and García-Navarro, 2021) or Dalian (China, see Fang et al., 2022) computed pollutant dispersions in flooded cities without any calibration data from field or laboratory measurements, so that turbulent diffusion coefficients had to be guessed or taken from more classical, but not well adapted, 1D flow experiments (see recent review by Huai et al., 2018). Within other applications, diffusion coefficients are known to be variable, being dependent on both the local turbulence as well as dispersion processes arising from the depth and/or width averaged treatment of the flow (Gualtieri et al., 2017). Consequently, calibrations using existing experimental studies of river or piped systems with significantly different aspect ratios and hydraulic characteristics are highly doubtful. The modelling community therefore requires robust datasets describing the spread and mixing of pollutant concentrations in shallow floodwaters typically encountered in the street network during urban floods, and a better understanding of the key physical processes associated with urban flood pollutant mixing/spreading such that appropriate modelling representations can be determined.
In recent years, a strong effort started to arise from the research community to establish operational numerical models able to simulate pollutant dispersions during urban floods, either from sewage overflows, accidents such as truck transporting dangerous materials, or floods invading industrial sites. Two types of numerical models are currently being tested. Both are based on shallow water equation models where the flow pattern is first computed over the whole domain over time, and dispersion of pollution is then computed with (i) Eulerian 2D models solving the advection-diffusion or with (ii) Lagrangian models based on random walk. Both models still require 2 turbulent diffusion coefficients, or associated turbulent Schmidt numbers, that are not known a priori (see in Fig. 1 the variability of turbulent Schmidt numbers used by the community of 3D RANS calculations, as collected by Gualtieri et al., 2017). Using the same hypothetical diffusion coefficients, Sämann et al (2019) recently compared both approaches and proved a fair agreement (see Fig 1). All these authors concur that there is a strong need for 2D experimental benchmarks in flow configurations typically encountered in urban floods - with the aim to identify the best-fitted streamwise and transverse turbulent diffusion coefficients to be applied when computing the flood flow in straight streets, in crossroads, skirting obstacles (trees, parked cars, parkings…), entering buildings, overflowing from the sewer network, etc. The project therefore aims to establish reference benchmark tests based on large scale experimental measurements using the above/below ground facility at UoS and develop an improved understanding of the underlying physical relationships governing the transport of pollutants in shallow floods environments. This will lead to increased confidence in the derivation of suitable model parameters in full scale systems. The proposal will simulate a range of possible urban flood scenarios, via physical experiments, that are associated with pollutant transmissions to establish unique datasets on the phenomena. The controlled environment in the lab will observe the spatiotemporal variations of key features in the process such as water depth, velocity, elevation, surface material and concentration of pollutants. The investigation will enhance the understanding of the influences of those parameters, which will be utilised in numerical modelling to strengthen the knowledge on model settings to better predict the dynamics of contamination concentrations in floodwaters. While measurements of the 2D fields of free-surface elevation (using ultrasonic probes or point gauge) and surface velocity (by LSPIV) are well-established methods, methodologies to measure the pollution dispersion in shallow flows are still under investigation by different research groups. Therefore, this proposal will utilise 3 different tracer methods: dye (for the PCA method), salt (for the conductivity method) and temperature (for the thermometry method). The PCA method is currently installed within the existing facility, with other instrumentation to be provided by visiting research groups. In first phase, testing will compare the 2D concentrations estimated by these 3 methods against benchmark probes, applied on a limited set of experimental configurations. Then, based on the results of this first phase, the project will test a number of further experimental configurations to establish reference experimental benchmarks to be used by numerical modellers to calibrate operational models. This proposal will bring together both the modelling (U. Hannover, ITWH, Liege Univ, Exeter, DHI and Innovyze) and the experimental (INRAE Lyon, LMFA Lyon, Sheffield) communities to generate new datasets associated with the spreading and mixing of pollutants in shallow floodwater, and develop new understanding of the sensitivity of these processes to physical characteristics such as roughness, flows around urban furniture, flows in street with various aspect ratios.