Experimental and computational study of velocity fields in narrow or compound section sewers
The urban managers have to manage the sewer networks and drainage systems in an optimal way, to avoid any problem of operation which could harm the quality of water surface and to have significant economic repercussions. The installation of the permanent instruments of measurement in the sewer networks not only allows ensuring a control of the operation of these networks but also allows optimising their management.
Sound management of the sewer networks and minimization of the pollution discharged into rivers through combined sewer overflows necessitate more in-depth knowledge of the flow rates and pollutant loads conveyed in sewers. The precise knowledge of the flux needs a better assessment of the representativeness of the sensor measured velocities versus the physical mean velocity in a cross section. Various European research teams are working on such topics. For sampling velocity fields, Hughes et al., (1996) have proposed an array of acoustic Doppler sensors but it makes only a limited number of measurements in a cross section and the location of these points can not be easily selected and they do not present any velocity field. Hrabak et al., (1998) have used the CFD tool to calibrate flowmeters. The current research project aims to improve the representativeness of the measurements made in the man entry sewers. The methodology proposes firstly to help selecting the measurements area and then to improve the data treatment of the measures made in that area, (Larrarte et al., 2004).
Numerical models open wide potentiality to represent the evolution of flow rates in a sewer network. Unfortunately the one dimensional computations that are now quite common in urban hydraulics are unable to represent all the complexity of the hydraulics phenomena occurring in narrow (width/height of water smaller than 5) or compound channel flows such as the dip phenomenon and secondary currents. Thus a three dimensional Navier Stokes solver is used in this project. Stovin et al., (2002) showed that isotropic turbulence models are unable to predict correctly the secondary current in trapezoidal channels. Very few data are available for egg shaped or circular channels, the current research project combines experimental investigation in real man entry sewer and numerical modelling. The experimental part gives the measurements that are necessary to validate the numerical part.
Figure 1a and 1b show respectively the experimental and numerical velocity distribution obtained by the solver with a volume of fluid method associated with the RSM non-isotropic model. It can be noticed that the numerical simulation is able to predict the velocity gradient close to the walls and the dip phenomenon centered at the mid depth of the cross section but underestimates the velocity close to the free surface and to the bench whereas it overestimate the maximal velocity area. These results are related to a numerical diffusion phenomenon at the free surface and to the mesh that need to be improved.
This paper : (1) presents the experimental velocity fields measured in real sewers whatever the water level is, (2) validates the application of three-dimensional numerical model to a narrow or compound channel, (3) investigate the effect of grid generation and turbulence modelling on the numerical results accuracy, (4) presents the limits of the k-e isotropic model and the potentialities of the RSM non-isotropic models on modelling the secondary currents and dip phenomenon in narrow or compound channel.
(a) experimental (b) numerical
Figure 1: Velocity distribution
References:
Hrabak D., Pryl K., Krejcik J., Richardson J., (1998), Calibration of flowmeters using FLOW-3D software, Novatech, 3rd International Conference on Innovative Technologies in Urban Storm Drainage, Lyon, pp 139-144.
Hughes A.W., Longair I. M., Ashley R. M., Kirby K., (1996), Using an array of ultrasonic velocity transducers to improve the accuracy of large sewer mean velocity measurements, Water Sci. Tech., Vol. 33, No. 1, pp1-12.
Larrarte F., Bonakdari H., Joannis C., (2004), How can CFD help for designing the implantation of sensors in sewer, 4rd International Confrence on Sewer Process and Networks, Funchal, Portugal.
Stovin V. R., Grimm J. P., Buxton A. P., Tait S. J., (2002), Parametric studies on CFD models of sewerage structures, 9rd International Conference on Urban Drainage, Portland, Oregon, U.S.A.
