Options
2014
Journal Article
Titel
Model based investigation of transport phenomena in water distribution networks for contamination scenarios
Abstract
Water distribution Networks (WDNs) are critical infrastructures that are exposed to deliberate or accidental contamination. In the immediate future water suppliers will install water quantity and water quality sensors throughout their networks that will provide a continuous and huge stream of data. This allows for the first time to develop a system that is able to monitor and protect WDNs in real-time. Obviously, in case of a toxic contamination, the WDN operators have to react very fast as humans have to be protected against toxic drinking water. Hence the operators are highly interested in getting an answer to these questions very quickly: Where is the contamination source? What impact will the contamination have on the water distribution network? Which actions are necessary (e.g. which main pipes have to be shut)? The SMaRT-OnlineWDN management toolkit [1] gives a high quality decision support regarding these questions. The online security management toolkit is based on sensor measurements (hydraulic and water quality sensors) and online hydraulic and transport models and will allow (i) an estimation of the localisation of the contamination source and (ii) the simulation of short-time future scenarios in order to estimate the impact of a contamination source. For both tasks the transport model is one of the central modules as the accuracy of predictions (or re-simulations) for the distribution of solute substances throughout the WDN depends on the accuracy of the transport model. Hence a main objective of research is to enhance the transport model for the use in online simulation. While the general phenomena of transport modelling are well understood (mathematically described by a partial differential equation which models spatio-temporal behaviour of diffusion, advection, reactions and sources resp. sinks). However, all actually available software packages for the simulation of hydraulics and water quality in WDN (e.g. EPANET [2]) are based on a one dimensional (1D) formulation. Due to that restriction in many cases where the 3D geometry of the WDN has an impact to spreading of substances (e.g. at junctions and crossings of pipes) the 1D description may not be accurate enough. Hence in recent years some research has been done in order to improve the performance of 1D transport models. E.g. in [3-5] a more realistic model of the axial dispersion in 1D-models has been introduced, as so far axial propagation for the most part is modelled by plug flow. Though this is a good approximation for turbulent flow it does not hold up for laminar and transitional flow. Furthermore, improvements of the modelling of the mixing behaviour in 1D models has been achieved. Classical mixing models used in most simulation software tools (e.g. EPANET) are based on the assumption of complete mixing of the substances. Experimental investigations have shown that for certain flow conditions at cross-junctions and double-T-junctions ideal mixing is not fulfilled [6]. Hence some more enhanced approaches have been investigated in order to find a more realistic model to describe an incomplete mixing of substances in 1D transport and flow models (e.g. estimations based on experimental results in AZRED [6], bulk-mixing model implemented in EPANET 2 BAM [7]). But in practice it is often difficult to apply these models as they contain one or more parameters which have to be determined by several experiments for each individual junction. Our approach is to find a more realistic model of transport processes at T- and cross junctions which can be applied more easily in practical applications. The mixing model depends on the flow conditions (inflow, outflow), the pipe and crossing geometry and wall roughness. Preliminary results will be presented using computer fluid dynamic (CFD) for 2D and 3D modelling of transport processes at T- and cross junctions in WDN for different flow conditions (laminar to turbulent flow). These results are evaluated in comparison to existing mixing and propagation models.