Modelling both the continual accumulation and erosion of discolouration material in drinking water distribution systems

Discolouration material has previously been shown to accumulate continually and ubiquitously as cohesive layers in all drinking water distribution network pipes. Discolouration risk cannot therefore be eliminated by one-off strategies such as pipe renewal or invasive cleaning. However maintenance programmes of regular controlled flow increases have potential for long-term risk management. To facilitate the design and optimisation of these programmes, a new discolouration model is proposed. This combines a novel sub-model of how the shear strength profile of material layers changes due to accumulation with a material erosion sub-model that has similar behaviour to the validated shear-stress-dependant PODDS erosion model. The accumulation sub-model reflects the observation that material with shear strengths exceeding the imposed shear stress accumulates at a rate invariant to this hydraulic force. These differing behaviours are facilitated by modelling how the amount of wall-bound material varies over time for distinct shear strength bands. This results in a model state that is more complex and powerful than PODDS but the new model has fewer parameters. The validity of the model was tested by calibrating model instances of a pipe rig experiment, four distribution main flushes and three long-term trunk main monitoring programmes using a verified software implementation of the model. The empirical model can be automatically calibrated using swarm optimisation (as has been demonstrated) but in the one case where its predictive power was tested it was found to be limited. However, the model's present value is as a framework for furthering the understanding of erosion and regeneration and, as more accurate and complete datasets become available, refinements could result in a tool for medium to long-term network management, including the design of flow conditioning programmes. The model is now implemented in commercial network modelling software.