The disadvantageous consequences of stormwater perturbations of receiving water quality in urban environments can be attenuated by exercising control at various locations across the sewer network, wastewater treatment plant, and the stream itself. As part of a long-standing programme of research on developing an integrated approach to the management and real-time control of water quality in river basins, the paper examines the sensitivity of the associated strategies to model uncertainty. Specifically, results are presented for a case study based on a 10km stretch of the River Cam as it passes through the city of Cambridge in eastern England. The options for control are restricted to design and operational features of the wastewater treatment facility. Assessment is according to maximum and cumulative values of mass flows of ammonium-N and biochemical oxygen demand, together with the duration of dissolved oxygen concentration below 4.0 gm−3, at the downstream boundary of the system. A straightforward analysis of the sensitivity of these criteria to changes in the parameterisation of a model for receiving water quality shows that the ranking of strategies is robust in the face of model uncertainty. Minor differences in ranking occur as a function of whether judgement is based on ammonium-N or the other two attributes of water quality and whether attention is focused on the treatment plant in isolation or performance across the system as a whole. However, such conclusions must be qualified by noting that our analysis has been limited in its scope and elementary in its treatment of uncertainty.

The paper summarises recent progress in a long-term programme of research on an integrated approach to the management, md real-time control of water quality in river basins. The focus of this progrmnme is the development, md application of simulation models for the dynamic behaviour of wastewater treatment plants, and in-stream water quality. The model for the latter is based on a multiple continuously stirred tank reactor (MCSTR) approximation of fluid and solute propagation along a river system. Results are presented for the identification (calibration) of this model with reference to field observations from the River Cam in eastern England. These results illustrate the benefits of significant changes to the hydraulic basis of the model (relative to earlier applications). They also provide a good test of the model's capabilities in respect of solute transport, md the biochemical interactions among the five state variables of water quality, i.e., biochemical oxygen demand, dissolved oxygen, ammonium-N, nitrate- N, and chlorophyll-a. The model is applied to the assessment of management and real-time control strategies for attenuating the adverse effects on stream water quality of storm sewage surges passing from the sewer network and through the wastewater plant. The assessment includes the coordinated manipulation of in-stream hydraulic structures to improve controlled performance.

The impacts of combined sewage discharges on river water quality are studied using the MCSTR (Multiple Continuously Stirred Tank Reactor) dynamic model. The potential for applying this model in a real-time context is demonstrated as a tool to support decisions regarding treatment plant operating during storm events, when it is often not possible to sustain full treatment of the incoming sewage flow. Discharges to the River Cam of treated and untreated urban wastewaters from Cambridge and the Cambridge Sewage Works are addressed as a hypothetical case study. Alternative treatment strategies are defined for improving receiving water quality and assessed through simulated water quality downstream of the discharge; the state variables of the model include the concentrations of biochemical oxygen demand, ammoniacal- and nitrate-nitrogen, and dissolved oxygen (chlorophyll-a concentrations are also calculated, but not considered herein). Strategies are assessed and ranked according to the reduction in maximum pollutant concentration (or the increase in minimum concentration, in the case of dissolved oxygen) promoted by each alternative, relative to conventional operation. The consequences of discharging overflows at an alternative position in the river, rather than together with the treatment-plant effluent, are also evaluated. Run times fr the MCSTR model are of the order of just a few minutes (at most), thus allowing the potential for its use in real time as a decision-support aid.

Monte Carlo simulations taking uncertainty in model parameters into account were performed on a river water quality model. The simulation results were used to rank wastewater treatment plant control strategies according to their impacts on river water quality. This impact is estimated by the maximum ammonium concentration and by the duration of dissolved oxygen concentration below 4 g/m3 at the downstream boundary of the system. The strategies were classified according to the previous criteria using 4 ranking methods, one of them being based on the concept of stochastic dominance. Results are presented for a case study based on a 10 km stretch of the River Cam as it passes through the city of Cambridge in Eastern England. It was found that ranking was robust in face of uncertainty in the parameter values for the control strategies considered as being superior in terms of river water quality impacts.

Wageningen Centre for Development Innovation supports value creation by strengthening capacities for sustainable development. As the international expertise and capacity building institute of Wageningen University & Research we bring knowledge into action, with the aim to explore the potential of nature to improve the quality of life. With approximately 30 locations, 6,500 members (5,500 fte) of staff and 12,500 students, Wageningen University & Research is a world leader in its domain. An integral way of working, and cooperation between the exact sciences and the technological and social disciplines are key to its approach.