Abstract for presentation at Urban Drainage Modelling and Water Sensitive Urban Design 2006

This paper brings together techniques that have been developed for more effective flood risk management with the need to better account for the uncertainty in the modelling, and hence control, of sediments in combined sewers.
The presence of persistent sediment deposits in main sewers is a common problem in many combined sewer systems. Knowledge about the build-up and erosion rates of sewer sediment is necessary to facilitate better sediment management strategies. Unfortunately sewer sediment phenomena are not well understood and as a consequence current transport modelling techniques often provide predictions which are significantly different from observation. This has resulted in a lack of confidence in sediment prediction tools amongst sewer operators and therefore planned sediment management is rare (Ashley et al. 2004).
Approaches that incorporate the uncertainties in modelling inputs and outputs are increasingly gaining interest in numerous engineering disciplines, e.g. for flood defence (HR Wallingford, 2004). Rather than basing design decisions on conservative deterministic estimates, these approaches acknowledge the possibility of a range of outcomes. This type of information supports more rational decision-making in the design of systems to manage risks and helps to ensure more cost effective solutions.
The aim of this study is to model sediment transport capacity, and any consequent erosion or deposition, in a main sewer in London that suffers from recurrent sediment problems. A probabilistic modelling approach has been selected, that can be utilised in addition to an existing deterministic design method. Model outputs provide information such as: ‘given a sediment transport capacity model and these input variables, the probability of sediment transport rates falling below X g/s/m³/s is Y %’. This type of analysis allows insight into the sensitivity of the model output to changes in the different variables, the contributions of the uncertainties in those variables to the exceedance probabilities and the incorporation of model uncertainties. It is also essential to assess the likelihood of how effective any remedial measures may be and the certainty or otherwise that may be attached to any sizing of control structures.
The hydraulic parameters in the sewer are modelled in a deterministic way using a calibrated Infoworks model of the main sewers in London. The sediment transport capacity has been modelled using in-house software, based on the commonly used Ackers sediment transport equations. Techniques for estimating the probability of non-exceedance of the Ackers predicted sediment transport rate have been developed from a probabilistic approach used in coastal engineering (Buijs et al, 2005).
Initially, uncertainties in the Ackers equation are addressed. In current applications the coefficients in the equation are considered to be deterministic. In fact, they are derived from a series of flume and pipe tests. The original data and endeavours of researchers to improve the coefficients for different design situations imply the contrary. As an illustration, applying a standard deviation of 10% of the mean value to the coefficients results in the plot below. Compared to the deterministic outcome of the model, this indicates a different model behaviour for variations in the coefficients in the Ackers transport equation. Although the standard deviation should be based on the original data, the results provide a first indication for which further investigations of coefficient sensitivity should be focused.
Secondly, the impact of the uncertainty in the field data is investigated. The data currently collected at this location are: rainfall data since 2002, flow data since May 2004 and detailed sediment build-up data (particle size analysis and sediment depths) since March 2005. These data are being used for modelling sediment erosion and build-up, using a combination of deterministic and probabilistic modelling techniques.
This paper demonstrates how the uncertainties with respect to coefficients embedded within the Ackers equation can influence the results. In addition, the results show how the variability in field data impact on the sediment transport capability predictions. These insights allow research to be focussed on the most influential coefficients in the model and help target data collection efforts. Furthermore, the paper explores how these results can be utilised to improve upon current design approaches for combined sewer sediment management.
Ashley R. M., Bertrand-Krajewski, J-L., Hvitved-Jacobsen T., Verbanck M., (eds) (2004), Solids in Sewers, Characteristics, effects and control of sewer solids and associated pollutants, Scientific and Technical Report No. 14, IWA Publishing, ISBN 1 900222 91 4.
Buijs, F.A., Segura-Domínquez, S., Sayers, P.B., Simm, J.D., Hall, J.W., (2005), Tiered reliability-based methods for assessing the structural performance of coastal defences, Proc. Coastlines, structures and breakwaters 19-22 April
HR Wallingford (2004), Risk Assessment for Flood and Coastal Defence for Strategic Planning, R&D Technical Report W5b-030/TR, A Summary, Defra / Environment agency