This paper describes the development of a methodology to theoretically assess the stormwater pollutant removal performances of structural best management practices (BMPs). The method combines the categorisation of the relative importance of the primary removal processes within 15 different BMPs with an evaluation of the ability of each process to remove a pollutant in order to generate a value representing the pollutant removal potential for each BMP. The methodology is demonstrated by applying it separately to a set of general water quality indicators (total suspended solids, biochemical and chemical oxygen demand, nitrates, phosphates and faecal coliforms) to produce a ranked list of BMP pollutant removal efficiencies. Given the limited amount of available monitoring data relating to the differential pollutant removal capabilities of BMPs, the resulting prioritisation will support stakeholders in making urban drainage decisions from the perspective of pollutant removal. It can also provide inputs to existing urban hydrology models, which aim to predict the treatment performances of BMPs. The level of resilience of the proposed approach is tested using a sensitivity analysis and the limitations in terms of BMP design and application are discussed.
The EU Water Framework Directive (WFD) (EU, 2000) identifies the control of diffuse pollution as a key factor in enabling good ecological status to be achieved in aquatic systems and therefore represents a clear driver for the review of current stormwater management practices. The conventional drainage approach to managing stormwater flows involves the direct removal of surface water through a series of pipes to the nearest watercourse to prevent local flooding, with little attention being paid to stormwater quality or its impact on receiving waters. However, the issue of urban water quality is increasingly taking centre stage and associated with this is a greater interest in the use of stormwater best management practices (BMPs). These represent a diverse range of source control procedures, which integrate stormwater quality and quantity control as well as enabling social and amenity perspectives to be incorporated into stormwater management approaches. The term ‘BMPs’ covers a wide range of structural systems, all of which have the ability to improve stormwater quality and attenuate flow volumes through a combination of biological, physical and chemical processes.
1.1. Selection of BMPs
There are a variety of approaches and guidelines available for selecting the most appropriate type of BMP for a particular site with, for example, many states/counties in the USA having their own stormwater design manuals (e.g. Maryland Department of the Environment, 2000). Typically, these make recommendations in relation to catchment-specific factors such as soil type, available space, capacity to store a design storm event, operation and maintenance requirements and cost (CIRIA, 2000, 2001). In contrast, the potential for specific types of BMPs to remove a particular pollutant of concern, or even the treatment efficiency of BMPs in general, is rarely, if ever, used as a discriminatory criterion. However, as a result of the implementation of the EU WFD, and the increasing importance of pollution reduction accountability in the context of River Basin Management Plans, knowledge relating to the ability of different BMP systems to remove a particular pollutant is becoming a prime requirement.
A range of urban hydrology models are available which incorporate assessment of the performance of BMPs as part of the drainage network, although these simulations are primarily based on an evaluation of their hydraulic behaviour. An appropriate example is InfoWorks CS, which models the hydraulic behaviour of BMPs such as soakaways, infiltration trenches, swales and permeable pavements. Additionally, it contains a water quality module which can feedback into the hydraulic simulation, but this relates primarily to the modelling of physical processes, such as sediment build-up, to support pollution control through the identification of CSO problems rather than contributing to the prediction of pollutant removal treatment efficiencies in different BMPs (Wallingford Software, 2006). The Storm Water Management Model (SWMM) integrates BMP hydrological modelling with a consideration of the associated treatment performance (EPA, 2006). However, this requires operators to input their own BMP removal efficiencies, which in the absence of field data would need to be estimated.
A model which more specifically addresses the role of BMPs by incorporating ponds, bioretention systems, infiltration buffer strips, sedimentation basins, pollutant traps, wetlands and swales into stormwater management strategies is the decision-support model MUSIC, Model for Urban Stormwater Improvement Conceptualisation (CRC, 2006). MUSIC models BMP performances using algorithms originally developed for Continuously Stirred Tank Reactors (CSTRs) with different numbers of CSTRs being used to mimic different types of BMPs. Pollutant removal is evaluated using the kC* model modification of the first-order kinetic uptake model developed by Kadlec and Knight (1996) to predict the removal of BOD in wetlands. The selection of appropriate values for the first-order rate constant (k) and the background pollutant concentration (C*) are therefore of critical importance in robustly predicting pollutant removal. However, because the removal processes which occur in BMPs are highly heterogeneous in terms of both space and time (for example, physical processes may dominate during storm events with biological and chemical processes being of greater importance in the longer term), choosing a single ‘k’ value which covers all these variables is extremely complex. The selection of C* is also problematic as it may vary greatly in relation to factors such as variations in the timeperiod between storm events and rainfall intensity. A range of default ‘k’ and C* values, based on considerations of the relevant physical processes, are provided within the model for total suspended solids (TSS), total phosphorous and total nitrogen for seven different BMPs. Users are recommended to select from a range of values (e.g. ‘k’ values from 500 to 5000 m year1 for TSS removal within a wetland) based on factors relating to the specific BMP characteristics and a sensitivity analysis to determine the impact of varying ‘k’ and ‘C*’ values on overall treatment performance. Such a pragmatic approach is necessitated by the current lack of comparable data on pollutant removals across the many different types of BMPs and indicates the need for a fuller investigation into the relative contributions of the pertinent biological, chemical and physical processes.
As a contribution to meeting these identified needs, this paper describes the development of a systematic approach, based on fundamental scientific principles, for the prediction of the comparative pollutant removal potentials operating within a range of BMPs. The results can be utilised to inform the selection of k values which integrate biological, chemical and physical processes within modelling routines, such as MUSIC, as well as contributing to the wider decision-making framework which typically involves the consideration of a diversity of factors ranging from catchment size to rainfall-runoff coefficients and costings. Hence, the approach described in this paper is not meant to be used as a stand-alone procedure, but one which can contribute to stormwater management decisionmaking processes with respect to the control of the discharge of specific substances to receiving water bodies until further field data becomes available. Full details on the development of this novel approach are presented, together with its application to TSS, BOD, COD, nitrates, phosphates and faecal coliforms. Results of this procedure are critically discussed and compared, where possible, to observed field data and the validity tested through a sensitivity analysis.
2. Unit removal processes in BMPs
2.1. Identification and controlling factors
The performances of individual stormwater BMPs may vary from site to site in relation to variables such as design specifications, local hydrologic and climatic conditions, and system age (e.g. Ellis et al., 2003). In developing the described methodology, the BMP-type descriptions are as outlined in Table 1. In addition, the individual BMP devices are assumed to be operating at their design efficiency and to be functioning on a ‘stand-alone’ basis, i.e. not part of a hybrid treatment train. The vulnerability of the treatment potential to variable hydraulic conditions has been taken into account by incorporating the impact of extreme events into the predicted efficiencies of the identified removal processes within different BMPs. The possible impacts of variables, such as system ageing, are partially incorporated through the inclusion of how the different removal processes are likely to be maintained throughout the lifetime of a BMP.
The primary biological, chemical and physical processes associated with pollutant removal in structural BMPs are identified in Fig. 1 and reflect the fundamental relative potential for each BMP to remove the pollutant under consideration. These combined values can then be ranked in a descending order to generate a hierarchy of BMPs with regard to removal of the specific pollutant of concern, which can also be used to support the relative selection of ‘k’ values for use in stormwater management modelling.unit operating processes (UOPs) familiar to traditional water and wastewater engineers. This UOP approach provides an alternative, and more radical, methodological basis for the selection of BMPs, but one which is being more widely considered within stormwater engineering (Quigley et al., 2005; Scholes et al., 2005). Fig. 1 shows how these fundamental UOP behavioural properties can be integrated with pollutant-specific characteristics to develop a combined value, which represents the potential for a specific pollutant to be removed within a particular BMP system. Repetition of this procedure for each BMP enables a combined value to be developed which represents the relative potential for each BMP to remove the pollutant under consideration. These combined values can then be ranked in a descending order to generate a hierarchy of BMPs with regard to removal of the specific pollutant of concern, which can also be used to support the relative selection of ‘k’ values for use in stormwater management modelling.