Framework for the validation of micropollutant removal in passive stormwater treatment systems

Stormwater is a valuable urban water resource that can be used to provide multiple benefits to the community: reduced potable water demand, improved stream health and amenity of urban landscapes. Meanwhile, stormwater biofilters (bio-retention or rain-gardens) and wetlands, are gaining popularity for stormwater harvesting. However, water quality treatment validation frameworks for these systems as yet do not exist, limiting their adoption for safe stormwater harvesting for high exposure human uses. In order to advance the acceptance of these systems, a three-stage validation framework was developed in this study for the validation of stormwater treatment systems for micropollutant removal: (1) pre-validation, (2) validation monitoring and (3) operational monitoring. In Stage 1: Pre-Validation, consists of five steps for gathering the necessary information for the next two stages: (i) identification of target micropollutants, (ii) specification of treatment targets, (iii) identification of potential removal mechanisms and influential factors, (iv) identification of potential surrogates, and (v) determination of operational and challenge conditions. Firstly, the potential target micropollutants were identified through a systematic literature review of stormwater quality, which showed that challenge concentrations of 8 pesticides, benzene, benzo(a)pyrene, pentachlorophenol (PCP), di-(2-ethylhexyl)-phthalate (DEHP) and total polychlorinated biphenyls (PCBs) were above different worldwide drinking water guideline limits, and hence the treatment targets (as percentage) could be set according to the guideline value. Potential removal mechanisms (e.g. adsorption and biodegradation) for different micropollutants, as well as potential surrogates for operational monitoring (e.g. delta-UV and delta-DOC), were also identified. MUSIC V5.1 was utilized to simulate the behavior of a model systems based on 30-year rainfall data in three distinct climate zones; outputs regarding the time and amount of flows in to/out of the systems were evaluated to identify the threshold of operational variables, including length of dry periods (LDPs) and volume of water treated per event. Stage 2: Validation Monitoring determines the system performance under challenge conditions in order to demonstrate that it can meet the treatment target from Stage 1. A series of laboratory, field experiments and modelling were conducted. Two small scale field biofilters (loamy sand without submerged zone, LS-noSZ; sand with submerged zone, S-SZ) were challenged for the removal of a range of micropollutants. Both of the tested biofilters had a removal efficiency of >80% for total petroleum hydrocarbons (TPHs), glyphosate, dibutyl phthalate (DBP), bis-(2-Ethylhexyl) phthalate (DEHP), pyrene and naphthalene under the range of operational challenge conditions, i.e. daily temperautre (5.0-33.0 oC), soil temperature (10.6-22.3 oC), volume of water treated per event (~3PVs for consecutive events and ~4PVs for single events), challenge wet conditions (LDPs-10 h) and challenge dry conditions (LDPs-496 h). The removal of pentachlorophenol (PCP) and phenol loads was >80% in LS-noSZ and 50%-80% in S-SZ, while chloroform had load removal rates between 20% and 50%. All of the above micropollutants (except PCPs) met the treatment targets as per Australian Drinking Water Guideline values under all conditions. PCP did not meet the target in S-SZ during the challenge dry and wet conditions. Biofilters were less effective in removing atrazine and simazine with load removal 20-50% in LS-noSZ and <20% in S-SZ. The removal of atrazine and simazine did not meet the treatment target under most operational conditions. A new validation modelling tool consisting of a process-based model and parameter estimation was developed. The tool was tested and was successful in reproducing the flow of well-designed biofilters (i.e. S-SZ) with a Nash-Sutcliffe coefficient of E=0.96. Fate parameters (adsorption and biodegradation) acquired from the laboratory batch experiments were used with the model to provide predictions of fluorescein behavior with E=0.67. The peak outflow concentrations were particularly well modelled, with the differences between the modelled and measured peak values being –3.9% to +7.4% for spiking tests and -4.4% to 28% for flushing/rinsing tests. Given that the peak outflow concentrations were sometimes slightly underestimated, if the validation modelling tool is to be used for validation, at least ±4.4% errors should be considered to ensure that the tool can produce conservative results. The removal efficiencies produced by the validation modelling tool indicated a conservative estimation. Lastly, a novel in-situ column (ISC) tool was developed for validation monitoring. The tool comprises of a stainless steel column that can be inserted into field biofilters in a non-destructive manner. The ISC was successful in reproducing the fluorescein field challenge tests (FCT) results through a series of continuous experiments that covered different operational conditions (Nash-Sutcliffe coefficient: 0.83-0.88). Specifically, good agreement was achieved for the outflow time lags, slopes and peaks, plateaus and the removal efficiencies between the ISC and FCT. ISC indicated conservative results in all tests except for the 1st spiking test, in which ISC had an overestimation of 5.3%. The later application of the tool indicated that ISC tests produced relatively lower outflow concentrations of herbicides (atrazine, simazine and prometryn) compared with the FCT, which was attributed to different climatic conditions, e.g. the ISC tests were conducted in winter while the FCTs were performed in summer. Therefore the ISC should be used only when the inherent boundary conditions of the test are carefully planned. Stage 3: Operational Monitoring ensures that the defined treatment targets are being continuously met during normal operation. A range of potential surrogates (total suspended solids (TSS), total phosphorus (TP), total nitrogen (TN), ammonia (NH3), nitrate (NOx), dissolved organic carbon (DOC) and UV absorbance at 254mn (UVA254)) were selected and tested in controlled field tests on two full scale biofilters (of different designs) that were exposed to both typical and challenge operational conditions for the removal of glyphosate, atrazine, simazine and prometryn. ∆TP was significantly and linearly related to glyphosate reduction (R2 = 0.75-0.98, P<0.01), while ∆TP and ∆UVA254 were linearly correlated (R2 = 0.44-0.84, P<0.05) to the reduction of triazines (atrazine, simazine and prometryn) in both field and laboratory tests. The linear regressions between ∆TP and glyphosate, and between ∆UVA254 and triazines, were further confirmed by laboratory tests. The performance of ∆TP and ∆UVA254 as surrogates for herbicides were reliable under normal and challenge dry conditions, but weaker correlations were observed under challenge wet conditions. In general, ∆TP is a potentially promising surrogate for glyphosate removal, while ∆UVA254 is a more suitable surrogate for triazines removal in stormwater biofilters. However, more work is needed since we had only one challenge test of this nature; e.g. the data set was rather limited and the poor correlation during challenge wet conditions could be due to very narrow range of removal rates observed in the field. In summary, this research presents the world first framework for validation of micropollutant removal in passive stormwater treatment systems. Further work is however needed to: (1) promote the adoption of the developed validation framework, (2) improve the current/develop new validation monitoring tools, (3) verify the promising surrogates by testing more sites and (4) develop operational monitoring schemes.