Modeling the quality of sewage during the leaking of stormwater surface runoff to the sanitary sewer system using SWMM: a case study

BMP

Best management particles

BOD

Biological oxygen demand

COD

Chemical oxygen demand

TSS

Total suspended solids

TDS

Total dissolved solids

TN

Total nitrogen

TKN

Total kjeldahl nitrogen

TP

Total phosphorus

EPA

Environmental Protection Agency

IDF

Intensity–duration–frequency

DEM

Digital elevation model

GAMSO

General Authority for Metrology and Seismic Observations

GIS

Geographic information system

M.a.s.l

Meter above sea level

NMSE

Normalized mean square error

R

Correlation coefficient

UPVC

Unplasticized polyvinyl chloride

SWMM

Storm water management model

USGS

U.S. geological survey

LID

Low impact development

NPS

Nonpoint sources

In urban areas, polluted stormwater is often discharged untreated to natural water systems, leading to quality depletion of receiving water bodies (Gasperi et al. 2010). As society progressed, agricultural lands, forests, and wetlands have transformed into municipal land uses. Hence, urbanization negatively impacts watershed hydrology, resulting in the degradation of stormwater quality (O'Driscoll et al. 2010). Pollutants are transported into water resources through stormwater runoff over land surfaces (Corbari et al. 2016). After a long dry season, contaminants accumulate on impervious surfaces before being washed away through heavy rain, causing surface runoff pollution, resulting in increased amounts of nonpoint source (NPS) contamination in the urban water systems (Ying & Sansalone 2010). Stormwater contains various NPS contaminants, sediments, nutrients, heavy metals, trace elements, and pathogens (Zhang et al. 2019). NPS pollutants are of multiple types and altered by several site-specific factors making them challenging to control due to complex uncertainties involved in their behavior (Abdulkareem et al. 2018).

On the other hand, activities associated with domestication, increasing impermeable surfaces (e.g., highways, parking lots, walkways, roofs, and compacted areas), and increased use of fertilizers and pesticides on municipal lawns and gardens all contribute to increased NPS pollution in urban cities (Obaid et al. 2015). On the other hand, gaps, holes, weak points, and manholes are causing the leaking of polluted stormwater into the sewer system. This increases the concentrations of total suspended solids (TSS) and biochemical oxygen demand (BOD), particularly in areas where unpaved roads are exposing the sewer systems uncovered (Alisawi et al. 2015).

Since runoff from urban areas plays a crucial role in water quality degradation in receiving environments, a reliable estimation of NPS loadings is critical for receiving water quality (Lee et al. 2010). Therefore, modeling pollutants loadings from stormwater surface runoff can effectively control NPS pollution and treatment (Geng et al. 2019). As a result, developing a simulation model of sewage pollution is crucial for the effective management of urban sewer systems, especially during heavy rain events (Li et al. 2015a). However, sewer pollution in cities is influenced by other factors such as rainfall, sewer leaking, and human activity on land (Huong & Pathirana 2013).

Apart from many models, the storm water management model (SWMM) of the U.S. Environmental Protection Agency is considered a potent tool. SWMM is a program that simulates surface runoff quantity and quality in urban drainage systems (Li et al. 2016). Since SWMM version 5.1.010, the model has included a quality model that accurately simulates an urban region's hydrologic output quality. The accumulation and transportation of stormwater contaminants from a specific catchment can be simulated using SWMM models. The quality model in the SWMM program represented the contaminants and their land-use properties (build-up and wash-off) per-square-meter basis. For instance, Tuomela et al. (2019) used SWMM to model concentrations of TSS, total phosphorus (TP), total nitrogen (TN), lead, copper, and zinc contributed by different land cover types and hydrological conditions of residential areas in southern Finland. In Gold Coast, Australia, Hossain et al. (2012) simulated the variation in the stormwater quality (TSS, TN, TP) from Saltwater Creek Catchment to the catchment outlet using SWMM. SWMM quality simulation also predicted a correlation between quality and quantity, whereas pollutants concentrations decreased with high load inflow and increased with low load inflow (Baek et al. 2020). In combined sewer systems, SWMM simulated total loads of SS, chemical oxygen demand (COD), and total kjeldahl nitrogen (TKN) after rainy weather in residential, commercial, and industrial catchments (Chow et al. 2012).

In addition to qualitative simulation, SWMM also assesses stormwater best management practices (BMPs) for pollutants reduction at source using low impact development (LID) strategies. The intensified waterlogging disasters demand urgent and essential improvements on the sustainable development and management of urban stormwater drainage systems (Li et al. 2015b). Applying LID in urban stormwater drainage systems, complex drainage flow processes, and related water quality issues have also aroused great interest and attention from researchers and practitioners (Li et al. 2019a). LID stands for localized small-scale treatment systems that minimize pollutants at runoff sources such as residential and roadways areas (Tuomela et al. 2019). Some of the newer types of LIDs that have been created to control urban stormwater at the source include vegetated swales, rain gardens (or bio-retention systems), permeable pavements, and green roofs (Qin & Li 2013). LID techniques can reduce the impact of imperviousness on the hydrology and water quality of urban stormwater runoff. LID strategies were created to imitate predevelopment climatic conditions and encourage storage, evaporation, and infiltration (Ahiablame et al. 2012).

In this case study, stormwater surface runoff is leaked into the sewer system due to gaps and holes in manholes, unpaved and completely exposed sewer lines, and illegal opening of sewer manholes drain floods. Therefore, the study's goal is to find the effect of stormwater surface runoff leaking to the sanitary sewer system of Al-Shuhada District of Samawah City in Iraq. The concentrations of major pollutants such as total suspended solids (TSS), chemical oxygen demand (COD), biochemical oxygen demand (BOD), and total dissolved solids (TDS) in the sanitary sewer system were simulated using SWMM during wet weather flow. Furthermore, LID strategies were applied to the model to investigate the effect of BMPs on the quality of sewage.

The research was conducted in the Al-Shuhada District of Samawah, with 8,892 people in 2016. It has a total area of approximately 1.04 km² and is located about 270 km (167.4 miles) Southeast of Baghdad (latitudes 31°71′34″N and longitudes 45°18′48″E), as shown in Figure 1(a). Samawah has a desert climate, with dry-hot summers and cold winters. While no month is entirely moist, November to April receives nearly all of the yearly precipitation. Samawah City receives approximately 106 mm of annual rainfall on average. The soil is sandy-clay, and the surface is smooth (Figure 1(b)). The region's elevations vary from 10 to 20 meters above sea level (Figure 1(c)).

For the study area of Al-Shuhada District, field data based on geographical information system (GIS) profiles were obtained from Samawah Sewerage Directorate. Data include details of sanitary sewer systems such as conduits and nodes, as well as their properties. Hydrological data was acquired from the intensity–duration–frequency (IDF) curve with 2, 5, 10, and 25 years of return periods, as illustrated in Figure 2. This IDF curve was developed from rainfall data for the past 30 years using Easy Fit 5.6 software. To obtain the topographic information, a geographically corrected satellite image of Al-Samawah was selected from the United States Geological Survey (USGS) website in August 2020 from the Landsat 8 satellite.

Furthermore, a land-use map of Al-Shuhada District was provided by Samawah Sewerage Directorate, as shown in Figure 3. The soil type and slopes of Samawah City were produced using a GIS arc map for SWMM simulation. For quality model calibration, sewer samples from the storage station at the study area downstream were collected on 12 January 2021 during one rainfall event when the depth of rain reached 30 mm over 2 h. In both dry and wet weather flow, sewer samples were tested for significant parameters such as BOD, COD, TSS, and TDS, according to standard procedures for water and wastewater analysis (APHA 2017). Time pattern, hourly or sub-hourly, rainfall data, topographic slope and elevation, soil, land use data, sewer system network map are all required to apply SWMM in an urban region. Data rainfall (for continuous modeling), hydrologic part parameters (subcatchments, pipes, junctions, etc.), and runtime controls are all necessary input parameters for the SWMM model to simulate efficiency (time step, start and end times, and so on).

SWMM V. 5.1 model has been operated to evaluate the change in sewage quality in the sewer system. The total amounts of pollutants in the sewer system resulting from mixed dry weather flow pollutants with leaked pollutants washed off the catchment surface during flood events can be simulated by SWMM. First, hydraulic data was imported from GIS arc map to SWMM simulation, then subcatchments areas of leaking surface runoff to the sanitary sewer system were drawn. The amount of surface runoff inflow can only be calculated inside the sewer system by (1) knowing the actual dry weather flow (sewage flow in joints before surface runoff leaking) provided by the sewage directorate department, (2) selecting affecting subcatchments based on the field observation around damaged manholes in the system, and (3) measuring the actual wet weather flow (sewage + surface runoff) in joints within 2 h of the rainfall event. Consequently, the SWMM model of the sanitary sewer system has 80 subcatchments, 283 UPVC conduits, and 276 nodes. Figure 4 shows the model of the sanitary sewer system with all points of leaking surface runoff into the sanitary sewer system, nodes, and conduits.

Furthermore, the primary input data consisted of assigned rain gage, outlet node for each subcatchment, set land-uses, tributary surface area, imperviousness, slope, characteristic width of overland flow, Manning's coefficient for overland flow on both pervious and impervious areas, depression storage on both pervious and impervious areas, and percent of impervious area for depression storage for each subcatchment in SWMM model (Table 1). The user assigned parameters such as rain gage, land-uses and imperviousness were determined by a site examination and the use of ArcGIS and Google Earth, and slope and width were determined using mathematics. In SWMM, the three main techniques for simulating infiltration are Horton, Green-Ampt, and Curve Number. The Green-Ampt model was utilized in SWMM's hydrological model to calculate the quantity of rainwater infiltrated into a permeable surface region's unsaturated upper soil zone, and surface runoff was calculated using Manning's equation.

Dry weather flow for each manhole was calculated from the population in each part of the study area and the planned water consumption rate of 250 L/day capita. Among time patterns of hourly, daily, and monthly sewage variations, this study chose an hourly pattern for SWMM simulation (Hassan et al. 2017). The proposed method provides the averaged or expected daily time series for dry weather inflow patterns in each manhole (Jia et al. 2021). The model was ready to use the rainfall data series after setting up the appropriate parameters. In wet weather flow, BOD, COD, TDS, and TSS (mg/L) were the primarily selected contaminants in SWMM's Data Browser's Quality category. The concentration groundwater infiltration and first-order decay were not considered in the simulation, and no co-pollutants have been identified. It was also assumed that groundwater infiltration is part of dry weather flow in the system.

Land use was specified for each catchment area leaking into the sanitary sewer system in the study area. Two types of land uses were identified in this research area: roads and residential (roof). Before simulating the amount of pollutants that runoff from the study location, an initial accumulation of pollutants must be identified to be washed away during a single rainfall case. Before the simulation, the number of antecedent dry days or the initial build-up mass on each subcatchment should be determined, whereas five antecedent dry days were selected in the analysis.

The coefficients for water quality in the exponential build-up and wash-off equations were calibrated for existing BOD, COD, TSS, and TDS data at several sites. B = C1 (1 − e^−C2t) is the exponential build-up equation, with B is the pollutant build-up mass per unit of subcatchment area (kg/ha), C1 represents the maximum build-up possible (kg/ha), C2 represents the build-up rate constant (1/days), and t is the time (days). W = C3 q^C4·B is the exponential wash-off equation, with W equaling the pollutant wash-off rate per unit area (kg/(h ha)), q equaling the runoff rate per unit area (mm/h), C3 equaling the wash-off coefficient, and C4 equaling the wash-off exponent. Following a sensitivity analysis of the coefficients, they were interactively calibrated for the events through a series of simulations (Rossman 2012).

It is important to note that using sensitive parameters from previous studies may not be possible because each case study has its characteristics, and sensitive parameters would change from one study to another. This study analyzed the model sensitivity to maximum build-up possible, build-up rate constant, wash-off coefficient, and wash-off exponent before the model calibration to determine which model output affects the parameters most. The parameters were tested one by one for land-use (residential and road) to conduct the sensitivity analysis. Initially, default parameters were applied, then values were varied by 10% for each parameter one by one, whereas other parameters remained constant. This process was repeated to identify the model's most sensitive parameters.

Several techniques exist to mitigate pollutants entering sanitary sewer systems during wet weather, including LID technology. The LID control is designed on a per-unit-area basis, allowing it to be applied to an unlimited number of subcatchments. After the model had been calibrated and validated, the selected LIDs were assigned to it. Rain garden, green roof, infiltration trench, permeable pavement, vegetative swale, cell for bio-retention, and rain barrel are examples of standard LIDs defined in SWMM. The Rain Garden LID was chosen in the final model. The LID is assigned to each subcatchment leaking runoff into the sanitary sewer system using the subcatchments LID control editor. The selected LID occupies 7% of each subcatchment in this study. Each subcatchment impervious area is treated by Rain Garden, which treats 40% of each subcatchment impervious area. An equally important input condition involves the rain garden setting parameters. The SWMM user's manual was used, as shown in Table 2 (Rossman 2012).

Sensitivity analysis is conducted to analyze the relative importance and individual influence of different uncertainty factors on the simulation process (Duan et al. 2016). Analysis of sensitivity was carried out before the calibration and validation. BOD, COD, TSS, and TDS were chosen contaminants to reflect water quality in this analysis. The sensitivity of the build-up and wash-off parameters for two land-use forms (residential and road) on the quantity of the pollutants are shown in Table 3. The results indicated that the maximum build-up possible (C1) was the most sensitive parameter (S_i,j > 0.25) on the SWMM quality model for both land uses. The maximum build-up possible (C1) for residential land-use was influential for all pollutants, but the build-up rate constant (C2) was significant only for BOD and COD. The maximum build-up possible (C1) for road use was influential for BOD, TSS, and TDS, while it was very influential for COD. Li et al. (2016) reported that maximum build-up possible was the most sensitive parameter for the TSS peak concentration quality model in support of this sensitivity analysis. The results also showed no wash-off (S_i,j = 0) for the exponential wash-off equation at the storm's end. This is because the majority of the pollutants were already washed away before the storm ended. However, it is essential to report that the sensitivity of the parameters depends on the formulas used for the build-up and wash-off calculation within the model simulation.

Modeled pollutants values were compared with actual pollutants concentrations for model calibration. Pollutants quality model was firstly simulated using default parameters, then sensitive parameters affecting the model were identified. Thereafter, sensitive parameters were varied to calibrate the model by making the difference between modeled and actual pollutants as less than possible. Table 4 displays the calibrated build-up and wash-off input parameters for simulating the quality model. The actual rainfall event recorded on 12 January 2021 was used for model calibration, and the test of model goodness-of-fit for the four pollutants is shown in Figure 5. For all pollutants, modeled concentrations perfectly matched the actual loads. Furthermore, the calibrated model outputs are validated using a second actual rainfall event recorded on 8 February to demonstrate the model efficacy. The model validation was conducted using statistical coefficients of R and NMSE, and the validation results are shown in Table 5. R and NMSE were validated within limits, with values ranging from 0.81 to 0.82, and 0.0173 to 0.022, respectively, indicating that statistical parameters were closest to perfect fit.

During rainfall seasons, the stormwater runoff loads pollutants through the sewer systems and directs them to streams, and finally, the water quality will deteriorate in these waterways (Wang et al. 2016). Using SWMM modeling, Al-Shuhada Quarter sanitary sewer system was evaluated through wet weather flow. BOD, COD, TSS, and TDS loadings are simulated using the model. Whenever it rains, the amount of sewage in the sewer system rises due to stormwater leaking into the system, resulting in pollutants overflow in some manholes and pollutants quality dilution due to the mixing of stormwater and sewage. Figure 6(a)–6(d) assesses the quality of BOD, COD, TSS, and TDS in Al-Shuhada Quarter sanitary sewer system at 2, 5, 10, and 25 years of return periods during 2 h of rainfall duration, respectively. This rainfall duration was selected because the rainfall events in the study area were not more than 2 h. At all return periods, it was noticed that pollutants concentration increased to their maximum level at 20 min and then gradually decreased to a slightly constant minimum value after 2 h. According to Li et al. (2015c), the first 40% of the surface runoff managed to wash 55% of TSS, 53% of COD, 58% of TN, and 61% of TP off the catchment surface, leading to change in pollutants concentrations. Similarly, Alisawi et al. (2015) reported that pollutants concentrations increased at the beginning of rainfall and gradually decreased to an almost constant value of pollutants concentration. Interestingly, the return period has a high effect on the quality of pollutants, whereas long return periods result in lower pollutants concentration and vice versa. The return period is the predicted rainfall intensity over a watershed, diluting pollutants concentration when the return periods are longer. Likewise, in a stormwater drainage system, Rezaei et al. (2019) revealed that pollutants concentration falls with increased rainfall intensities resulting from long return periods.

Figure 6(a) shows the concentration of BOD in the sanitary sewer system during wet weather flow resulting from surface runoff leaking. The quality of BOD concentrations were ranged as 858–1,500, 433–1,070, 250–610, and 143–348 mg/L for 2, 5, 10, and 25 years of return periods, respectively. Figure 6(b) shows the concentrations of COD were in the range of 1,198–2,358, 780–1,928, 451–1,097, and 258–626 mg/L at return periods of 2, 5, 10, and 25 years, respectively. Overall, by comparing the BOD and COD concentrations from Figure 6(a) and 6(b), it can be observed that the short return period of 2 years did not affect the biodegradability (BOD/COD ratio) of sewage. In contrast, further dilution on the quality of BOD and COD at longer return periods resulted in gradually lower sewage biodegradability due to the high amount of stormwater leaking to the system compared to the sewage flow.

Furthermore, Figure 6(c) illustrates that the concentrations of TSS were within 2,131–4,520, 1,702–4,090, 1,000–2,334, and 585–1,337 mg/L for return periods of 2, 5, 10, and 25 years, respectively. On the other hand, Figure 6(d) reported that the TDS concentrations were in the range of 4,918–9,900, 3,489–8,625, 2,018–4,909, and 1,155–2,803 mg/L at 2, 5, 10, and 25 years of return periods. As a result, it can be noticed that the concentrations of TSS and TDS increased at the beginning of rain event due to pollutants washing, followed by a gradual decrease in the amount of pollutants as it is discharged in the system. It is also important to report that high concentrations of TSS and TDS are due to the highly dusty streets and unpaved roads. This should be considered during the design of sewage treatment plants to withstand high quantities of wastewater and solid pollutants during wet weather flow. Aging sewer systems and undeveloped areas are a major cause of the leaking of stormwater surface runoff in the urban system, increasing the concentration of the pollutant in sanitary sewer systems (Hussein et al. 2015).

Stormwater surface runoff leaking into sewer systems occurs for various causes, including gaps, pits, and cracked manhole covers, in addition to illegally opened manhole covers to discharge surface runoff quickly. This will add further pollutants washed off the surfaces of the sewer system that may disturb the sewage treatment processes. There are several methods to treat surface runoff pollutants, including LID technique. After SWMM model calibration and validation, the selected LID was applied to the SWMM simulation. Rain gardens were selected due to their effectiveness in reducing pollutants wash off and applicability in the study area. After that, BOD, COD, TSS, and TDS concentrations were evaluated, as shown in Figure 7(a)–7(d), respectively. At 2, 5, 10, and 25 years of return periods, in respective manner, the BOD concentration was reduced by 88, 76, 53, and 9%; whereas the COD concentration was dropped by 84, 76, 52, and 7%; while TSS concentration was lowered by 82, 75, 52, and 7%; and TDS were decreased by 85, 77, 55, and 10%.

As can be observed, LID was very effective at reducing all pollutants concentrations compared with initial values. The findings of this study showed that LID technique is critical in the urban region to offset imperviousness's effects on both amounts of quantity and quality. As a result, LID strategies may dramatically reduce urban runoff and improve stormwater quality. This is attributed to the application of only a few LID units in the study area. In another way, more LIDs are needed in critical areas to reduce peak runoff and further pollutants (Li et al. 2019a). However, the expense of implementing LID should be taken into account as well. To determine the ideal numbers and location of LIDs, optimization methods could be used.

According to the literature, LIDs have an excellent capability for reducing runoff and removing pollutants in metropolitan settings. Using a rain garden and a vegetated swale, TSS and TN can be reduced by up to 89 and 58%, respectively (Martin-Mikle et al. 2015). Bio-retention cells could reduce expected peak flows by at least 45% in Maryland and North Carolina (Hunt et al. 2008). Sediment and nutrient levels can be reduced by up to 99% using bioretentions. In contrast, Jiang et al. (2017) studied two rain gardens (each 24 m²) in the field and found average reduction ratios of 52, 40, 57, and 57% for TSS, COD, TN, and TP, respectively. According to Li et al. (2019b), TSS, COD, TN, and TP reduction ratios with a rainfall intensity of 26 mm/h were 19.7, 19.2, 18.6, and 16.9%. In a laboratory setting, Davis et al. (2006) found that rain gardens’ TN and TP removal ratios on urban rainwater runoff were 50–55% and 70–85%, respectively.

SWMM was confirmed to be an efficient tool to model the sewage quality in the sanitary sewer system of the Al-Shuhada Quarter. Among four parameters, only maximum build-up possible was the most sensitive parameter for road and roof land-uses. Thereafter, using actual data from two rainfall events, the model was successfully calibrated and validated. The model assessment indicated changes in pollutants quality in the sanitary sewage system during the leaking of stormwater surface runoff into the system, especially a high increase in pollutants concentrations during the initial phase of street washing that lasted for 20 min. This may cause a burden on the efficiency of sewage treatment plants during rainy seasons. Nevertheless, the modeled rain garden proved to be an effective LID technique for the reduction of pollutants concentrations, whereas about 82–88, 75–77, 52–55, and 7–10% of all pollutants were reduced at return periods of 2, 5, 10, and 25 years. The findings of this study are expected to benefit a variety of stakeholders, including the city government, sanitary sewer network designers, sanitary managers, and planners.

M.H.M. conceived the study and wrote the first draft. H.M.Z. supervised, analyzed the data, wrote the first draft, and revised the successive drafts of the manuscript. W.H.H. supervised, conceptualized the methodology, and interpreted the results.

The authors would like to thank Samawah Sewerage Directorate for providing information.

The authors declare that no conflicts of interest.

All relevant data are included in the paper or its Supplementary Information.