Modeling phosphorus transport in soil columns amended with polyaluminum chloride and anionic polyacrylamide water treatment residuals: Implications for stormwater bioretention systems

Phosphorus pollution in stormwater runoff has been a global environmental concern for decades, as stormwater flows over land surfaces containing phosphorus-rich substances during storm events (Xu et al. 2020; Biswal et al. 2022; Azad et al. 2024). Phosphorus in stormwater runoff primarily originates from the application of fertilizers in lawns and gardens in urban and suburban areas as well as in agricultural farms, various animal wastes, and industrial and commercial activities (Xu et al. 2020; Ali & Pickering 2023). Excessive phosphorus in stormwater runoff can increase the risks of eutrophication, algal blooms, particularly harmful algal blooms, and the disruption of aquatic ecosystems in surface water bodies (Xu et al. 2020; Ali & Pickering 2023). Therefore, it is crucial to develop proper treatment methods to address phosphorus pollution in stormwater runoff before it directly flows into surface water bodies (Xu et al. 2020; Biswal et al. 2022; Azad et al. 2024).

Currently, best management practices (BMPs) are primarily strategies and techniques designed to mitigate various contaminants in stormwater runoff and protect water quality in surface water bodies (Hunt et al. 2008; Agyin-Birikorang et al. 2009; Liu et al. 2018; Okaikue-Woodi et al. 2020). One widely utilized BMP globally is the stormwater bioretention system, alternatively known as a stormwater biofilter or a rain garden (Yang et al. 2009; Fang et al. 2021; Li et al. 2021; Vijayaraghavan et al. 2021; Biswal et al. 2022). The stormwater bioretention system not only lowers contaminant concentrations but also reduces peak flow and overall stormwater runoff volumes during storm events (Roy-Poirier et al. 2010; Vijayaraghavan et al. 2021). Additionally, the stormwater bioretention system can provide ecological and aesthetic benefits to the surrounding environment (Roy-Poirier et al. 2010; Vijayaraghavan et al. 2021). As a result, stormwater bioretention systems have garnered global attention for research and subsequent field engineering applications in addressing diverse forms of pollution, including phosphorus, nitrogen, heavy metals, and pathogens in stormwater runoff (Roy-Poirier et al. 2010; Xu et al. 2020; Vijayaraghavan et al. 2021).

Stormwater bioretention systems typically consist of several key components, including plants, a ponding layer, a mulch layer, a filling media, and a drainage structure at the bottom (Roy-Poirier et al. 2010; Vijayaraghavan et al. 2021). Among these components, the filling media is a critical component for the treatment of stormwater runoff (Okaikue-Woodi et al. 2020; Biswal et al. 2022). Traditional soil-based media have been the standard for stormwater bioretention systems worldwide for several decades; however, previous studies have indicated that their treatment performance can be inadequate (Okaikue-Woodi et al. 2020; Biswal et al. 2022). Consequently, global researchers have been actively exploring innovative media to either entirely replace traditional soil-based media or to integrate them with traditional soil-based media to enhance the capacity and effectiveness of stormwater bioretention systems in controlling stormwater runoff pollution (Okaikue-Woodi et al. 2020; Biswal et al. 2022).

Researchers worldwide have proposed the utilization of water treatment residuals (WTRs) as an amendment to traditional soil-based filling media in stormwater bioretention systems to address stormwater runoff contaminants (Xu et al. 2020). WTRs are waste products generated during drinking water treatment processes. Recycling WTRs into stormwater bioretention systems can not only enhance their treatment capabilities but also offer a sustainable solution for managing waste in the water treatment industry (Xu et al. 2020; Duan et al. 2022). Numerous studies have demonstrated that aluminum-based or iron-based WTRs exhibit high phosphorus adsorption capacities, making them promising candidates for phosphorus removal in stormwater runoff (Xu et al. 2020). Moreover, recent upgrades and expansions of municipal water treatment plants have led to the substantial generation of WTRs containing polyaluminum chloride (PAC) and anionic polyacrylamide (APAM) (Duan & Fedler 2021a). These compounds are commonly used as coagulants and dewatering agents, respectively, in municipal drinking water treatment plants in the world. Previous batch study has highlighted the high phosphorus removal capabilities of PAC-APAM WTRs, suggesting their potential as an amendment in traditional soil-based stormwater bioretention systems to effectively enhance phosphorus removal from stormwater runoff (Li et al. 2022).

However, batch studies on phosphorus adsorption by various WTRs have limitations in providing comprehensive insights for the recycling of WTRs into stormwater bioretention systems (Duan & Chen 2022). Column studies offer distinct advantages over batch studies by more closely simulating phosphorus transport in the filling media of stormwater bioretention systems, providing deeper insights into the dynamics of phosphorus transport under field conditions. To date, there has been a notable absence of column studies and modeling research focusing on phosphorus transport in soils amended with WTRs. This lack of comprehensive research has hindered the full utilization of WTRs as amendments in traditional soil-based stormwater bioretention systems. Addressing this gap through column studies and modeling efforts is essential for unlocking the potential benefits of WTRs in stormwater management.

Therefore, the main objective of this study was to investigate phosphorus transport in three soil types amended with PAC-APAM WTRs under saturated steady-state flow conditions through column studies. Additionally, this study aimed to provide a deeper understanding of the impact of flow rate variations on phosphorus transport. To achieve this, Hydrus-1D, a robust modeling software (Šimůnek & Genuchten 2008; Chotpantarat et al. 2012; Behroozi et al. 2020), was employed to simulate and analyze phosphorus transport under different flow rates. This approach not only contributes to the existing body of knowledge but also aids in optimizing the design and management of bioretention systems for enhancing phosphorus removal efficiency.

The samples of PAC-APAM WTRs and the three types of soils used in this study were the same as those reported in a previous study on phosphorus adsorption by soils amended with PAC-APAM WTRs (Li et al. 2022). They were collected from a local municipal water treatment plant and agricultural farms, respectively, located in Taiyuan city, Shanxi Province, China. Additionally, the samples of PAC-APAM WTRs used in this study were the same as those reported in a series of previous studies. Therefore, the pretreatment method and physicochemical properties of the PAC-APAM WTRs samples can be found in these studies (Duan & Fedler 2021a, b, 2022; Duan et al. 2022). The bulk density, pH, specific surface area, total pore volume, and average pore width of the PAC-APAM WTRs were 1.144 g/cm³, 7.6, 36.693 m²/g, 0.0623 cm³/g, and 6.7957 nm, respectively (Duan & Fedler 2021a). The three types of soils used in this study were silty clay soil (Soil 1), loam soil (Soil 2), and silty loam soil (Soil 3). Detailed information regarding the pretreatment of these soil samples can be found in a previous study (Li et al. 2022). The bulk densities of Soil 1, Soil 2, and Soil 3 were 1.144, 1.107, and 0.902 g/cm³, respectively. The specific surface areas were 5.357, 3.506, and 26.208 m²/g for Soil 1, Soil 2, and Soil 3, respectively. The total pore volumes (PVs) were 0.0093, 0.0073, and 0.0307 cm³/g. The average pore widths were 6.9616, 8.2849, and 4.6915 nm. The pH (m/v = 1:2) values were 7.78, 7.75, and 7.20; the electrical conductivities (EC) (m/v = 1:2) were 84.3, 79.5, and 181.0 mS/cm; and the total organic carbon contents were 21.235, 20.810, and 20.690 g/kg for Soil 1, Soil 2, and Soil 3, respectively (Li et al. 2022).

Column experiments were conducted using glass chromatographic columns, each with a net height of 10 cm and an inner diameter of 2.5 cm. A previous study demonstrated that an optimized weight ratio of PAC-APAM WTRs to a specific type of soil for effective phosphorus adsorption was established at 20%:80% (Li et al. 2022). Consequently, the mixed media within the columns was composed of a specific type of soil amended with PAC-APAM WTRs in the same ratio.

After thorough mixing of the soil and PAC-APAM WTRs, the mixed media were wet-packed in the columns, layer by layer, with each layer approximately 2 cm in height. Subsequently, the columns, once filled with the media, experienced saturation and stabilization using at least six PVs of ultrapure water. The ultrapure water was pumped into the columns from the bottom by a peristaltic pump (Precision Pump Co., Baoding, China) to ensure that any entrapped air within the column profile during the media packing process was effectively expelled (Lin & Xu 2020).

To determine the longitudinal hydrodynamic dispersion coefficient (D_L, cm²/h) and the average longitudinal dispersivity (α_L, cm) for modeling phosphorus transport and to confirm solute transport as an equilibrium process (Chotpantarat et al. 2012), tracer (KBr) transport experiments were conducted prior to the phosphorus transport experiments. A bromide solution with an initial concentration (C₀) of 0.4 mmol/L was continuously pumped into the columns at the bottom with the flow rate of 6, 8, or 10 mL/h. At different time intervals, the bromide concentration (C) in the effluent at the column top was measured using ion chromatography (Thermo Scientific, USA) following collection using an automatic fraction collector (BSZ-100, Shanghai Precision Instrument Co., Ltd, Shanghai, China), centrifugation, and filtration. Once C/C₀ reached its maximum, the columns were continuously flushed with ultrapure water until the bromide concentration in the effluent approached zero.

The phosphorus transport experiments were conducted in triplicate following a similar protocol to the bromide transport experiments. The specific conditions for phosphorus transport were set at a solution pH of 6, an initial phosphorus concentration (C₀) of 1,000 mg/L, and flow rates of 6, 8, and 10 mL/h.

Bromide transport in soil columns amended with PAC-APAM WTRs was modeled using the computer program CXTFIT, version 2.0 (Toride et al. 1999), employing non-linear least squares optimization (Chotpantarat et al. 2012). Throughout the modeling process, the convection-dispersion equation (CDE) model (Song et al. 2018; Ma et al. 2021) and the two-region physical non-equilibrium model (Chotpantarat et al. 2012) were utilized to estimate D_L and α_L (Duan & Chen 2022).

The transport of phosphorus in soil columns amended with PAC-APAM WTRs was simulated using the software HYDRUS-1D (Šimůnek & Genuchten 2008; Jiang et al. 2019b; Ma et al. 2021). Three models were employed to model the breakthrough curves (BTCs) of phosphorus transport: the linear CDE model, the non-linear CDE model based on the Langmuir isotherm, and the chemical non-equilibrium two-site model (TSM) (Duan & Chen 2022). These BTCs were plotted with the number of PVs as the x-axis and the normalized concentration C/C₀ as the y-axis (Duan et al. 2022).

The linear CDE model assumes that phosphorus adsorption by a soil amended with PAC-APAM WTRs is a linear chemical reaction, whereas the non-linear CDE model suggests that phosphorus adsorption is a monolayer adsorption process. The TSM model encompasses two solid-phase processes for phosphorus adsorption onto a soil amended with PAC-APAM WTRs. It proposes that phosphorus adsorption at type-1 sites reaches equilibrium instantaneously, while at type-2 sites, phosphorus adsorption is controlled by the rate of interaction between phosphorus and the soil amended with PAC-APAM WTRs (Šimůnek & Genuchten 2008; Song et al. 2018; Behroozi et al. 2020).

In these equations, R represents the retardation factor, C (mg/L) denotes the phosphorus concentration, t (h) signifies the transport time, ρ (g/cm³) and θ (%) are the bulk density and the porosity of the soil amended with PAC-APAM WTRs, x (cm) represents the column height, f_e indicates the fraction of equilibrium phosphorus adsorption sites, K_d is the solid–liquid distribution coefficient, β is the Freundlich exponent, η represents the Langmuir coefficient, and ω (1/h) denotes the first-order rate coefficient (Šimůnek & Genuchten 2008; Song et al. 2018; Behroozi et al. 2020; Duan & Chen 2022).

The results of modeling bromide transport in the columns of soils amended with PAC-APAM WTRs using the two-region physical non-equilibrium model showed that all dimensionless mass transfer coefficients (ω_L) were 100, as shown in Table 1. This suggests that the transport of bromide in all experimental columns occurred under physical equilibrium conditions (Shahmohammadi-Kalalagh & Taran 2019). Moreover, the findings imply that the chemical reaction sites within the filling media were immediately available for phosphorus adsorption during the subsequent phosphorus transport experiments (Shahmohammadi-Kalalagh & Taran 2019). This instantaneous availability is crucial for understanding the dynamics of phosphorus adsorption and transport in the soils amended with PAC-APAM WTRs.

These modeling outcomes further validate the necessity to employ the deterministic equilibrium CDE model to characterize bromide transport in all columns in this study (Toride et al. 1999). By applying this model, two important parameters, D_L and α_L, were estimated as presented in Table 1 for the subsequent phosphorus transport modeling processes. These results laid a solid foundation for optimizing modeling approaches and enhancing understanding for phosphorus transport in the soils amended with PAC-APAM WTRs.

The results of modeling phosphorus transport in the columns of Soil 1 amended with PAC-APAM WTRs showed that the CDE model using the Langmuir isotherm performed slightly better than the CDE model using the linear isotherm and the chemical non-equilibrium TSM when the flow rate was 6 mL/h since it has a marginally higher R² (Table 2). However, the TSM model had better performance than the other two models when the flow rate was 8 and 10 mL/h to describe phosphorus transport in the columns of Soil 1 amended with PAC-APAM WTRs (Table 2). This implies that the primary governing mechanism for phosphorus transport when the flow rate was 6 mL/h was instantaneous adsorption and it transferred to two-site adsorption including instantaneous adsorption and first-order kinetic adsorption when the flow rate increased to 8 and 10 mL/h (Chotpantarat et al. 2012).

The results showed that the value of K_d obtained from the modeling of phosphorus transport using three models decreased with increasing flow rate (Table 2). This was attributed to the decrease in contact time between phosphorus and the mixed media in the columns, which caused less phosphorus adsorption (Jiang et al. 2019a). All values of b were small (Table 2), which are related to the maximum monolayer adsorption capacity, indicating that the phosphorus adsorption was weak under saturated steady-state flow conditions in the range of flow rate between 6 and 10 mL/h (Šimůnek & Genuchten 2008). Also, the value of f_e decreased with an increase in flow rate (Table 2). This indicated that a higher flow rate caused less instantaneous phosphorus adsorption. Unfortunately, low values of ω (Table 2) indicated that the phosphorus adsorption reactions at the non-equilibrium sites, type-2 sites, were slow.

The modeling results for phosphorus transport in Soil 2 amended with PAC-APAM WTRs showed that at a flow rate of 6 mL/h, both the TSM model and the CDE model with linear isotherm provided a better fit for describing phosphorus BTC data than the CDE model using the Langmuir isotherm (Table 3). Notably, the R² values of these two models were nearly identical (Table 3). At a flow rate of 8 mL/h, the CDE model with linear isotherm was superior in describing phosphorus transport data compared with the other two models (Table 3). Conversely, at a flow rate of 10 mL/h, the TSM model demonstrated the optimal performance in modeling phosphorus transport (Table 3). This suggests that with the increasing flow rate, the primary mechanism governing phosphorus transport shifted from instantaneous adsorption to two-site adsorption as observed in the columns of Soil 1 and PAC-APAM WTRs.

The value of K_d decreased with an increase in flow rate. This is because a higher flow rate reduced contact time between phosphorus and the filling media and resulted in lower phosphorus adsorption (Jiang et al. 2019a). The low value of b in the TSM model at the flow rate of 10 mL/h indicated weak affinity of phosphorus to the filling media, as observed in the columns of Soil 1 amended with PAC-APAM WTRs. However, both K_d and b (Table 3) were higher than those observed in the columns of Soil 1 amended with PAC-APAM WTRs (Table 2) at the flow rate of 10 mL/h. This demonstrates that phosphorus adsorption in Soil 2 amended with PAC-APAM WTRs was higher than that in Soil 1 amended with PAC-APAM WTRs, resulting in more phosphorus retention. This is further supported by the higher values of K_d in the CDE model with linear isotherm (Table 3) compared with those in Soil 1 with PAC-APAM WTRs (Table 2) at flow rates of 6 and 8 mL/h. Comparing the values of f_e in the TSM model at flow rates of 6 and 10 mL/h, it can be found that a high flow rate caused a low fraction of equilibrium phosphorus adsorption sites, potentially resulting in reduced phosphorus retention and increased phosphorus mobility.

The results of modeling phosphorus transport in Soil 3 amended with PAC-APAM WTRs showed that the TSM model provided a better description of phosphorus transport data across the three flow rates (Table 4). Notably, at a flow rate of 8 mL/h, the CDE model using linear isotherm (R² = 0.8041) exhibited a marginally better performance compared with the TSM model (R² = 0.7907). Thus, the performance of both models was deemed to be relatively equivalent. These results imply that two-site adsorption was the predominant mechanism influencing phosphorus transport in Soil 3 amended with PAC-APAM WTRs throughout the flow rate range of 6–10 mL/h.

In comparison with observations from the other columns (Tables 2 and 3), the value of K_d in Soil 3 with PAC-APAM WTRs was lower (Table 4), indicating lower phosphorus adsorption. The low f_e values in the TSM models (Table 4) indicated that the majority of phosphorus adsorption occurred at type-2 sites and was rate-limited by adsorption kinetics. The low ω values further demonstrated that adsorption rates at these sites were quite slow. These findings collectively suggest that during phosphorus transport processes, there was a tendency for phosphorus to move readily to the column outlet with minimal retention and high mobility. Consistent with observations from the other columns, K_d decreased as the flow rate increased.

Although numerous previous studies have investigated the performance of WTRs recycled into the filling media of stormwater bioretention systems for removing various pollutants, including total suspended solids and total nitrogen, limited research has focused on phosphorus removal from stormwater runoff by WTRs (Xu et al. 2020). Moreover, previous studies regarding phosphorus removal by WTRs have been limited to laboratory batch adsorption studies (Xu et al. 2020; Li et al. 2021; Duan & Fedler 2021a), with a lack of research employing column studies. This study addresses this gap by employing column studies to investigate phosphorus transport in three types of soils amended with PAC-APAM WTRs. The findings provide valuable implications for the engineering application of WTRs recycled into bioretention systems as a component of the filling media.

It is commonly and widely accepted that selecting a soil type with high phosphorus adsorption capacity is one of the primary factors when designing a traditional soil-based stormwater bioretention system for removing phosphorus from stormwater runoff (Roy-Poirier et al. 2010). This is because soils with high phosphorus adsorption capacity can effectively capture and retain phosphorus, preventing it from entering water bodies and causing pollution. However, in order to successfully recycle PAC-APAM WTRs into the traditional soil-based filling media of bioretention systems for removing phosphorus, soil selection cannot be based solely on the maximum phosphorus adsorption capacity of soils. Soil selection requires an evaluation of the maximum phosphorus adsorption capacity of a soil amended with PAC-APAM WTRs. For instance, a previous study found that Soil 1 (silty clay soil) had a higher maximum phosphorus adsorption capacity (1.593 mg/g) than Soil 2 (loam soil) (1.517 mg/g) and Soil 3 (silty loam soil) (1.282 mg/g) (Li et al. 2022). However, after being amended with 20% PAC-APAM WTRs, Soil 2 exhibited a higher maximum phosphorus adsorption capacity (2.001 mg/g) in batch static phosphorus adsorption experiments (Li et al. 2022). Moreover, this column study demonstrated that Soil 2 amended with PAC-APAM WTRs also had more phosphorus adsorption and less phosphorus mobility than the other two soil types amended with PAC-APAM WTRs as mentioned above. In this regard, the results of this column study were consistent with those of batch phosphorus adsorption study using the same soil and PAC-APAM WTRs by Li et al. (2022).

This study highlights the critical role of designing and managing the depth of the ponding layer in stormwater bioretention systems. The depth of the ponding layer is a pivotal parameter that regulates the flow rate of stormwater runoff within the bioretention cell under saturated flow conditions. Generally, an increase in the depth of the ponding layer leads to a higher pressure and results in a higher flow rate (Roy-Poirier et al. 2010). As discussed above, increased flow rates can result in increased phosphorus leaching, potentially contributing to phosphorus pollution in surface water bodies. However, the ponding layer also plays a crucial role in mitigating the overall volume of stormwater runoff and attenuating peak flow during stormwater events. The storage capacity provided by a deeper ponding layer can temporarily detain larger volumes of runoff, reducing the frequency and magnitude of peak flows that can lead to downstream flooding and erosion. Therefore, optimizing the design of the ponding layer's depth is essential for balancing the dual objectives of phosphorus removal and stormwater management and requires careful consideration. Future research should continue to explore the relationship between ponding layer depth, pollutant removal, and hydrological performance to inform best practices in bioretention system design.

This study determined that the TSM model demonstrated superior performance in characterizing phosphorus transport in soils amended with PAC-APAM WTRs. This finding suggests that future research should more clearly distinguish between instantaneous phosphorus adsorption and the slower adsorption processes that are kinetically limited. Consequently, soils amended with PAC-APAM WTRs with high instantaneous phosphorus adsorption capacities should be strongly considered as candidates for filling media in bioretention systems in engineering application. Regarding the modeling aspect, the TSM model exhibited a relatively weaker capacity to describe phosphorus transport in filling media with low instantaneous phosphorus adsorption and slow adsorption reaction rates. Future research should aim to improve the predictive ability of the TSM model or develop more advanced models to accurately simulate phosphorus transport in filling media akin to Soil 3 amended with PAC-APAM WTRs. Moreover, advanced modeling techniques should be developed to address phosphorus transport in stormwater bioretention systems with aging filling media, preferential flow due to plant root influences, and complex configurations, such as internal water storage.

The transport of phosphorus in columns containing three distinct soil types amended with PAC-APAM WTRs under saturated steady-state flow conditions was effectively modeled using three different models: the CDE model with linear isotherm, the CDE model with Langmuir isotherm, and the TSM model. In both Soil 1 (silty clay soil with a specific surface area of 5.357 m²/g) and Soil 2 (loam soil with a specific surface area of 3.506 m²/g) amended with PAC-APAM WTRs, the primary mechanism governing phosphorus transport shifted from instantaneous adsorption to two-site adsorption as the flow rate increased. In contrast, Soil 3 (silty loam soil with a specific surface area of 26.208 m²/g) amended with PAC-APAM WTRs was predominantly governed by two-site adsorption throughout the experiments. The increase in flow rate not only influenced the dominant mechanism of phosphorus transport but also reduced K_d and f_e at type-1 sites, resulting in decreased phosphorus adsorption, increased phosphorus mobility, and consequently, an elevated risk of phosphorus pollution.

This study highlights several critical points relevant to the design and management of stormwater bioretention systems. It strongly recommends the selection of soil amended with PAC-APAM WTRs that exhibits higher instantaneous phosphorus adsorption. Additionally, it emphasizes the importance of designing the depth of the ponding layer appropriately, utilizing the TSM model, and refining modeling techniques. These considerations are essential for optimizing phosphorus removal and mitigating associated pollution risks in stormwater bioretention systems utilizing soil amended with PAC-APAM WTRs as the filling media. More advanced models are required in the future to accurately describe phosphorus transport in stormwater bioretention systems under field conditions, which will improve the management of phosphorus in stormwater runoff.

This work is supported by Open Project of State Key Laboratory of Urban Water Resource and Environment (Grant No. HC202150), China.

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

The authors declare there is no conflict.