Different erosion characteristics of sediment deposits in combined and storm sewers

Urban wet-weather discharges from combined sewer systems to receiving waters alter river hydro-geomorphology, water quality and sanitary quality (Becouze-Lareure et al. 2016). In general, a storm drainage system is designed to discharge storm runoff into a receiving water. On dry weather days, however, non-storm water discharges also enter storm drainage systems, resulting in the release of untreated sewage in urban receiving waters, especially in the metropolises of China (Field et al. 1994; Li et al. 2014). Both combined and storm sewer overflows contribute significant wet weather pollutants to urban rivers. A large fraction of these pollutants are trapped on sediment deposits (Zgheib et al. 2011; David et al. 2013). Although Field & Struzeski (1972) showed 20–30% of the annual production of wastewater solids settled and were discharged during storm events in Buffalo (NY, USA), it was still unclear as to what type of particles were actually eroded during rain events in combined and storm sewers (SS).

The amounts and types of eroded solids could be affected by many factors during periods of rainfall, particularly in terms of the biochemical and physical properties of deposits, antecedent dry weather period, and rainfall intensity. For instance, Biggs et al. (2005) placed real deposits (sediments from Loenen, The Netherlands and Dundee, UK) in the base of an annular flume with varying conditions for consolidation. However, the transport and amount of eroded sediments also play an important role in this research field, but are rarely assessed. In addition, Banasiak et al. (2005) made use of sewer sediment (from a combined sewer system in the city of Dendermonde, Belgium), observing strong biochemical reactions during the deposit formation period. Unfortunately, these researchers did not discuss the differences between sediments from combined and SS. Different combined and SS systems may have different dimensions of illicit connection, which may result in differences in the types of sediments.

The aim of this project is to investigate the different erosion patterns and initial effects of combined sewer sediments (CSSD) and storm sewer sediments (SSSD) under different consolidation periods. The flux on eroded particle size and flush pollutants are determined by continuous monitoring. We hypothesize discovering some differences in the initial effects between CSSD and SSSD. This study seeks to determine the parameters of flushing gates or weirs in different combined and SS, leading to a decrease in pollution from wet weather discharges.

Between 2008 and 2012, according to the local government, the length of pipe for dredging the urban drainage system in Shanghai was much longer than in rural areas, and the amount of siltation in urban drainage systems was also higher than in rural areas (0.02–0.04 t/m pipe >0.006–0.027 t/m pipe). This finding indicates that the sedimentation in urban areas is heavier than in rural areas. Thus, the local government made urban drainage systems focus on serious sedimentation. Therefore, it is more reasonable for us to select urban drainage systems as a research project.

To avoid the destruction of the structure of the sewer sediments, deposit samples were pumped with a suction device, consisting of a peristaltic pump and a polyethylene tube (see Figure S1, Supplementary material, available with the online version of this paper). The tube was sunk into the sediment using a mallet to the depth required (15 cm in this study). The peristaltic pump was used to create a vacuum that sucked the sediment. A new tube was used for each sample. All used tubes were carefully transferred to polypropylene containers and stored on ice until arrival in the laboratory.

To avoid the influence of overlying water on samples, a period of consolidation was set in the laboratory. After several hours, the overlying water was excluded, and the sediments were ready for the following tests. Wei (2012) found that the water-solid interface would be stable after a consolidation of approximately 4 h (see Figure S2, Supplementary material, available with the online version of this paper). Twenty samples were collected in separated drainage systems (ID400, No.1–4; ID600, No.1–4; ID800, No.1–4; ID1000, No.1–4; ID1200, No.1–4, here ID400 means internal diameter of 400 mm). Twenty samples were also collected in combined sewer systems (ID400, No.1–4; ID600, No.1–4; ID800, No.1–4; ID1000, No.1–4; ID1200, No.1–4).

A flush model was designed for the erosion test (see Figure S3, Supplementary material, available with the online version of this paper). The pipe diameter (outside diameter) was 200 mm, and the inside diameter was 190 mm. The length of the pipe was 7.0 m, and the pipe material was un-plasticized polyvinyl chloride. The pipe slope was 8.03‰. The roughness of the inside pipe wall was n = 0.018 (n is the Manning coefficient. Units for values of n are often left off; however it is not dimensionless, having a unit of s/[m^1/3]). The pipeline was divided into two parts. The length of one section was 6.8 m, and the top 7 cm part was cut off, primarily for the convenience of such parameters as placing sediments and calibrating the flow. The rest of the pipe remained intact, and its length was 0.2 m. The diameter of the return pipe was 50 mm; it was made of PVC, and was used for adjusting flow. The flush pipe and the return pipe were connected with a PVC bucket (volume = 685 L). An agitator was installed in the bucket, and the distance was 30 cm from the blades to the bottom of the bucket. An online particle test probe was also set in the bucket for testing turbidity every 20 s. A single-stage vertical centrifugal pump (Rated Power = 21 m³/h) and the bucket were connected by a PVC pipe. To regulate the flow, a butterfly valve and a Holykell electromagnetic flowmeter were sequentially arranged behind the centrifugal pump. The control process can be seen in the Supplementary material.

After consolidation, the sediments were laid on the flush pipe. Two different flows (1.30 L/s and 5.95 L/s) were set in these tests. An online particle test probe (Range: 0–10 g/L; Resolution: 0.1 mg/L; Model: SYCAMIN 7200 Germany) was set in the bucket for testing turbidity every 20 s. Simultaneously, aqueous samples were collected for analysing particle size by a Mastersizer 3000 laser diffraction grain size analyser (Range: 0.01–3,500 μm). The volume of every sample was 200 mL. DO was measured in situ using a Hydrolab DS5X multi-probe sonde (Hach, USA).

Critical shear stress was measured to analyse how it is affected by particle size, moisture, relative density and organic matters. Five sources (Storm sewer ID400, ID800, ID1000 and Combined sewer ID800, ID1000) of original sediments were measured. Critical shear stress was defined as the value of the shear stress that would yield a zero transport rate (Vanoni 1964). Different methods of detecting critical shear stress exist and can produce different values. The method for measuring critical shear stress was described by Seco et al. (2014). Moisture and organic content of sediments (volatile suspended solids/total suspended solids, VSS/TSS) were measured according to the Chinese standard method (CJ/T 221–2005). Oxygen uptake rates (OUR) of sediments were determined and modeled according to the description by Vollertsen et al. (1999). Relative density was measured by a densitometer (SARTORIUS BSA-DS3). In addition, redundancy analysis was used for correlation analysis by R language.

To compare the erosion characteristics between combined and SS, parts of samples (Storm sewer ID400, ID800 and Combined sewer ID800) were stratified for particle size and moisture analysis.

The turbidity value significantly increases under the higher flow condition (Q = 5.95 L/s), and the value of SS (ID1000) reaches its peak relatively later than other samples (Figure 3(a)). Under the higher flow condition, the turbidity value of CSSD after consolidation of 30 days increases more than without consolidation; thus, it reaches its peak in a shorter time. This may suggest that its sub-layer is exposed earlier to a higher flow rate. In addition, the consolidation process may cause structural changes in the deposits and results in a separation between the surface layer and the bottom layer. Our results are consistent with the findings of Ristenpart (1995) and De Sutter et al. (2003) that the consolidation process could result in an increase of erosional strength with depth.

After 30 days' consolidation, the turbidity variation trend of the SSSD is steadier than that of CSSD (Figure 3(b)). The moisture content of SSSD regulated by different rainfall conditions has a positive effect on the erosion resistance of SSSD (ID800, ID1000) (Figure 4). Nevertheless the effects of VSS on these two types of sediments are not significant (Figure 4). A high proportion of illicit connections could cause an increase in organic matter in CSOs (Li et al. 2010). Therefore, this may imply few illicit connections remained in the study sewer networks, which is consistent with the previous perambulation. The particle size of the SSSD is less than that of the CSSD on average (Figure S5, Supplementary material, available with the online version of this paper). A smaller particle has a larger specific surface area; therefore, the static friction between particles of small size is greater than that of large size particles. In addition, organic matter in the sediment could also produce viscosity between particles. The critical shear stresses of sediments from SS-ID1000 and SS-ID800 are observed to be negatively affected by their particle size (Figure 4). Thus, the small size may improve the erosion resistance of the sediment from the above two types of pipelines. It could further illustrate that the viscosity was considerably smaller than the static friction force between the particles in this study. Because VSS of SSSD is less than that of CSS (Figure 2(d)), VSS content may have little effect on the erosion resistance of SSSD according to the above analysis. Therefore, after 30 days of consolidation, the variation of turbidity of the SS is not significant to some extent (Figure 3(b)). In the urban areas of Shanghai, SS of ID400 are typically used to connect rain grates and main pipes, which causes the characteristic differences between the sediments from a branch pipe (ID400) and main pipes (ID1000 or ID800). Simultaneously, water gushing occurs much more frequently in rain grates and branch pipes, where rainwater accumulates severely. Thus, on rainy days in urban areas, a significant amount of sediments frequently accumulate with high water content. This finding is consistent with the results shown in Figure 2(c). Overall, these findings indicate that the grain size and relative density of sediments are both important factors that affect the erosion resistance of sediments from the storm sewer of ID400 (Figure 4).

The variation of particle size against time is not as regular as the turbidity. Therefore, values of particle size are sorted from small to large order (Figure S5a, b, Supplementary material). The results show that CS-ID800 > CS-ID1000 > SS-ID800 > SS-ID1000 (Figure S5a, b, Supplementary material), but this is opposite to the peak time of turbidity (CS-ID800 < CS-ID1000 < SS-ID800 < SS-ID1000, Figure 3(b)). Therefore, the particle size of sediments may affect the interaction between the particles, thus modifying the erosion resistance of sediments. The effect of particle size on the interaction between particles may exist in the following two aspects: (1) small particles size leads to large specific area, which increases the friction between particles; (2) particles with large specific area could attach larger amounts of viscous organic matter (e.g., humic acid), which may increase the amount of microorganisms that inhabit their surface. Both microorganisms themselves and the EPS they secrete exhibit stickiness under adverse conditions, which will increase the viscosity of particles. In addition, after a consolidation of 30 days, it was found that the particle size was higher under the condition of large flow than under small flow (Figure S5a, b, Supplementary material). Therefore, it can be concluded that the distribution of particles has been changed to a certain extent after a period of consolidation, which is consistent with the conclusion drawn from the change of turbidity. This change may be related to the increase of particle size, caused by the aggregation of organic matter and the consolidation of inorganic particles.

Stratified analysis was carried out on the variation of the characteristic of sediments from the bottom to the surface (five layers) over the 30 days. After consolidation of 30 days, the D50 of each layer of sediment from all sources (except surface sediments from the storm sewer of ID400) had increased (Figure S6a, b, c, Supplementary material). In contrast, the water contents of each layer had a different degree of reduction (Figure S6d, e, f, Supplementary material). The bottom water content decreased more rapidly than the surface for both SSSD and CSSD (Figure S6d, e, f, Supplementary material). Thus, it can be concluded that the surface layer has a relatively higher water-holding capacity than that of the sub-layer, though the surface layer is always considered to be prone to losing water on account of its regular exposure to air. Cells and EPS from a biofilm layer are related to the particle concentration, which might be explained by the water holding capacity of cells and EPS (Cetin & Erdincler 2004). Actually, Chen et al. (2003) confirmed that a biofilm layer was present at the sediment surface in a sanitary gravity sewer. The water contents of 1–4 layer sediments from SS-ID400 all decreased significantly, but their growth rates of D50 all increased (Figure S6a, d, Supplementary material). This finding implies that most of the surface particles may settle into the lower space in the deposition process of sediments from SS-ID400. This is probably due to the lack of viscous organic matter in the storm sewer.

Overall, the growth rates of D50 and the water content in each layer of the combined sewer are lower than those of the storm sewer. Water content is a good reflection of the living state of microbial life (Erdincier & Vesilind 2000). Microorganisms can grow well under the condition of sufficient organic matter, which helps to maintain moisture. Under adverse conditions, microbials could produce EPS, which could form aggregate structures for adhesion around the particles, thereby increasing the particle size. This finding may be the reason for the increasing particle size of SSSD. Additionally, water loss in the sediment (especially in bottom sediments) accelerates the mineralization rate, which promotes the consolidation of the bottom layer. Thus, in the scour process, the turbidity has a gentle change of storm sewer (Figure 3(b)).

• CSSD and SSSD have different erosion patterns, which are affected by the consolidation period, pipe types, organic matter, particle size, moisture and relative density. Additionally, illicit connection is an important factor affecting the viscosity and static friction force of particles, altering the erosion resistance of sewer sediments.

• In the flux experiment, the quantity of eroded samples was determined by continuous monitoring. Under the higher flow condition, the turbidity values of CSS after 30 days of consolidation showed a greater increase than without consolidation. Additionally, the sub-layer is exposed earlier to the higher flow rate. As the shear stress increases, the surface layer no longer covers the whole surface and releases the material beneath itself. After consolidation, the reduction ratio of organic matter is lowest at the surface layer of the sediment and increases continuously with the increase in sediment depth after 30 days of consolidation. This ratio is one of the reasons for the differentiation of sediments, leading fewer particles to be eroded from the deposits before the shear stress increases. Based on previous researches, it is assumed that the consolidation process will result in an increase in erosional strength with depth (Ristenpart 1995; De Sutter et al. 2003). In contrast, this phenomenon is not obvious for SSSD. The VSS content of CSSD is higher than that of SSSD.

• The results of the variation in particle size show that CS-ID800 > CS-ID1000 > SS-ID800 > SS-ID1000; this finding is the opposite of the peak time of the turbidity. Therefore, the particle size of sediments may have some relationship with the interaction between the particles, thus affecting the erosion resistance of the sediments. Under adverse conditions, microbials can produce EPS, which could form aggregate structures for adhesion around the particles, thereby increasing the particle size. This could be the reason for the increasing particle size of SSSD. Additionally, water loss of the sediment (especially in bottom sediments) accelerates the mineralization rate, which makes the bottom consolidation.

• The OUR of SSSD is lower than that from CSSD, which has a relationship with critical shear stress of sediments. However, there is no significant difference between the storm and CS of large diameter. Therefore, understanding quantitatively the direct impacts of microorganisms on the critical shear stress requires further analysis.

• The different illicit connection ratio causes the difference between CSSD and SSSD, which has strong effects on the CSOs and SSOs with different rainfall intensities. Elucidating the differences of time-varying pollutant loads between combined and SS can help to control overflows. Future laboratory investigation is needed, considering the influence of types of organic matters and microbial populations on erosion. In addition, a larger number of tests are needed to confirm the influence on critical shear stresses.

• Overall, this study will help to determine the parameters of flushing gates or weirs, based on different characteristics of sediments, in different combined and SS, which reduces the sedimentation in dry weather, and finally leads to a decrease in pollution from wet weather discharges.

These studies were financially supported by national water pollution control and management technology major projects (2014ZX07303-003). The authors would like to extend their appreciation to research team members for their help. We also appreciate Hong Kou Municipal Sewerage Co. Ltd for great help.